JP5185854B2 - Road friction coefficient estimation device - Google Patents

Description

  The present invention relates to a road surface friction coefficient estimation device that estimates a friction coefficient of a road surface on which a vehicle is traveling.

  As a technique for estimating the friction coefficient of the road surface on which the vehicle is traveling (hereinafter sometimes simply referred to as “μ”), for example, techniques disclosed in Patent Documents 1 and 2 have been proposed by the applicant of the present application.

  In the technique of Patent Document 1, road reaction force (cornering force (lateral force of the vehicle) and braking / driving force (vehicle lateral force) acting on each wheel from the road surface using tire characteristics set according to the estimated value of μ (vehicle) )), And based on the estimated value of the road surface reaction force, an estimated value of the lateral acceleration of the vehicle as a vehicle motion state quantity generated by the resultant force of the road surface reaction force and the vehicle The estimated value of the yaw rate change speed (the yaw rate change speed at the center of gravity of the vehicle) is calculated. In the technique of Patent Document 1, the deviation between the lateral acceleration detected value by the acceleration sensor and the lateral acceleration estimated value, and the differential value of the yaw rate detected value by the yaw rate sensor (the detected value of the yaw rate change speed) and the yaw rate change. A new estimated value of μ is obtained by updating the previous estimated value of μ in accordance with the larger deviation of the deviations from the estimated value of speed.

  In the technique of Patent Document 2, the road surface reaction force (cornering force and braking / driving force) acting on each wheel from the road surface is estimated using a tire model set according to the estimated value of μ. Based on the estimated force value, an estimated value of the lateral acceleration of the vehicle and an estimated value of the longitudinal acceleration of the vehicle are calculated as the amount of motion state of the vehicle generated by the resultant force of the road reaction force. . And in the technique of patent document 2, when the slip angle (side slip angle) of the rear wheel is small, the estimated value of μ according to the magnitude relationship between the estimated value of the longitudinal acceleration of the vehicle and the detected value of the longitudinal acceleration by the sensor. If the slip angle of the rear wheels is large, the estimated value of μ is incremented by a predetermined value according to the magnitude relationship between the estimated value of the lateral acceleration of the vehicle and the detected value of the lateral acceleration by the sensor. By increasing or decreasing, the estimated value of μ is sequentially updated.

  The road surface reaction force acting on the wheel depends not only on μ, but also on the slip ratio and side slip angle (slip angle) of the wheel. For this reason, in the techniques found in Patent Documents 1 and 2, the slip ratio of the wheel is estimated, and the side slip angle of the vehicle and the side slip angle of the wheel are also estimated using a vehicle motion model.

Japanese Patent No. 3669668 JP 2003-118554 A

  By the way, the techniques found in Patent Documents 1 and 2 are based on the premise that the lateral acceleration, longitudinal acceleration, and yaw rate change speed of the vehicle change depending on the change in μ on the road surface.

  On the other hand, according to the knowledge of the present inventor, the ratio of the change in the lateral acceleration, longitudinal acceleration, and yaw rate change speed of the vehicle to the change in μ on the road surface, that is, the lateral acceleration, longitudinal acceleration, yaw rate change speed of the vehicle relative to the change in μ. The sensitivity changes depending on the behavior state of the vehicle and the like. In a situation where the sensitivity is low (a situation where the sensitivity is close to “0”), the deviation between the estimated value and the detected value of the lateral acceleration, the longitudinal acceleration, and the yaw rate change speed of the vehicle has a low dependency on μ. It will be a thing. Therefore, in a situation where the sensitivity of the lateral acceleration, longitudinal acceleration, and yaw rate change speed of the vehicle with respect to changes in μ is low, the deviation between the estimated value and the detected value of the lateral acceleration, longitudinal acceleration, yaw rate change speed of the vehicle, or the It is considered desirable to suppress the update of the estimated value of μ in accordance with the magnitude relationship between the estimated value and the detected value.

  However, the above-mentioned sensitivity is not taken into consideration in the techniques found in Patent Documents 1 and 2. For this reason, the estimated value of μ may be excessively updated even in a situation where the sensitivity of the lateral acceleration, longitudinal acceleration, and yaw rate change speed of the vehicle with respect to the change of μ is low. As a result, there is an inconvenience that the estimated value of μ exhibits an unstable change or the accuracy of the estimated value of μ decreases.

  The present invention has been made in view of such a background, and prevents an estimated value of a friction coefficient of a road surface on which a vehicle is traveling from causing unstable fluctuations and a decrease in accuracy of the estimated value. An object of the present invention is to provide a road surface friction coefficient estimating device capable of estimating the friction coefficient.

A road surface friction coefficient estimation device according to the present invention is a road surface friction coefficient estimation device that estimates while updating the friction coefficient of a road surface on which a vehicle is traveling in order to achieve such an object.
A predetermined type of external force component acting on the vehicle by the resultant force of the road surface reaction force acting on each wheel of the vehicle from the road surface is set as a comparison target external force, and the first estimated value of the comparison target external force is set between the vehicle wheel and the road surface. First comparison target external force estimation obtained using a friction characteristic model representing the relationship between slip and road reaction force, an estimated value of the already determined friction coefficient, and an observation value of a predetermined type of observation target quantity related to vehicle behavior Means,
From the observed value of the motion state quantity of the vehicle that defines the inertial force corresponding to the external force to be compared among the inertial force generated by the motion of the vehicle, the value of the external force component commensurate with the inertial force is obtained, and the obtained external force component A comparison target external force second estimation unit that obtains a value of a second estimation value of the comparison target external force;
Μ sensitivity calculation means for obtaining a value of μ sensitivity, which is a ratio of the increase amount of the external force to be compared to the increase amount of the friction coefficient of the road surface, or a value obtained by dividing the ratio by the value of the friction coefficient of the road surface;
The deviation between the first estimated value and the second estimated value, or the first estimated filtered value and the second estimated value obtained by passing the first estimated value through the first filter for adjusting the frequency component are used for adjusting the frequency component. At least the deviation, the value of the μ sensitivity, or the value of the μ sensitivity, and a third filter for frequency component adjustment so that the deviation of the second estimated filtering value passed through the two filters converges to “0”. Friction coefficient increasing / decreasing manipulated variable determining means for determining an increasing / decreasing manipulated variable of the estimated value of the friction coefficient of the road surface according to the μ sensitivity dependent value passing through one or both of the saturation characteristic elements
Friction coefficient estimated value update means for determining a new estimated value of the friction coefficient by updating the estimated value of the friction coefficient of the road surface according to the increase / decrease operation amount,
The friction coefficient increasing / decreasing manipulated variable determining means determines the increasing / decreasing manipulated variable so as to decrease the increasing / decreasing manipulated variable as the μ sensitivity value or μ sensitivity dependent value decreases. (First invention).

  The “observed value” in the present invention is an indirect value using a suitable model or a natural law from a detected value directly observed from a certain sensor output or one or more sensor outputs related to the observation target quantity. Means the estimated value observed in

  In addition, examples of the first to third filters include filters having a high cut characteristic, a low cut characteristic, or a band pass characteristic. It is desirable that these first to third filters basically have frequency characteristics having the same tendency. Further, the saturation characteristic element has such a characteristic that the larger the magnitude (absolute value) of the μ sensitivity value, the smaller the rate of change of the output of the saturation characteristic element with respect to the change of the μ sensitivity value. It is an element that has. In this case, the ratio of the change in the output of the saturation characteristic element to the change in the value of the μ sensitivity may change continuously with the change in the value of the μ sensitivity value, but changes discontinuously. Also good.

  In the first invention, the comparison target external force first estimation means uses the friction characteristic model representing the relationship between the slip between the vehicle wheel and the road surface and the road surface reaction force as the estimated value of the comparison target external force, It is determined using the determined estimated value of the friction coefficient (hereinafter sometimes referred to as a determined estimated value) and the observed value of a predetermined type of observation target quantity related to the behavior of the vehicle. As a result, the first estimated value as the value of the comparison target external force specified depending on the determined estimated value of the friction coefficient is obtained. In this case, more specifically, the slip in the friction characteristic model is specified (estimated) from the observed value of the predetermined type of observation target amount related to the behavior of the vehicle, and the slip and the determined estimated value of the friction coefficient are By inputting the friction characteristic model, it is possible to estimate the road surface reaction force acting on each wheel of the vehicle. And the value of the said comparison external force calculated | required from the estimated road surface reaction force should just be obtained as a 1st estimated value. Therefore, the observation value of the predetermined type of observation target amount may be an observation value of the observation target amount necessary for specifying the slip between the wheel and the road surface in the friction characteristic model. Such an observation target amount may be selected in accordance with the structure of the friction characteristic model.

  In addition, as an index value showing the slip between a wheel and a road surface, the slip ratio of each wheel, a skid angle, etc. can be used, for example. Examples of the external force to be compared include a translation force in a predetermined direction that acts on the entire vehicle due to the road reaction force, and a moment around a predetermined axis. Further, the determined estimated value of the friction coefficient is preferably the latest value of the already determined estimated values, but within a sufficiently short period such that the estimated value of the friction coefficient is kept substantially constant. As long as it is a value, it may be a past value than the latest value.

  On the other hand, the comparison target external force second estimation means balances the inertial force based on the observed value of the motion state quantity of the vehicle that defines the inertial force corresponding to the comparison target external force out of the inertial force generated by the motion of the vehicle. A value of the external force component is obtained, and the obtained value of the external force component is obtained as a second estimated value of the comparison target external force. Accordingly, the second estimated value of the comparison target external force can be obtained from the observed value of the motion state quantity of the vehicle that defines the inertial force corresponding to the comparison target external force without using the estimated value of the friction coefficient of the road surface. .

  The first invention is characterized in that at least a deviation between the first estimated value and the second estimated value, or a deviation between the first estimated filtering estimated value and the second estimated filtered value by the friction coefficient increasing / decreasing manipulated variable determining means, In accordance with the μ sensitivity value or the μ sensitivity dependent value, an increase / decrease operation amount of the estimated value of the friction coefficient of the road surface is determined so that the deviation converges to “0”. At this time, the increase / decrease operation amount is determined such that the smaller the μ sensitivity value or the μ sensitivity dependent value, the smaller the increase / decrease operation amount. Further, according to the first aspect of the present invention, a new estimated value of the friction coefficient is determined by updating the estimated value of the friction coefficient of the road surface according to the increase / decrease operation amount by the friction coefficient estimated value update means.

  Here, in the first aspect of the invention, the increase / decrease operation amount is determined such that the smaller the μ sensitivity value or the μ sensitivity dependent value, the smaller the increase / decrease operation amount. The smaller the sensitivity value or μ sensitivity-dependent value is, the smaller the update amount of the estimated value of the friction coefficient by the increase / decrease operation amount is, and the update of the estimated value of the friction coefficient is suppressed. In other words, the smaller the μ sensitivity value or the μ sensitivity dependent value is, the smaller the gain value (feedback gain) that is the ratio of the change in the increase / decrease operation amount to the change in the deviation is suppressed. . For this reason, the estimated value of the friction coefficient of the road surface is excessive in a situation where the value of the μ sensitivity value or the μ sensitivity dependent value is small (that is, the dependence amount of the comparison target with respect to the road friction coefficient is low) It is possible to prevent the change. As a result, it is possible to estimate the friction coefficient while preventing the estimated value of the friction coefficient of the road surface on which the vehicle is traveling from causing unstable fluctuations and a decrease in the accuracy of the estimated value.

  Supplementally, examples of the external force to be compared include, for example, a moment around the yaw axis at a predetermined point of the vehicle (for example, a neutral steer point described later), and a road surface reaction force (driving / controlling force) acting on the front wheel of the vehicle. The combined force of power and lateral force or lateral force) is suitable.

  Further, the μ sensitivity can be calculated as follows, for example. That is, using the friction characteristic model, the friction coefficient value shifted by a predetermined amount from the determined estimated value, and the observed value of the predetermined type of observation target amount, the comparison corresponding to the friction coefficient value Find the value of the off-target force. In other words, instead of the determined estimated value, the friction coefficient value shifted from the determined estimated value by a predetermined amount is used to perform the same process as the comparison target external force first estimating unit, thereby The comparison target external force value corresponding to the coefficient value is obtained. Then, by dividing the deviation between the calculated external force value to be compared and the first estimated value by the predetermined amount, or by dividing the divided result value by the determined estimated value, the μ Calculate the sensitivity.

  Further, when the moment around the yaw axis at the vehicle's neutral steer point (hereinafter sometimes referred to as NSP) is used as the external force to be compared, the μ sensitivity is set as the steering of the vehicle wheel. It can also be obtained by linearly combining the observed value of the wheel steering angle and the observed value of the vehicle yaw rate. In this case, the weighting coefficient applied to the observed value of the steering angle and the observed value of the yaw rate in the linear combination is set so that the mutual ratio of the two weighting coefficients changes according to the vehicle speed of the vehicle. Preferably, at least one of the coefficients is set according to the observed value of the vehicle speed, and the linear combination is calculated using the set weight coefficient.

  In the first aspect of the invention, more specifically, the friction coefficient increase / decrease manipulated variable determining means determines the deviation according to a deviation / μ sensitivity dependent value that is a product of the deviation and the μ sensitivity value or μ sensitivity dependent value. It is preferable to determine the amount of increase / decrease operation (second invention).

  According to the second invention, the deviation / μ sensitivity product approaches “0” as the μ sensitivity value or the μ sensitivity dependent value approaches “0”. Therefore, by determining the increase / decrease operation amount according to the deviation / μ sensitivity product, the gain value becomes smaller as the μ sensitivity value or the μ sensitivity dependent value is closer to “0”. It is possible to determine the increase / decrease operation amount (so that the increase / decrease operation amount becomes smaller).

  In the second invention, more specifically, the friction coefficient increasing / decreasing manipulated variable determining means determines the increasing / decreasing manipulated variable according to the deviation / μ sensitivity product so as to be proportional to the deviation / μ sensitivity product. It is preferable to do this (third invention).

  According to the third aspect of the invention, since the increase / decrease operation amount is proportional to the deviation / μ sensitivity product, the increase / decrease operation amount approaches “0” as the μ sensitivity value approaches “0”. Will be determined.

  In the first to third inventions described above, the friction coefficient increasing / decreasing manipulated variable determining means uses at least the deviation of the first estimated filtering value and the second estimated filtering value and the μ sensitivity value to the third filter. When it is means for determining the increase / decrease operation amount in accordance with the μ sensitivity dependent value that is passed through, it is one form that the first filter, the second filter, and the third filter are filters having low cut characteristics. (4th invention).

  The low cut characteristic means a frequency characteristic that cuts off a frequency component equal to or lower than a predetermined frequency.

  As described above, when the first to third filters have low cut characteristics, they are constantly included in the first estimated value, the second estimated value, and the μ sensitivity due to the offset or drift of the sensor output. The increase / decrease operation amount can be determined by eliminating the low frequency or DC error component. As a result, the system and stability of the estimated value of the friction coefficient can be enhanced.

  Alternatively, the friction coefficient increasing / decreasing manipulated variable determining means may obtain a deviation between the first estimated filtering value and the second estimated filtering value and a μ sensitivity dependent value obtained by passing the μ sensitivity value through at least the third filter. In the case where the increase / decrease operation amount is determined in response, the first filter, the second filter, and the third filter may be filters having a high-cut characteristic as another embodiment (fifth invention). ).

  The high cut characteristic means a frequency characteristic that cuts off a frequency component of a predetermined frequency or higher.

  As described above, when the first to third filters have high cut characteristics, they are included in the first estimated value, the second estimated value, and the μ sensitivity due to high-frequency noise components included in the output of the sensor. The amount of increase / decrease operation can be determined by eliminating the high frequency error component. As a result, the system and stability of the estimated value of the friction coefficient can be enhanced.

  The first to third filters in the fourth invention or the fifth invention may be filters having frequency characteristics of both a low cut characteristic and a high cut characteristic (for example, a bandpass characteristic filter).

  In the fifth invention, wherein the first to third filters have high cut characteristics, the friction coefficient increasing / decreasing manipulated variable determining means determines a provisional value of the increasing / decreasing manipulated variable according to the deviation and the μ sensitivity dependent value. On the other hand, it is preferable to determine the increase / decrease operation amount by passing the provisional value through a phase compensation filter having a function of advancing the phase of the provisional value (sixth invention).

  That is, when the first to third filters have a high-cut characteristic, the deviation between the first estimated filtering value and the second estimated filtering value with respect to the external force to be compared and μ sensitivity in a relatively high frequency range, and the μ Sensitivity-dependent value phase lag is likely to occur. For this reason, if the increase / decrease operation amount is determined using the deviation and the μ sensitivity dependent value as they are, the estimated value of the friction coefficient may oscillate, and the estimated value may become unstable. There is. Accordingly, in the sixth aspect of the invention, the provisional value of the increase / decrease manipulated variable is determined according to the deviation and the μ sensitivity dependent value, and the provisional value is passed through the phase compensation filter to thereby delay the phase of the provisional value. Was solved. Thereby, the situation where the estimated value of the friction coefficient of the road surface vibrates can be prevented, and the stability of the estimated gear value can be improved more effectively.

  In the fourth to sixth inventions, the friction coefficient increasing / decreasing manipulated variable determining means determines that the deviation value is “0” when the deviation is within a predetermined range as a range in the vicinity of “0”. Preferably, the increase / decrease operation amount is determined as being present (seventh aspect).

  That is, when the deviation between the first estimated filtering value and the second estimated filtering value is a value near “0”, the deviation is caused by an error component of the first estimated value or the second estimated value. Many ingredients are easily included. That is, the S / N ratio of the deviation tends to be low. Therefore, in the seventh invention, when the deviation is within a predetermined range as a range in the vicinity of “0”, the increase / decrease operation amount is determined on the assumption that the value of the deviation is “0”. Thereby, it is possible to prevent the estimated value of the friction coefficient from being updated inappropriately in a situation where the S / N ratio of the deviation is low, and to prevent the estimated value from becoming unstable.

  In the fourth to seventh inventions, the friction coefficient increasing / decreasing manipulated variable determining means determines that the μ sensitivity dependent value is within a predetermined range as a range near “0”. It is preferable to determine the increase / decrease operation amount as “0” (eighth invention).

  That is, when the μ sensitivity dependent value is a value in the vicinity of “0”, the S / N ratio of the μ sensitivity dependent value tends to be low. Accordingly, in the eighth invention, when the μ sensitivity dependent value is within a predetermined range as a range in the vicinity of “0”, the increase / decrease operation amount is determined on the assumption that the μ sensitivity dependent value is “0”. I tried to do it. As a result, it is possible to prevent the estimated value of the friction coefficient from being updated inappropriately and prevent the estimated value from becoming unstable in a situation where the S / N ratio of the μ sensitivity dependent value is low.

The figure which shows schematic structure of the vehicle in embodiment. 2A and 2B are diagrams visually showing typical reference symbols used in the description of the embodiment. The block diagram which shows the main functions of the control apparatus in 1st Embodiment. The flowchart which shows the process of the control apparatus in 1st Embodiment. The block diagram which shows the function of the vehicle model calculating means shown in FIG. 6A and 6B are graphs for explaining the processing of the wheel slip ratio estimation unit shown in FIG. 7A and 7B are graphs for explaining the processing of the wheel skid angle estimating unit shown in FIG. The graph for demonstrating the process of the other form of the wheel slip ratio estimation part shown in FIG. The flowchart which shows the process of the bank angle estimation means shown in FIG. The flowchart which shows the process of the gradient angle estimation means shown in FIG. FIG. 4 is a block diagram showing functions of μ estimation means shown in FIG. 3. The flowchart which shows the process of the micro estimation means shown in FIG. The flowchart which shows the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 2nd Embodiment. The figure for demonstrating the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 3rd Embodiment. The flowchart which shows the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 3rd Embodiment. The flowchart which shows the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 4th Embodiment. The flowchart which shows the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 5th Embodiment. The block diagram which shows the principal part of the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 6th Embodiment. The block diagram which shows the principal part of the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 7th Embodiment. The block diagram which shows the principal part of the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 8th Embodiment. The block diagram which shows the principal part of the determination process of the friction coefficient increase / decrease manipulated variable (DELTA) micro in 9th Embodiment.

  Hereinafter, embodiments of the present invention will be described. First, a schematic configuration of a vehicle according to an embodiment of the present specification will be described with reference to FIG.

  As shown in FIG. 1, the vehicle 1 includes a plurality of wheels 2-i (i = 1, 2,...), And these wheels 2-i (i = 1, 2,. The vehicle body 1B is supported. More specifically, the vehicle 1 according to the embodiment includes a total of four wheels 2-i (i = 1, i.e., a pair of left and right front wheels 2-1 and 2-2 and a pair of left and right rear wheels 2-3 and 2-4. 2, 3, 4). In this case, the front wheels 2-1 and 2-2 of the wheels 2-i (i = 1, 2, 3, 4) are driving wheels and steering wheels, and the rear wheels 2-3, 2-4 are It is a non-steering wheel as well as a driven wheel.

  In the following explanation, the left front wheel 2-1 of the vehicle 1 is the first wheel 2-1, the right front wheel 2-2 is the second wheel 2-2, and the left rear wheel 2-3 is the third wheel. 2-3, the wheel 2-4 on the right rear side may be referred to as a fourth wheel 2-4. In addition, when expressing an arbitrary wheel of the wheels 2-i (i = 1, 2, 3, 4), a description such as “(i = 1, 2, 3, 4)” is omitted. May be simply referred to as “wheel 2-i” or “i-th wheel 2-i”. In addition, reference numerals of elements other than the wheels 2-i (i = 1, 2, 3, 4) (components, physical quantities, etc.) and related to the individual i-th wheels 2-i are subscripts “ i ”is added. In this case, for an element corresponding to a specific one of the wheels 2-i (i = 1, 2, 3, 4), instead of the subscript “i” in the reference numeral of the element, The value of i (1 or 2 or 3 or 4) corresponding to the specific wheel is added.

  The vehicle 1 is provided with a drive system for rotationally driving the drive wheels. In the embodiment, the drive system includes an engine 3 as a power generation source mounted on the vehicle body 1B. The drive system transmits the power (output torque) of the engine 3 to the front wheels 2-1 and 2-2 as drive wheels via the power transmission mechanism 4 including the transmission 4a, thereby causing the front wheels 2-1 to move. , 2-2 are rotated. In this case, the power of the engine 3 is controlled in accordance with a depression operation amount of an accelerator pedal (not shown) of the vehicle 1.

  The vehicle 1 is also provided with a steering system for steering the steered wheels. In this embodiment, the steering system includes a steering wheel 5 disposed in front of the driver's seat of the vehicle body 1B. The front wheel 2-1 as a steering wheel is operated by a steering mechanism (not shown) in conjunction with a rotation operation of the steering wheel 5. , 2-2 is steered. The steering mechanism is configured by a mechanical steering mechanism such as a rack and pinion, or a steering mechanism with an actuator having a steering actuator such as an electric motor (so-called power steering device).

  The vehicle 1 is also provided with a braking system for braking the running. In this embodiment, the braking system includes a frictional braking mechanism 7-i (i = 1, 2, 3, 4) such as a disc brake for each wheel 2-i. These braking mechanisms 7-i (i = 1, 2, 3, 4) are connected to a braking system hydraulic circuit 6, and depending on the hydraulic pressure (brake pressure) applied from the braking system hydraulic circuit 6, respectively. A braking force for braking the rotation of the corresponding wheel 2-i is generated. In this case, the braking system hydraulic circuit 6 basically applies the brake pressure corresponding to the depression amount (depression force) of the brake pedal in conjunction with the depression operation of the brake pedal (not shown) of the vehicle 1. Granted to mechanism 7-i. In the vehicle 1 according to the embodiment, the brake system hydraulic circuit 6 applies a brake pressure to be applied to each brake mechanism 7-i (and thus a brake force of each wheel 2-i) from the control device 20 described later. It is possible to adjust according to the control command.

  Furthermore, the vehicle 1 includes various sensors for detecting an observation target amount, which will be described later, and a control device 20 that performs behavior control of the vehicle 1 in addition to the drive system, the steering system, and the braking system. In the embodiment, as a sensor, for example, a wheel rotation angular velocity sensor 8-i (i = 1, 2, 3, 4) that generates an output corresponding to the rotation angular velocity of each wheel 2-i, and each wheel 2-i. A brake pressure sensor 9-i (i = 1, 2, 3, 4) for generating an output corresponding to the brake pressure applied to the brake mechanism 7-i, and a steering angle (rotation angle) of the steering wheel 5 Steering steering angle sensor 10 that generates output, transmission sensor 11 that generates output according to the operating state (gear ratio, etc.) of the transmission 3, output according to the amount of depression of an accelerator pedal (not shown) of the vehicle 1 Accelerator sensor 12 for generating the output, yaw rate sensor 13 for generating an output corresponding to the yaw rate that is the angular velocity around the yaw axis of the vehicle 1 (around the vertical axis of the vehicle body 1B), and the roll axis direction of the vehicle 1 (the vehicle body A longitudinal acceleration sensor 14 for generating an output corresponding to the acceleration in the direction B), and a lateral acceleration sensor 15 for generating an output corresponding to the acceleration in the pitch axis direction of the vehicle 1 (the lateral direction (left and right direction) of the vehicle body 1B). It is mounted on the vehicle 1.

  The control device 20 is an electronic circuit unit including a CPU, a RAM, a ROM, and the like, and outputs (detection data) of each sensor described above are input. Then, the control device 20 controls the behavior of the vehicle 1 by executing a predetermined calculation process based on a pre-installed program while using the input detection data and the setting data stored and held in advance. . In this case, the control device 20 controls the rotational force (turning motion) around the yaw axis of the vehicle 1 by controlling the braking force of each wheel 2-i by each braking mechanism 7-i via the braking system hydraulic circuit 6, for example. ) And side-slip motion, etc. In addition, the control device 20 also has a function of sequentially estimating a friction coefficient of a road surface on which the vehicle 1 is traveling in order to execute a behavior control process of the vehicle 1. The estimated friction coefficient is used, for example, to estimate the state quantity (side slip angle, side slip speed, etc.) of the side slip motion of the vehicle 1 or to determine the behavior of the target vehicle 1.

  The above is the schematic configuration of the vehicle 1 in each embodiment described in this specification.

  The vehicle to which the present invention is applied is not limited to the vehicle 1 having the above configuration. For example, the power generation source of the drive system of the vehicle 1 may be an electric motor. Alternatively, both the engine and the electric motor may be mounted on the vehicle 1 as power generation sources. Further, the drive wheels of the vehicle 1 may be the rear wheels 2-3, 2-4, or may be both the front wheels 2-1, 2-2 and the rear wheels 2-3, 2-4. Good. Further, the drive system may be configured such that the drive force applied to each drive wheel from the power generation source can be adjusted individually. In addition to steering the front wheels 2-1 and 2-2 in conjunction with the rotation operation of the steering wheel 5, the steering system of the vehicle 1 appropriately steers the rear wheels 2-3 and 2-4 by an actuator. It may be configured to. Further, the number of wheels may not be four.

  Next, with reference to FIGS. 2A and 2B, main reference symbols (variables) and terms used in the following description of each embodiment will be described.

  Variables prefixed with “↑”, such as ↑ V1, ↑ F1, etc. in FIGS. 2A and 2B, represent vector quantities. The vector quantity is expressed in the form of a column vector (transposition vector of row vector) when the component is displayed using an appropriate coordinate system. In the description of the embodiment, “×” is used as an arithmetic symbol for multiplication (ie, outer product) of vector quantities, and an arithmetic symbol for multiplication other than outer products, such as multiplication of scalar quantities or multiplication of a scalar quantity and a vector quantity. “*” Is used. Further, in order to indicate transposition of a row vector, the subscript “T” is added to the upper right of the component display of the row vector.

  The “vehicle body coordinate system” is a coordinate system in which the front-rear direction of the vehicle body 1B is the X-axis direction and the lateral direction (left-right direction) of the vehicle body 1B is the Y-axis direction. In this case, the forward direction of the vehicle body 1B is the positive direction of the X axis, and the left direction of the vehicle body 1B is the positive direction of the Y axis. Note that the X-axis direction of the vehicle body coordinate system may simply be the front-rear direction of the vehicle 1 or the roll axis direction. In addition, the Y-axis direction of the vehicle coordinate system may simply be the lateral direction or the pitch axis direction of the vehicle 1. Further, the yaw axis direction of the vehicle 1 (the vertical direction of the vehicle body 1B) is orthogonal to the XY plane of the vehicle body coordinate system (orthogonal to the X axis and the Y axis).

  The “i-th wheel coordinate system” is parallel to the rotation surface of the i-th wheel 2-i (a surface orthogonal to the rotation axis of the i-th wheel 2-i) when the vehicle 1 is viewed from above in the yaw axis direction. The direction (front-rear direction of the i-th wheel 2-i) is the x-axis direction, and the direction parallel to the rotation axis of the i-th wheel 2-i (the left-right direction (lateral direction) of the i-th wheel 2-i) is the y-axis. Coordinate system. In this case, the forward direction of the i-th wheel 2-i is the positive direction of the x-axis, and the left direction of the i-th wheel 2-i is the positive direction of the y-axis. It is assumed that the xy plane of the i-th wheel coordinate system is parallel to the XY plane of the vehicle body coordinate system and is orthogonal to the yaw axis direction of the vehicle 1.

  Supplementally, “orthogonal” and “parallel” in this specification do not mean only orthogonal or parallel in a strict sense, but may be approximate orthogonal or parallel.

  “Δi” represents the rudder angle of the i-th wheel 2-i (hereinafter, sometimes simply referred to as the wheel rudder angle). More specifically, each wheel rudder angle δi is an angle formed by the rotational surface of the i-th wheel 2-i with respect to the X-axis direction of the vehicle body coordinate system when the vehicle 1 is viewed from above in the yaw axis direction. In the vehicle 1 of the embodiment, since the rear wheels 2-3 and 2-4 are non-steering wheels, δ3 = δ4 = 0 is always satisfied.

  “↑ Vg” represents a moving speed vector (hereinafter referred to as a vehicle center-of-gravity speed vector) of the center of gravity of the vehicle 1 with respect to the road surface as viewed by projection onto the XY plane of the vehicle body coordinate system. The vehicle center-of-gravity velocity vector ↑ Vg is a vector composed of an X-axis direction component and a Y-axis direction component of the vehicle body coordinate system. In this case, the X-axis direction component of the vehicle center-of-gravity velocity vector ↑ Vg is expressed as Vgx, the Y-axis direction component is expressed as Vgy, and these are referred to as vehicle center-of-gravity longitudinal velocity Vgx and vehicle center-of-gravity skid velocity Vgy. In other words, the vehicle center-of-gravity longitudinal speed Vgx has a meaning as the traveling speed (vehicle speed) of the vehicle 1. Although not shown in FIGS. 2A and 2B, the temporal change rate (differential value) of the vehicle center-of-gravity longitudinal speed Vgx is the temporal change rate of the vehicle center-of-gravity longitudinal speed Vgdot_x and the vehicle center-of-gravity skid speed Vgy. The rate of change (differential value) is referred to as vehicle center-of-gravity skid speed change rate Vgdot_y.

“Βg” represents a side slip angle of the center of gravity of the vehicle 1 (hereinafter referred to as a vehicle center of gravity side slip angle). More specifically, the vehicle center-of-gravity skid angle βg is an angle formed by the vehicle center-of-gravity velocity vector ↑ Vg with respect to the X-axis direction of the vehicle body coordinate system. Therefore, βg = tan ⁻¹ (Vgy / Vgx).

  “↑ Vi” is a moving speed vector of the contact portion of the i-th wheel 2-i with respect to the road surface projected on the XY plane of the vehicle body coordinate system (hereinafter referred to as a traveling speed vector of the i-th wheel 2-i, or simply Wheel speed vector). Each wheel traveling speed vector ↑ Vi is a vector composed of an X-axis direction component and a Y-axis direction component of the vehicle body coordinate system. In this case, although not shown in FIGS. 2A and 2B, the X axis direction component of each wheel traveling speed vector ↑ Vi is expressed as Vx_i, and the Y axis direction component is expressed as Vy_i.

  “↑ Vsub_i” is a moving speed vector (hereinafter referred to as a wheel traveling speed vector on the wheel coordinate system) of the ground contact portion of the i-th wheel 2-i with respect to the road surface, as projected onto the xy plane of the i-th wheel coordinate system. Represent. The wheel traveling speed vector ↑ Vsub_i on each wheel coordinate system is a vector composed of the x-axis direction component and the y-axis direction component of the i-th wheel coordinate system. In this case, although illustration in FIGS. 2A and 2B is omitted, the x-axis direction component of the wheel traveling speed vector ↑ Vsub_i on each wheel coordinate system is denoted as Vsubx_i, and the y-axis direction component is denoted as Vsuby_i. The wheel traveling speed vector ↑ Vsub_i on the wheel coordinate system of each wheel 2-i and the wheel traveling speed vector ↑ Vi differ only in the coordinate system representing them, and the spatial orientation and size are different. They are the same vector quantity.

“Βi” represents the side slip angle of the i-th wheel 2-i (hereinafter, simply referred to as the wheel side slip angle). More specifically, each wheel side slip angle βi is an angle formed by the wheel traveling speed vector ↑ Vsub_i on the wheel coordinate system of the i-th wheel 2-i with respect to the x-axis direction of the i-th wheel coordinate system. Therefore, βi = tan ⁻¹ (Vsuby_i / Vsubx_i).

  “Β0i” represents an angle (= βi + δi, hereinafter referred to as a wheel position side slip angle) formed by the wheel traveling speed vector ↑ Vi of the i-th wheel 2-i with respect to the X-axis direction of the vehicle body coordinate system. In the embodiment, since the rear wheels 2-3 and 2-4 are non-steering wheels, β03 = β3 and β04 = β4. For this reason, illustration of β03 and β04 is omitted.

  “Γ” represents an angular velocity around the yaw axis of the vehicle 1, that is, a yaw rate.

  “Df” is the distance between the front wheels 2-1 and 2-2 in the lateral direction of the vehicle 1 (Y-axis direction of the vehicle body coordinate system) (that is, the tread of the front wheels 2-1 and 2-2), and “dr” is the vehicle 1 represents the distance between the rear wheels 2-3 and 2-4 in the horizontal direction (Y-axis direction of the vehicle body coordinate system) (that is, the tread of the rear wheels 2-3 and 2-4). Hereinafter, df is referred to as a front wheel tread, and dr is referred to as a rear wheel tread.

  “Lf” is the distance between the axles (rotating shafts) of the front wheels 2-1 and 2-2 in the state where δ1 = δ2 = 0 and the center of gravity of the vehicle 1 (distance in the front-rear direction of the vehicle 1), “ Lr ″ represents the distance between the axle (rotary axis) of the rear wheels 2-3 and 2-4 and the center of gravity of the vehicle 1 (the distance in the front-rear direction of the vehicle 1). Hereinafter, Lf is referred to as the distance between the front wheel axle and the center of gravity, and Lr is referred to as the distance between the rear wheel axle and the center of gravity.

“↑ Pi” is a position vector of the i-th wheel 2-i viewed from the center of gravity of the vehicle 1 in a state where the vehicle 1 is viewed from above in the yaw axis direction (hereinafter, sometimes simply referred to as a wheel position vector). Represent. Each wheel position vector ↑ Pi is a vector composed of an X-axis direction component and a Y-axis direction component of the vehicle body coordinate system. In this case, although not shown in FIGS. 2A and 2B, the X-axis direction component of each wheel position vector ↑ Pi is expressed as Px_i, and the Y-axis direction component is expressed as Py_i. When the position of the center of gravity of the vehicle 1 in the Y-axis direction of the vehicle body coordinate system is on the center line of the vehicle width of the vehicle 1, ↑ P1 = (Lf, df / 2) ^T , ↑ P2 = (Lf, −df / 2) ^T , ↑ P3 = (− Lr, dr / 2) ^T , ↑ P4 = (− Lr, −dr / 2) ^T

  “↑ Fi” represents the road surface reaction force of the i-th wheel 2-i (translational force vector acting on the i-th wheel 2-i from the road surface) as projected on the XY plane of the vehicle body coordinate system. Hereinafter, ↑ Fi is referred to as a wheel two-dimensional road surface reaction force or a two-dimensional road surface reaction force. This wheel two-dimensional road surface reaction force ↑ Fi is a vector composed of an X-axis direction component and a Y-axis direction component of the vehicle body coordinate system. Here, the road surface reaction force acting on each wheel 2-i from the road surface spatially (three-dimensionally) is a driving / braking force that is a translational force component in the x-axis direction of the i-th wheel coordinate system, and y It is a resultant vector of lateral force, which is a translational force component in the axial direction, and ground contact load, which is a translational force component in the yaw axis direction. Accordingly, the wheel two-dimensional road surface reaction force ↑ Fi is a resultant vector of the driving / braking force and lateral force of the i-th wheel 2-i (this corresponds to a friction force acting on the i-th wheel 2-i from the road surface). Is a vector that is expressed in the vehicle body coordinate system. In this case, although illustration in FIGS. 2A and 2B is omitted, the X-axis direction component of the wheel two-dimensional road surface reaction force ↑ Fi is expressed as Fx_i, and the Y-axis direction component is expressed as Fy_i. In the following description, the spatial road surface reaction force as a resultant vector of the driving / braking force, lateral force, and contact load of each wheel 2-i is referred to as a wheel three-dimensional road surface reaction force or a three-dimensional road surface reaction force. . Moreover, the ground contact load as a yaw-axis direction component of the three-dimensional road surface reaction force of each wheel 2-i is described as Fz_i.

  “↑ Fsub_i” represents the road surface reaction force of the i-th wheel 2-i projected on the xy plane of the i-th wheel coordinate system (hereinafter referred to as a wheel coordinate system upper wheel two-dimensional road surface reaction force). Each wheel coordinate system upper wheel two-dimensional road surface reaction force ↑ Fsub_i is a vector composed of an x-axis direction component and a y-axis direction component of the i-th wheel coordinate system. In this case, although illustration in FIGS. 2A and 2B is omitted, the x-axis direction component of each wheel coordinate system upper wheel two-dimensional road surface reaction force ↑ Fsub_i is represented as Fsubx_i, and the y-axis direction component is represented as Fsuby_i. . In other words, the x-axis direction component Fsubx_i is the driving / braking force of the i-th wheel 2-i, and the y-axis direction component Fsuby_i is the lateral force of the i-th wheel 2-i. The wheel coordinate system upper wheel two-dimensional road surface reaction force ↑ Fsub_i of the i-th wheel 2-i and the wheel two-dimensional road surface reaction force ↑ Fi of the i-th wheel 2-i are different in the coordinate system expressing them. These are vector quantities having the same spatial orientation and size.

  “↑ Fg_total” is the resultant force of the road surface reaction force acting on the wheel 2-i (i = 1, 2, 3, 4) (the resultant force of the wheel three-dimensional road surface reaction force (i = 1, 2, 3, 4)) ) Represents a spatial translation force vector (hereinafter referred to as a total road surface reaction force synthesis translation force vector) acting on the center of gravity of the vehicle 1. In this case, although illustration in FIGS. 2A and 2B is omitted, the X-axis direction component of the vehicle body coordinate system of the total road surface reaction force combined translation force vector ↑ Fg_total is Fgx_total, and Y of the vehicle body coordinate system The axial direction component is expressed as Fgy_total, and the yaw axis direction component is expressed as Fgz_total. Also, Fgx_total may be referred to as total road reaction force combined longitudinal force, and Fgy_total may be referred to as total road reaction force combined lateral force.

  “Mgz_total” is the resultant of the road surface reaction force acting on the wheel 2-i (i = 1, 2, 3, 4) (the i-th wheel three-dimensional road surface reaction force (i = 1, 2, 3, 4)). The resultant force represents a moment acting around the yaw axis at the center of gravity of the vehicle 1 (hereinafter referred to as a total road surface reaction force combined yaw moment). The yaw axis direction component Fgz_total of the resultant force of the wheel three-dimensional road surface reaction force (i = 1, 2, 3, 4) does not contribute to the total road surface reaction force combined yaw moment Mgz_total. Therefore, the total road surface reaction force combined yaw moment Mgz_total is substantially equal to the resultant force of the wheel two-dimensional road surface reaction force ↑ Fi (i = 1, 2, 3, 4), that is, all the wheels 2-i (i = 1, 2, 3, 4) represents the moment acting around the yaw axis at the center of gravity of the vehicle 1 by the resultant force of driving / braking force and lateral force.

  Supplementally, in the embodiment of the present specification, the resultant force of the road surface reaction force acting on the wheel 2-i (i = 1, 2, 3, 4) is regarded as the entire external force acting on the vehicle 1. More specifically, the external force acting on the vehicle 1 includes air resistance and the like in addition to the road surface reaction force acting on each wheel 2-i from the road surface, but in the embodiment, the external force other than the road surface reaction force is: Compared to the resultant force of the road surface reaction force acting on the wheel 2-i (i = 1, 2, 3, 4), it is considered to be sufficiently small to be negligible. Therefore, ↑ Fg_total and Mgz_total have meanings as a translational force vector and a moment acting on the center of gravity of the vehicle 1 due to the entire external force acting on the vehicle 1, respectively.

  “NSP” represents the neutral steer point of the vehicle 1. When the vehicle 1 is traveling with δ1 = δ2 = 0 and the vehicle center-of-gravity skid angle βg (≠ 0) is generated, the NSP generates all wheels 2-i (i = 1, 2, 3, 4). ) Means the point of application (action point) of the resultant force of the lateral force Fsuby_i (i = 1, 2, 3, 4). More specifically, the NSP has a straight line extending in the X-axis direction of the vehicle body coordinate system (the front-rear direction of the vehicle 1) through the center of gravity of the vehicle 1 when the vehicle 1 is viewed from above in the yaw axis direction. It means the intersection with the action line of the resultant force of the lateral force Fsuby_i (i = 1, 2, 3, 4) acting on all the wheels 2-i (i = 1, 2, 3, 4), respectively.

  “Lnsp” represents the distance between the center of gravity of the vehicle 1 and the NSP in the X-axis direction (the front-rear direction of the vehicle 1) of the vehicle body coordinate system (hereinafter referred to as the vehicle center-of-gravity / distance between NSPs). When the NSP exists behind the center of gravity of the vehicle 1, the vehicle center-NSP distance Lnsp is set to a positive value, and the NSP exists ahead of the center of gravity of the vehicle 1. The vehicle center of gravity / NSP distance Lnsp is set to a negative value.

  “Mnsp” is the resultant force of the road surface reaction force acting on the wheel 2-i (i = 1, 2, 3, 4) (the resultant force of the wheel three-dimensional road surface reaction force (i = 1, 2, 3, 4) or By the wheel two-dimensional road surface reaction force ↑ Fi (the resultant force of i = 1, 2, 3, 4), a moment acting around the yaw axis by NSP (hereinafter referred to as NSP yaw moment) is represented. In other words, the NSP yaw moment Mnsp is the sum of the total road surface reaction force combined yaw moment Mgz_total and the total road surface reaction force combined translational force vector ↑ Fg_total generated around the yaw axis by the NSP (= Lnsp * Fgy_total). Moment.

  Supplementally, in the embodiment, state quantities (δi, βi, γ, etc.) related to rotational movement around the yaw axis, such as angles, angular velocities, angular accelerations around the yaw axis, and moments (Mgz_total, Mnsp, etc.) around the yaw axis, With respect to, in the state where the vehicle 1 is viewed from above in the yaw axis direction, the counterclockwise direction is defined as a positive direction.

  Although not shown in FIGS. 2A and 2B, the following variables are used in the following description in addition to the variables (reference numerals) described above.

  “Θh” represents the steering angle (rotation angle; hereinafter referred to as the steering angle) of the steering wheel 5.

  “Γdot” represents angular acceleration around the yaw axis of the vehicle 1 (hereinafter referred to as yaw angular acceleration).

  “Ωw_i” may be referred to as the rotational angular velocity of the i-th wheel 2-i (hereinafter, simply referred to as wheel rotational angular velocity), and “Rw_i” may be referred to as the effective radius of the i-th wheel 2-i (hereinafter simply referred to as effective wheel radius). ), “Vw_i” is the wheel speed of the i-th wheel 2-i defined as the product of ωw_i and Rw_i (= ωw_i * Rw_i) (that is, i-th viewed from the rotation center of the i-th wheel 2-i). (Circumferential speed of the ground contact portion of the wheel 2-i). Each wheel speed Vw_i matches the x-axis direction component Vsubx_i of the wheel traveling speed vector ↑ Vsub_i on the wheel coordinate system in a state where the i-th wheel 2-i is not slipped.

  “Κi” is a slip ratio of the i-th wheel 2-i (longitudinal slip ratio; hereinafter, simply referred to as a wheel slip ratio), and “Tq_i” is given to the i-th wheel 2-i from the drive system of the vehicle 1. The total torque of the driving torque and the braking torque applied from the braking system of the vehicle 1 (hereinafter simply referred to as wheel torque), “Iw_i” is the moment of inertia of the i-th wheel 2-i (hereinafter simply referred to as the wheel). Inertia moment).

  “M” represents the mass of the entire vehicle 1 (hereinafter referred to as vehicle mass), and “Iz” represents the moment of inertia around the yaw axis of the entire vehicle 1 at the center of gravity of the vehicle 1 (hereinafter referred to as vehicle yaw inertia moment).

  “Accx” is the X-axis direction component of the vehicle body coordinate system of the acceleration generated at the center of gravity of the vehicle 1 due to the centrifugal force accompanying the turning motion of the vehicle 1 (= -Vgy * γ) is added to represent the acceleration (= Vgdot_x−Vgy * γ). Further, “Accy” is the Y axis direction component of the vehicle body coordinate system of the acceleration generated at the center of gravity of the vehicle 1 due to the centrifugal force accompanying the turning motion of the vehicle 1 to the vehicle center of gravity side slip velocity change rate Vgdot_y. Acceleration (= Vgdot_y + Vgx * γ) obtained by adding (= Vgx * γ). In other words, “Accx” and “Accy” are respectively the X-axis directions of the acceleration of the motion of the center of gravity of the vehicle 1 as viewed in the vehicle body coordinate system (the second derivative of the position of the center of gravity in the vehicle body coordinate system). Component, Y-axis direction component. Hereinafter, “Accx” is referred to as vehicle center-of-gravity longitudinal acceleration, and “Accy” is referred to as vehicle center-of-gravity lateral acceleration.

  “Μ” represents a friction coefficient of the road surface (a friction coefficient between the wheels 2-i. Hereinafter, it may be referred to as a road surface friction coefficient). In the embodiment, the road surface friction coefficient μ is a relative value based on a friction coefficient between a road surface in a certain reference state (hereinafter referred to as a reference road surface) such as a standard dry road surface and each wheel 2-i. It is a coefficient of friction. Further, it is considered that the road surface friction coefficient μ is the same at the contact point of any wheel 2-i (i = 1, 2, 3, 4).

  “Θbank” represents a road bank angle (hereinafter, sometimes referred to as a road surface bank angle), and “θslope” represents a road surface slope angle (hereinafter, sometimes referred to as a road surface gradient angle). The road surface bank angle θbank is an inclination angle of the road surface with respect to the horizontal plane as viewed in the roll axis direction of the vehicle 1, and the road surface gradient angle θslope is an inclination angle of the road surface with respect to the horizontal plane as viewed in the pitch axis direction of the vehicle 1. The road surface bank angle θbank is generally referred to as a road surface cant angle in the field of automotive engineering, but the term bank angle is used in this specification. In the embodiment of the present specification, the road bank angle θbank when the vehicle 1 on the road surface is inclined to the right is assumed to be a positive angle. Further, the road surface slope angle θslope when the vehicle 1 on the road surface is in a forward-declining inclination posture is set as a positive angle.

“Rot (δi)” is a vector quantity (vector quantity consisting of an x-axis direction component and a y-axis direction component of the i-th wheel coordinate system) expressed in the i-th wheel coordinate system, expressed in the vehicle body coordinate system ( This represents a coordinate conversion matrix for conversion into a vector quantity consisting of an X-axis direction component and a Y-axis direction component of the vehicle body coordinate system. The coordinate transformation matrix R (δi) is a matrix (second-order square matrix) determined depending on the steering angle δi of the i-th wheel 2-i, and a column vector (cos (δi), sin (δi)) ^T , (−sin (δi), cos (δi)) A matrix in which ^T is a component in the first column and a component in the second column, respectively. In this case, if the notation of a vector quantity ↑ A in the i-th wheel coordinate system is (ax, ay) ^T and the notation in the vehicle body coordinate system is (Ax, Ay) ^T , (Ax, Ay) ^T and (ax , the relationship between the ay) ^T is a ^{(Ax, Ay) T = Rot} (δi) * (ax, ay) T. Therefore, the relationship between the wheel traveling speed vector ↑ Vi of each wheel 2-i and the wheel traveling speed vector ↑ Vsub_i on the wheel coordinate system is given by ↑ Vi = Rot (δi) * ↑ Vsub_i. Similarly, the relationship between the wheel two-dimensional road surface reaction force ↑ Fi of each wheel 2-i and the wheel coordinate system upper wheel two-dimensional road surface reaction force ↑ Fsub_i is represented by ↑ Fi = Rot (δi) * ↑ Fsub_i. Given. A coordinate transformation matrix for converting a vector quantity expressed in the vehicle body coordinate system into a vector quantity expressed in the i-th wheel coordinate system, that is, an inverse matrix of Rot (δi) is Rot (−δi).

  In the following description, when expressing an actual value (true value) state quantity, vector quantity, etc., the head of the name (name) of the state quantity, vector quantity, etc., such as “actual yaw rate”, etc. May be marked with "real". In this case, “_act” is added to the end of a variable (reference code) representing the state quantity, vector quantity, etc., such as “γ_act”. Furthermore, when expressing an observed value (detected value or estimated value) of a state quantity or vector quantity, the name of the state quantity or vector quantity (for example, “yaw rate detected value”, “yaw rate estimated value”, etc.) "Detected value" or "estimated value" is added to the end of the name. In this case, in principle, an “estimated value” is used for an observed value calculated by the vehicle model calculating means 24 described later or another observed value generated based on the calculated observed value. Further, the “detected value” is used for the observed value obtained based on the output of a certain sensor without using the observed value calculated by the vehicle model calculating means 24. “Detected value” is appended with “_sens” at the end of the variable (reference sign) such as “γ_sens”, and “estimated value” is a variable (referenced) such as “γ_estm”. Add "_estm" to the end of the sign. Further, when expressing a temporal change rate (differential value depending on time) of a state quantity, “dot” is added to a variable (reference code) of the state quantity, such as “γdot”.

  Based on the above description, embodiments of the present invention will be described in detail below.

[First Embodiment]
First, the process of the control device 20 in the first embodiment will be specifically described. In the present embodiment, as shown in the block diagram of FIG. 3, the control device 20 includes, as its main functional means, an observation target amount detection means 22, a vehicle model calculation means 24, a μ estimation means 26, and a bank angle estimation means 28. And gradient angle estimation means 30.

  The observation target amount detection means 22 executes a process of detecting a predetermined type of observation target amount related to the behavior of the vehicle 1 from the outputs (detection data) of the various sensors of the vehicle 1 and obtains the detected value of the observation target amount. Means for generating.

  In the present embodiment, the observation target quantity by the observation target quantity detection means 22 includes the actual steering angles δ1_act and δ2_act of the steered wheels (front wheels) 2-1 and 2-2 and the actual wheel speed Vw_i_act (i = 1, 2, 3, 4), the actual yaw rate γ_act and actual yaw angular acceleration γdot_act of the vehicle 1, the actual vehicle center-of-gravity longitudinal acceleration Accx_act, the actual vehicle center-of-gravity lateral acceleration Accy_act, and the actual wheel torque Tq_i_act (i = 1, 2, 3, 4). And are included.

  In order to generate detection values of these observation target amounts, the observation target amount detection means 22 includes wheel steering angle detection means 22a for generating wheel steering angle detection values δ1_sens and δ2_sens of the front wheels 2-1, 2-2, Wheel speed detection means 22b for generating wheel speed detection value Vw_i_sens (i = 1, 2, 3, 4), yaw rate detection means 22c for generating yaw rate detection value γ_sens, and yaw angle for generating yaw angular acceleration detection value γdot_sens Acceleration detection means 22d, longitudinal acceleration detection means 22e for generating vehicle center-of-gravity longitudinal acceleration detection value Accx_sens, lateral acceleration detection means 22f for generating vehicle center-of-gravity lateral acceleration detection value Accy_sens, and wheel torque detection value Tq_i_sens (i = 1, 2, 3, 4) for generating wheel torque detection means 22 g.

  The vehicle model calculation means 24 includes a friction characteristic model that expresses the relationship between the slip between each wheel 2-i and the road surface and the road surface reaction force acting on the wheel 2-i from the road surface, and the external force acting on the vehicle 1. Road surface reaction force acting on each wheel 2-i using a dynamic model of the vehicle 1 (hereinafter, simply referred to as a vehicle model) including a vehicle motion model that expresses a relationship between the vehicle 1 and the motion of the vehicle 1 And a process for estimating a state quantity of motion of the vehicle 1 that is dynamically generated by the road surface reaction force acting on the vehicle 1 as an external force. For this process, the vehicle model calculation means 24 has a detection value of a predetermined type of observation target quantity generated by the observation target quantity detection means 22 (in this embodiment, δ1_sens, δ2_sens, Vw_i_sens of the detection values). , Γ_sens, Accy_sens, Accy_sens, Tq_i_sens), and the latest road friction coefficient estimated value μ_estm already determined by the μ estimating means 26 are input. And the vehicle model calculating means 24 estimates the road surface reaction force of each wheel 2-i and the state quantity of the motion of the vehicle 1 using these input values and the vehicle model.

  The estimated value obtained by the vehicle model calculating means 24 includes a road surface reaction force estimated value that is an estimated value regarding the road surface reaction force, a translational motion of the vehicle 1 in the front-rear direction (roll axis direction) and the lateral direction (pitch axis direction), and It is roughly classified into a vehicle motion state quantity estimated value that is an estimated value of a state quantity related to the rotational motion around the yaw axis.

  In this case, the estimated road surface reaction force includes the driving / braking force Fsubx_i and lateral force Fsuby_i of each wheel 2-i and the ground load Fz_i, and the total road surface reaction force combined translational force vector estimated value ↑ Fg_total_estm (Fgx_total_estm And Fgy_total_estm) and the total road surface reaction force combined yaw moment estimated value Mgz_total_estm. In addition, the vehicle motion state amount estimated value includes a yaw rate estimated value γ_estm, a vehicle gravity center velocity vector estimated value ↑ Vg_estm (Vgx_estm and Vgy_estm), a vehicle gravity center longitudinal acceleration estimated value Accx_estm, and a vehicle gravity center lateral acceleration estimated value Accy_estm. included.

  The μ estimation means 26 is a means for executing processing for estimating the friction coefficient μ (road surface friction coefficient μ) of the road surface on which the vehicle 1 is traveling. For this processing, the μ estimation means 26 calculates δ1_sens, δ2_sens, γ_sens, γdot_sens, Accy_sens among the detection values of the observation target amount generated by the observation target amount detection means 22 and the vehicle model calculation means 24. Total road surface reaction force combined translation force vector estimated value ↑ Fg_total_estm (more specifically, all road surface reaction force combined lateral force estimated value Fgy_total_estm of ↑ Fg_total_estm) and total road surface reaction force combined yaw moment estimated value Mgz_total_estm, and vehicle model The vehicle center-of-gravity longitudinal speed estimated value Vgx_estm, which is the X-axis direction component (the vehicle 1 longitudinal direction component) Vgx_estm of the vehicle center-of-gravity velocity vector estimated value ↑ Vg_estm among the vehicle motion state amount estimated values calculated by the computing means 24, is input. Is done. Then, the μ estimation means 26 calculates a road surface friction coefficient estimated value μ_estm that is an estimated value of the road surface friction coefficient μ using these input values.

  The bank angle estimation means 28 is a means for executing a process for estimating the road surface bank angle θbank (the bank angle θbank of the road surface on which the vehicle 1 is traveling). For this processing, the bank angle estimation means 28 calculates the vehicle center-of-gravity lateral acceleration detection value Accy_sens among the detection values of the observation target quantity generated by the observation target quantity detection means 22 and the vehicle model calculation means 24. The estimated vehicle center-of-gravity lateral acceleration estimated value Accy_estm is input. Then, the bank angle estimation means 28 calculates a road surface bank angle estimated value θbank_estm that is an estimated value of the road surface bank angle θbank using these input values.

  The gradient angle estimation means 30 is a means for executing a process for estimating the road surface gradient angle θslope (the gradient angle θslope of the road surface on which the vehicle 1 is traveling). For this processing, the gradient angle estimation means 30 calculates the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens among the detection values of the observation target quantity generated by the observation target quantity detection means 22 and the vehicle model calculation means 24. The estimated vehicle acceleration / deceleration acceleration value Accx_estm is input. Then, the gradient angle estimation means 30 calculates a road surface gradient angle estimated value θslope_estm, which is an estimated value of the road surface gradient angle θslope, using these input values.

  The control device 20 uses the observation target amount detection means 22, vehicle model calculation means 24, μ estimation means 26, bank angle estimation means 28, and gradient angle estimation means 30 to perform the processing shown in the flowchart of FIG. Execute sequentially in cycles. In the following description, the values (detected values, estimated values, etc.) obtained in the current (current) computation processing cycle of the control device 20 and the previous (previous) computation processing cycle are obtained. In order to distinguish between values, the former may be referred to as “current value” and the latter as “previous value”. Then, a suffix “_p” is added to the reference sign of the previous value, for example, “γ_estm_p”. In this case, “previous value” means the latest value among the values already obtained in the past calculation processing cycle of the control device 20. A value that does not particularly refer to “current value” and “previous value” means the current value.

  Referring to FIG. 4, control device 20 first executes the process of observation target amount detection means 22 in S100. The observation target amount detection means 22 includes a wheel rotation angular velocity sensor 8-i (i = 1, 2, 3, 4), a brake pressure sensor 9-i (i = 1, 2, 3, 4), a steering steering angle. From the outputs of various sensors such as the sensor 10, the transmission sensor 11, the accelerator sensor 12, the yaw rate sensor 13, the longitudinal acceleration sensor 14, and the lateral acceleration sensor 15, the detected values δ1_sens, δ2_sens, Vw_i_sens (i = 1, 2) , 3, 4), γ_sens, γdot_sens, Accy_sens, Accy_sens, and Tq_i_sens.

  More specifically, the wheel steering angle detection values δ1_sens and δ2_sens are generated from the output of the steering steering angle sensor 10 by the wheel steering angle detection means 22a. Here, in the present embodiment, the actual rudder angle δ1_act of the first wheel 2-1 and the actual rudder angle δ2_act of the second wheel 2-2 are the same as each other, and hence δ1_sens = δ2_sens. Therefore, hereinafter, the steering angles δ1 and δ2 of the front wheels 2-1 and 2-2 are generically referred to as front wheel steering angles δf, and the wheel steering angle detection values δ1_sens and δ2_sens are generically referred to as front wheel steering angle detection values δf_sens. Then, the wheel steering angle detection means 22a calculates the steering steering angle θh and the front wheel steering angle δf from the steering steering angle detection value θh_sens which is a steering steering angle value (converted value) indicated by the output value of the steering steering angle sensor 10. A front wheel steering angle detection value δf_sens (= δ1_sens = δ2_sens) as a common steering angle detection value of the front wheels 2-1 and 2-2 is obtained based on a preset relationship (model, map, etc.).

  For example, when the steering mechanism of the vehicle 1 is configured such that the actual steering angles δ1_act and δ2_act of the front wheels 2-1 and 2-2 are substantially proportional to the actual steering steering angle θh_act, a proportional constant set in advance to θh_sens. Δf_sens is calculated by multiplying (so-called overall steering ratio).

  When the steering mechanism includes a steering actuator such as a power steering device, in addition to the steering steering angle detection value θh_sens or instead of the steering steering angle detection value θh_sens, It is also possible to detect the operating state of the actuator or the state quantity that defines it, and use the detected value to determine the detected front wheel steering angle value Δf_sens.

  Further, the steering angle detection values δ1_sens and δ2_sens of the front wheels 2-1 and 2-2 may be individually obtained by using a stricter steering system model or the like. The average value (= (δ1_sens + δ2_sens) / 2) of the detected steering angle values δ1_sens and δ2_sens of the front wheels 2-1 and 2-2 is representative of the actual steering angles δ1_act and δ2_act of the front wheels 2-1 and 2-2. You may make it obtain | require as the front-wheel steering angle detected value (delta) f_sens to do.

  The wheel speed detection value Vw_i_sens (i = 1, 2, 3, 4) is generated by the wheel speed detection means 22b from the output of the corresponding wheel rotation angular velocity sensor 8-i. Specifically, for each wheel 2-i, the wheel speed detection means 22b sets a wheel rotation angular velocity detection value ωw_i_sens that is an angular acceleration value (converted value) indicated by an output value of the wheel rotation angular velocity sensor 8-i. The wheel speed detection value Vw_i_sens is obtained by multiplying the value of the effective radius Rw_i of the i-th wheel 2-i set in advance.

  The yaw rate detection value γ_sens and the yaw angular acceleration detection value γdot_sens are respectively generated from the output of the yaw rate sensor 13 by the yaw rate detection means 22c and the yaw angular acceleration detection means 22d. That is, the yaw rate detection means 22c generates the angular velocity value (converted value) indicated by the output value of the yaw rate sensor 13 as the yaw rate detection value γ_sens. Further, the yaw angular acceleration detection means 22d differentiates the yaw rate detection value γ_sens (determines the temporal change rate) or the value of the angular acceleration indicated by the value obtained by differentiating the output value of the yaw rate sensor 13 ( Conversion value) is generated as the detected yaw angular acceleration value γdot_sens.

  Note that the yaw angular acceleration detection value γdot_sens can be generated from the output of a sensor different from the yaw rate sensor 13. For example, two acceleration sensors are mounted on the vehicle body 1B with a gap Lacc in a direction orthogonal to the yaw axis direction of the vehicle 1 (for example, the roll axis direction or the pitch axis direction of the vehicle 1). In this case, these two acceleration sensors are arranged so as to be sensitive to acceleration in a direction orthogonal to the interval direction of the two acceleration sensors and the yaw axis direction. In this case, the yaw angular acceleration detection value γdot_sens can be generated by dividing the difference between the acceleration detection values indicated by the output values of the two acceleration sensors by the interval Lacc.

  The vehicle center-of-gravity longitudinal acceleration detection value Accx_sens is generated from the output of the longitudinal acceleration sensor 14 by the longitudinal acceleration detection means 22e. Further, the vehicle center-of-gravity lateral acceleration detection value Accy_sens is generated from the output of the lateral acceleration sensor 15 by the lateral acceleration detection means 22f. Here, in the present embodiment, the position of the center of gravity of the vehicle 1 is specified in advance, and the longitudinal acceleration sensor 14 and the lateral acceleration sensor 15 are fixed to the vehicle body 1B so as to be positioned at the center of gravity. The longitudinal acceleration sensor 14 and the lateral acceleration sensor 15 may be an integral structure acceleration sensor (biaxial acceleration sensor).

  Then, the longitudinal acceleration detection means 22e generates the acceleration value (converted value) indicated by the output value of the longitudinal acceleration sensor 14 as the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens. Further, the lateral acceleration detection means 22f generates an acceleration value (converted value) indicated by the output value of the lateral acceleration sensor 15 as a vehicle center-of-gravity lateral acceleration detection value Accy_sens.

  Even when the longitudinal acceleration sensor 14 or the lateral acceleration sensor 15 is arranged at a position shifted from the center of gravity of the vehicle 1, the acceleration detection value indicated by the output value of the sensor 14 or 15 is used as the yaw angular acceleration detection. By correcting according to the value γdot_sens (or the differential value of the yaw rate detection value γ_sens), the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens or the vehicle center-of-gravity lateral acceleration detection value Accy_sens can be generated. For example, in the case where the longitudinal acceleration sensor 14 is arranged at a position having a Ly interval on the left side from the center of gravity of the vehicle 1, an acceleration detection value (acceleration at the position of the sensor 14) indicated by the output value of the longitudinal acceleration sensor 14. The vehicle center-of-gravity longitudinal acceleration detection value Accx_sens can be generated by adding the value obtained by multiplying the yaw angular acceleration detection value γdot_sens (or the differential value of the yaw rate detection value γ_sens) by Ly. Similarly, when the lateral acceleration sensor 15 is disposed at a position having a distance Lx forward from the center of gravity of the vehicle 1, the acceleration detection value (the position of the sensor 15 is indicated by the output value of the lateral acceleration sensor 15). The vehicle center-of-gravity lateral acceleration detected value Accy_sens can be generated by subtracting the value obtained by multiplying the detected yaw angular acceleration value γdot_sens (or the differential value of the yaw rate detected value γ_sens) by Lx from the detected acceleration value.

  Supplementally, the acceleration detected (sensitive) by the longitudinal acceleration sensor 14 is an acceleration vector generated at the center of gravity of the vehicle 1 by the entire external force acting on the vehicle 1 (the resultant force) (the center of gravity of the vehicle 1 by the entire external force). Of the acceleration vector obtained by dividing the acting translational force vector by the vehicle mass m), it has a meaning as a longitudinal component of the vehicle body 1B (X-axis direction component of the vehicle body coordinate system). In this case, the acceleration to which the longitudinal acceleration sensor 14 is sensitive is the actual vehicle center-of-gravity longitudinal acceleration Accx_act as an original detection target if the actual road surface slope angle θslope_act is “0”. On the other hand, when the actual road surface gradient angle θslope_act is not “0”, the longitudinal direction (X-axis direction) of the vehicle body 1B, which is the sensitive direction of the longitudinal acceleration sensor 14, has an inclination of θslope_act with respect to the horizontal plane. . For this reason, the longitudinal acceleration sensor 14 includes not only the actual vehicle center-of-gravity longitudinal acceleration Accx_act but also the acceleration component in the direction parallel to the longitudinal direction of the vehicle body 1B (= −g * sin (θslope_act). It is also sensitive to acceleration constant. Therefore, the vehicle center-of-gravity longitudinal acceleration detected value Accx_sens as the acceleration indicated by the output of the longitudinal acceleration sensor 14 is actually the actual vehicle center-of-gravity longitudinal acceleration Accx_act in the direction parallel to the longitudinal direction of the vehicle body 1B of the gravitational acceleration. This is the detected value of acceleration (= Accx_act−g * sin (θslope_act)) formed by superimposing the acceleration component (including the case of θslope_act = 0).

  Similarly to the above, the acceleration detected (sensitive) by the lateral acceleration sensor 15 is the acceleration vector generated in the center of gravity of the vehicle 1 due to the entire external force acting on the vehicle 1 (the resultant force) in the lateral direction of the vehicle body 1B. It has meaning as a component (Y-axis direction component of the vehicle body coordinate system). In this case, the acceleration to which the lateral acceleration sensor 15 is sensitive is the actual vehicle center-of-gravity lateral acceleration Acty_act as an original detection target if the actual road bank angle θbank_act is “0”. On the other hand, when the actual road surface bank angle θbank_act is not “0”, the lateral direction (Y-axis direction) of the vehicle body 1B that is the sensitive direction of the lateral acceleration sensor 15 has an inclination of θbank_act with respect to the horizontal plane. . For this reason, the lateral acceleration sensor 15 is sensitive not only to the actual vehicle center-of-gravity lateral acceleration Accy_act but also to the acceleration component (= g * sin (θbank_act)) of the gravitational acceleration in a direction parallel to the lateral direction of the vehicle body 1B. . Therefore, the vehicle center-of-gravity lateral acceleration detected value Accy_sens as the acceleration indicated by the output of the lateral acceleration sensor 15 is actually the actual vehicle center-of-gravity lateral acceleration Accy_act in the direction parallel to the lateral direction of the vehicle body 1B of the gravitational acceleration. This is a detected value of acceleration (= Accy_act + g * sin (θbank_act)) formed by superimposing acceleration components (including the case where θbank_act = 0).

  In the following description, the sum (= Accx−g * sin) of the vehicle center-of-gravity longitudinal acceleration Accx and the acceleration component (= −g * sin (θslope)) of the gravitational acceleration in the direction parallel to the longitudinal direction of the vehicle body 1B. The acceleration defined as (θslope)) (that is, the acceleration to which the longitudinal acceleration sensor 14 is sensitive) is referred to as sensor-sensitive longitudinal acceleration Accx_sensor. Similarly, it is defined as the sum (= Accx + g * sin (θbank)) of the vehicle center-of-gravity lateral acceleration Accy and the acceleration component of the gravitational acceleration in the direction parallel to the lateral direction of the vehicle body 1B (= g * sin (θbank)). The acceleration (that is, the acceleration to which the lateral acceleration sensor 15 is sensitive) is referred to as a sensor-sensitive lateral acceleration Accy_sensor, where the sensor-sensitive longitudinal acceleration Accx_sensor matches the vehicle center-of-gravity longitudinal acceleration Accx when θslope = 0. The acceleration Accy_sensor coincides with the vehicle center-of-gravity lateral acceleration Accy when θbank = 0. Therefore, the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens generated by the longitudinal acceleration detection means 22e and the vehicle generated by the lateral acceleration detection means 22f. The center-of-gravity lateral acceleration detection value Accy_sens strictly means the detection values of the sensor-sensitive longitudinal acceleration Accx_sensor and the sensor-sensitive lateral acceleration Accy_sensor, respectively.

  The wheel torque detection value Tq_i_sens (i = 1, 2, 3, 4) is calculated by the wheel torque detection means 22g from the output of the brake pressure sensor 9-i and the output of the accelerator sensor 12 and the transmission sensor 11 respectively. Generated. Specifically, the wheel torque detection means 22g recognizes the output torque (requested torque) of the engine 3 from the detected value of the depression amount of the accelerator pedal indicated by the output value of the accelerator sensor 12, and outputs the transmission sensor 4a. The reduction ratio between the engine 3 and each wheel 2-i is recognized from the detected value of the transmission ratio of the transmission 4a indicated by the value. The wheel torque detection means 22g is configured to drive torque transmitted from the engine 3 to each wheel 2-i based on the recognized output torque of the engine 3 and the reduction ratio (each wheel 2 by the drive system of the vehicle 1). -Drive torque applied to -i). Further, based on the brake pressure detection value indicated by the output value of the brake pressure sensor 9-i, the braking torque applied to each wheel 2-i from each braking mechanism 7-i (each wheel 2-by the braking system of the vehicle 1). braking torque applied to i). Then, for each wheel 2-i, the value of the sum of the obtained drive torque and braking torque (synthetic torque) is calculated as a wheel torque detection value Tq_i_sens.

  The above is the details of the processing of S100 (processing of the observation target amount detection means 22).

  In the processing of the observation target amount detection unit 22, the sensor output may be input to the detection units 22a to 22g after passing through a filter such as a high cut filter for removing high frequency noise components. . Alternatively, the detection value of the observation target amount obtained by using the output of the sensor as it is is used as a provisional detection value, and the provisional detection value is passed through a filter such as a high cut filter, so that A detection value may be generated.

  In particular, for the vehicle center-of-gravity lateral acceleration detection value Accy, means for detecting or estimating the roll angle of the vehicle body 1B (relative inclination angle around the roll axis of the vehicle body 1B with respect to the road surface) (for example, detecting the stroke of the suspension with a sensor, In the case of providing a means for calculating the roll angle of the vehicle body 1B from the detected value), the influence of the output of the lateral acceleration sensor 15 accompanying the roll motion of the vehicle body 1B (the lateral acceleration sensor 15 is inclined by the roll angle of the vehicle body 1B). Accordingly, the influence of gravity acceleration included in the output of the acceleration sensor 15 is estimated using the observed value of the roll angle, and the estimated influence is subtracted from the acceleration detection value indicated by the output value of the lateral acceleration sensor 15. Thus, it is desirable to obtain the vehicle center-of-gravity lateral acceleration detection value Accy.

  After executing the processing of the observation target amount detection unit 22 as described above, the control device 20 executes the processing of S102 to S116 by the vehicle model calculation unit 24.

  Hereinafter, this process will be described in detail with reference to FIGS.

  As shown in FIG. 5, the vehicle model calculation means 24 has, as its function, a wheel contact load estimation unit 24 a for obtaining a contact load estimated value Fz_i_estm of each wheel 2-i and a wheel coordinate system upper wheel of each wheel 2-i. A wheel friction force estimator 24b for obtaining a driving / braking force estimated value Fsubx_i_estm which is an estimated value of the x-axis direction component of the two-dimensional road reaction force ↑ Fsub_i and a lateral force estimated value Fsuby_i_estm which is an estimated value of the y-axis direction component; A resultant force calculation unit 24c for obtaining a road surface reaction force combined translational force vector ↑ Fg_total_estm and a total road surface reaction force combined yaw moment Mgz_total_estm, a vehicle motion estimation unit 24d for determining a vehicle motion state quantity estimated value, and a wheel traveling speed of each wheel 2-i Wheel estimated speed vector estimator 24e for determining vector estimated value ↑ Vi_estm, wheel motion estimator 24f for determining wheel speed estimated value Vw_i_estm for each wheel 2-i, and estimated wheel slip angle βi for each wheel 2-i A wheel side slip angle estimating unit 24g for obtaining _estm and a wheel slip rate estimating unit 24h for obtaining a wheel slip rate estimated value κi_estm of each wheel 2-i are provided.

  In the processing of S102 to S116, first, in S102, the ground contact load estimation value Fz_i_estm of each wheel 2-i is calculated by the wheel ground load estimation unit 24a.

  In this case, in this embodiment, the wheel contact load estimating unit 24a uses the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens and the vehicle center-of-gravity lateral acceleration detection value Accy_sens among the detection values of the observation target amount obtained in S100. The ground load estimated value Fz_i_estm (i = 1, 2, 3, 4) is calculated by the following equation 1-1.

Fz_i_estm = Fz0_i + Wx_i * Accx_sens + Wy_i * Accy_sens ...... Formula 1-1

Here, Fz0_i in Formula 1-1 is the value of the ground load Fz_i of the i-th wheel 2-i when the vehicle 1 is stopped (still) on a horizontal road surface (hereinafter referred to as a ground load reference value). , Wx_i is a weighting factor that defines the change in the ground load Fz_i (change from Fz0_i) of the i-th wheel 2-i depending on the vehicle center-of-gravity longitudinal acceleration Accx, and Wy_i is the i-th wheel depending on the vehicle center-of-gravity lateral acceleration Accy. 2-i is a weighting factor that defines a change in the ground load Fz_i (change from Fz0_i). These values of Fz0_i, Wx_i, and Wy_i are predetermined values set in advance.

  Therefore, Formula 1-1 is the amount of change from the ground load reference value Fz0_i (the amount of change from the ground load reference value Fz0_i) due to the acceleration of the center of gravity of the vehicle 1 (acceleration in the direction orthogonal to the yaw axis direction). ) Is obtained by a linear combination of the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens and the vehicle center-of-gravity lateral acceleration detection value Accy_sens, and a value obtained by adding the change to the ground load reference value Fz0_i is calculated as a ground load estimated value Fz_i_estm. is there.

  The relationship between the vehicle center-of-gravity longitudinal acceleration Accx and the vehicle center-of-gravity lateral acceleration Accy and the ground load Fz_i is mapped, and the vehicle center-of-gravity longitudinal acceleration detected value Accx_sens and the vehicle center-of-gravity lateral acceleration detected value Accy_sens are mapped to the map. Based on this, the ground load estimated value Fz_i_estm of each wheel 2-i may be obtained.

  Further, Fz_i_estm may be obtained by reflecting the dynamic characteristics of the suspension device (not shown) of the vehicle 1. For example, the dynamic characteristics of the suspension device of the vehicle 1 are modeled in association with the rotational motion (roll motion) around the roll axis of the vehicle body 1B and the rotational motion (pitch motion) around the pitch axis. Then, a motion state quantity related to roll motion or pitch motion, for example, an observed value of the tilt angle of the vehicle body 1B around the roll axis and its change rate, an observed value of the tilt angle of the vehicle body 1B around the pitch axis and its change rate, Using the model showing the dynamic characteristics of the suspension device, the translational force in the vertical direction (yaw axis direction) acting on each wheel 2-i from the suspension device is estimated. Then, for each wheel 2-i, the estimated translational force and the gravity acting on the wheel 2-i are added together to obtain a ground load estimated value Fz_i_estm of each wheel 2-i. In this way, the accuracy of the ground load estimated value Fz_i_estm (i = 1, 2, 3, 4) can be further increased.

  Further, when the change in the ground load Fz_i of each wheel 2-i can be considered to be sufficiently small, the processing of S102 is omitted, and the ground load estimated value Fz_i_estm is set to a predetermined value (for example, the ground load standard). The value Fz0_i) may be set.

  As described above, when the ground load estimated value Fz_i_estm (i = 1, 2, 3, 4) is determined without using the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens and the vehicle center-of-gravity lateral acceleration detection value Accy_sens, The input of Accx_sens and Accy_sens to the vehicle model calculation means 24 is not necessary.

  Next, in S104, the wheel traveling speed vector estimation unit 24e calculates the wheel traveling speed vector ↑ Vi_estm of each wheel 2-i.

In this case, the wheel traveling speed vector estimation unit 24e is a vehicle in the vehicle motion state quantity estimated value (previous value) calculated by the process of S114 (process by the vehicle motion estimation unit 24d) described later in the previous calculation processing cycle. From the center-of-gravity velocity vector estimated value ↑ Vg_estm_p (= (Vgx_estm_p, Vgy_estm_p) ^T ), the yaw rate estimated value γ_estm_p, and each wheel position vector ↑ Pi (= (Px_i, Py_i) ^T ), 2, each wheel traveling speed vector estimated value ↑ Vi_estm (= (Vx_i_estm, Vy_i_estm) ^T ) is calculated.

↑ Vi_estm = ↑ Vg_estm_p + (− Py_i * γestm_p, Px_i * γestm_p) ^T ...... Equation 1-2

Here, the second term on the right side of Expression 1-2 is the i-th wheel with respect to the center of gravity of the vehicle 1 caused by the rotational motion around the yaw axis of the vehicle 1 (rotational motion where the yaw rate value is γestm_p). 2-i relative speed (relative speed in a direction perpendicular to the yaw axis direction).

  Note that the yaw rate detection value γ_sens (previous value or current value) may be used instead of the estimated yaw rate value (previous value) γ_estm_p in Expression 1-2.

  Next, in S106, the wheel slip ratio estimation unit 24h calculates the wheel slip ratio estimated value κi_estm of each wheel 2-i.

  In this case, the wheel slip ratio estimation unit 24h detects the front wheel steering angle detection value (current value) δf_sens (= δ1_sens = δ2_sens) among the detection values of the observation target amount obtained in S100, and the later calculation processing cycle. The wheel speed estimated value (previous value) Vw_i_estm_p (i = 1, 2, 3, 4) calculated in the process of S116 (the calculation process by the wheel motion estimation unit 24f) and the wheel traveling speed vector estimation calculated in S114 Each wheel slip ratio estimated value κi_estm is calculated from the value (current value) ↑ Vi_estm (i = 1, 2, 3, 4).

  Specifically, the wheel slip ratio estimation unit 24h first converts the wheel traveling speed vector estimated value ↑ Vi_estm into the wheel coordinate system by the following equation 1-3 for each wheel 2-i, thereby Calculate the wheel traveling speed vector estimated value ↑ Vsub_i_estm on the coordinate system.

↑ Vsub_i_estm = Rot (−δi_sens) * ↑ Vi_estm …… Formula 1-3

In this case, in Expression 1-3, the front wheel steering angle detection value δf_sens is used as the values of δ1_sens and δ2_sens for the front wheels 2-1 and 2-2. In the present embodiment, since the rear wheels 2-3 and 2-4 are non-steering wheels, the values of δ3_sens and δ4_sens in Expression 1-3 are set to “0”. Accordingly, for the rear wheels 2-3 and 2-4, since ↑ Vsub_3_estm = ↑ V3_estm and ↑ Vsub_4_estm = ↑ V4_estm, the calculation process of Expression 1-3 may be omitted.

  In addition, when the wheel traveling speed vector estimated value ↑ Vsub_i_estm of each wheel coordinate system is not used in the y-axis direction component estimated value Vsuby_i_estm in the calculation processing (processing of S108, etc.) described later, the wheel traveling speed vector on each wheel coordinate system. Only the x-axis direction component estimated value Vsubx_i_estm of the estimated value ↑ Vsub_i_estm may be calculated.

  For each wheel 2-i, the wheel slip ratio estimation unit 24h calculates the wheel coordinate system wheel traveling speed vector estimated value ↑ Vsub_i_estm x-axis direction component estimated value Vsubx_i_estm calculated as described above, and the wheel speed estimated value ( The wheel slip ratio estimated value κi_estm is calculated from the previous value) Vw_i_estm_p according to the following expression 1-4.

κi_estm = (Vsubx_i_estm−Vw_i_estm_p) / max (Vsubx_i_estm, Vw_i_estm_p)
... Formula 1-4

In this case, when the vehicle 1 that applies driving force to the front wheels 2-1 and 2-2, which are drive wheels, from the drive system of the vehicle 1 is accelerated, Vsubx_i_estm ≦ Vw_i_estm_p, and therefore κi_estm ≦ 0. Further, when the vehicle 1 decelerates the braking force applied from the braking system of the vehicle 1 to each wheel 2-i, Vsubx_i_estm ≧ Vw_i_estm_p, and therefore κi_estm ≧ 0.

  Note that the wheel speed detection value Vw_i_sens (previous value or current value) may be used instead of the estimated wheel speed value (previous value) Vw_i_estm_p in Formula 1-4. In such a case, the wheel motion estimation unit 24f, which will be described in detail later, is unnecessary.

  Next, in S108, the wheel side slip angle estimating unit 24g calculates the wheel side slip angle estimated value βi_estm of each wheel 2-i.

  In this case, the wheel skid angle estimation unit 24g detects the front wheel steering angle detection value δf_sens (= δ1_sens = δ2_sens) among the detection values of the observation target amount obtained in S100, and the wheel traveling speed vector estimation value calculated in S104. ↑ Each wheel side slip angle estimated value βi_estm is calculated from Vi_estm (i = 1, 2, 3, 4).

  Specifically, the wheel side slip angle estimation unit 24g first calculates, for each wheel 2-i, from the X-axis direction component estimated value Vx_i_estm and the Y-axis direction component estimated value Vy_i_estm of the wheel speed progress velocity vector estimated value ↑ Vi_estm. A wheel position side slip angle estimated value β0i_estm is calculated by Formula 1-5.

β0i_estm = tan ⁻¹ (Vy_i_estm / Vx_i_estm) ...... Formula 1-5

Then, for each wheel 2-i, the wheel side slip angle estimating unit 24g estimates the wheel side slip angle from the wheel position side slip angle estimated value β0i_estm calculated as described above and the rudder angle detected value δi_sens by the following equation 1-6. The value βi_estm is calculated.

βi_estm = β0i_estm−δi_sens ...... Formula 1-6

In this case, in Expression 1-6, the front wheel steering angle detection value δf_sens is used as the values of δ1_sens and δ2_sens for the front wheels 2-1 and 2-2. In the present embodiment, since the rear wheels 2-3 and 2-4 are non-steering wheels, the values of δ3_sens and δ4_sens in Expression 1-6 are set to “0”. Therefore, β3_estm = β03_estm and β4_estm = β04_estm.

  It should be noted that the wheel side slip angle is calculated by the following equation 1-7 from the estimated value Vsubx_i_estm of the wheel traveling speed vector estimated value ↑ Vsub_i_estm on the wheel coordinate system calculated by the equation 1-3 and the y-axis direction component estimated value Vsuby_i_estm. The estimated value βi_estm may be calculated.

βi_estm = tan ⁻¹ (Vsuby_i_estm / Vsubx_i_estm) ...... Formula 1-7

Next, in S110, the wheel frictional force estimating unit 24b calculates an upper wheel two-dimensional road surface reaction force estimated value ↑ Fsub_i (= (Fsubx_i_estm, Fsuby_i_estm) ^T ) of each wheel 2-i.

  Here, the wheel frictional force estimation unit 24b includes a friction characteristic model that expresses a relationship between slippage between the wheels 2-i and the road surface reaction force acting on the wheels 2-i from the road surface. In the present embodiment, the friction characteristic model includes a driving / braking force Fsubx_i and a lateral force Fsuby_i in the wheel coordinate system upper wheel two-dimensional road surface reaction force ↑ Fsub_i as a friction force acting on each wheel 2-i from the road surface. As shown in the following formulas 1-8 and 1-9, the wheel slip ratio κi and the wheel slip angle βi indicating the slip state of the i-th wheel 2-i, the ground load Fz_i, and the road surface friction coefficient μ are input. It is a model expressed as a function as a parameter.

Fsubx_i = func_fxi (κi, βi, Fz_i, μ) ...... Formula 1-8
Fsuby_i = func_fyi (κi, βi, Fz_i, μ) ...... Equation 1-9

In this case, the function func_fxi (κi, βi, Fz_i, μ) on the right side of Expression 1-8, that is, the function func_fx_i that defines the relationship between Fsubx_i and κi, βi, Fz_i, and μ is In the example, it is represented by the following formula 1-8a.

func_fx_i (κi, βi, Fz_i, μ) = μ * Cslp_i (κi) * Cattx_i (βi) * Fz_i
... Formula 1-8a

In this expression 1-8a, Cslp_i (κi) is a coefficient that defines the change characteristic of the driving / braking force Fsubx_i accompanying the change in the wheel slip ratio κi, and Cattx_i (βi) is the change in the wheel skid angle βi (and hence the lateral force). This is a coefficient that defines the change characteristic of the driving / braking force Fsubx_i accompanying the change in Fsuby_i. The relationship between Cslp_i (κi) and κi is set, for example, as shown in the graph of FIG. That is, the relationship is set so that the coefficient Cslp_i (κi) becomes a monotone decreasing function with respect to the wheel slip rate κi. More specifically, in the situation where κi> 0 (the situation when the vehicle 1 is decelerated), the value of the function func_fx_i (= driving / braking force Fsubx_i) becomes negative ( In the situation where the braking force increases (in the direction of increasing braking force) and κi <0 (the situation during acceleration of the vehicle 1), the value of the function func_fx_i (= drive · The relationship between Cslp_i (κi) and κi is set so that the braking force Fsubx_i changes in the positive direction (increase in driving force). In the relationship shown in FIG. 6A, the coefficient Cslp_i (κi) has a saturation characteristic with respect to the wheel slip rate κi. That is, as the absolute value of κi increases, the ratio of the change in Cslp_i (κi) to the change in κi (the value obtained by differentiating Cslp_i (κi) by κi) decreases.

  Further, the relationship between the coefficient Cattx_i (βi) and the wheel side slip angle βi is set as shown in the graph of FIG. 6B, for example. That is, this relationship is set so that the value of the coefficient Cattx_i (βi) approaches “0” from “1” as the absolute value of the wheel slip angle βi increases from “0”. In other words, the relationship between Cattx_i (βi) and βi is reduced so that the value of the function func_fx_i (= drive / braking force Fsubx_i) decreases as the absolute value of the wheel slip angle βi increases. Is set. This corresponds to the fact that the magnitude of the lateral force Fsuby_i generally increases and the magnitude of the driving / braking force Fsubx_i decreases as the absolute value of the wheel side slip angle βi increases.

  Therefore, the friction characteristic model represented by the expressions 1-8 and 1-8a is that the driving / braking force Fsubx_i of the i-th wheel 2-i is proportional to the road surface friction coefficient μ and the ground load Fz_i, and the wheel slip In this model, Fsubx_i becomes a monotonically decreasing function with respect to the rate κi, and the magnitude of Fsubx_i decreases as the absolute value of the wheel slip angle βi increases.

  Supplementally, the friction characteristic model represented by the equations 1-8 and 1-8a in this way corresponds to the first model of the friction characteristic models in the present invention.

  The function func_fyi (κi, βi, Fz_i, μ) on the right side of Expression 1-9, that is, the function func_fyi that defines the relationship between Fsuby_i and κi, βi, Fz_i, and μ is an example of this embodiment. Then, it is represented by the following formula 1-9a.

func_fy_i (κi, βi, Fz_i, μ) = μ * Cbeta_i (βi) * Catty_i (κi) * Fz_i
... Formula 1-9a

In this equation 1-9a, Cbeta_i (βi) is a coefficient that defines the change characteristic of the lateral force Fsuby_i accompanying the change in the wheel side slip angle βi, and Catty_i (κi) is the change in the wheel slip ratio κi (and hence the driving / braking force). This is a coefficient that defines the change characteristic of the lateral force Fsuby_i accompanying the change of Fsubx_i. The relationship between Cbeta_i (βi) and βi is set as shown in the graph of FIG. 7A, for example. That is, the relationship is set so that the coefficient Cbeta_i (βi) is a monotonically decreasing function with respect to the wheel side slip angle βi. More specifically, in a situation where βi> 0 (a situation where Vsuby_i> 0), the value of the function func_fy_i (= lateral force Fsuby_i) becomes negative (i-th wheel) as the wheel slip angle βi increases. 2-i to the right) and when βi <0 (situation where Vsuby_i <0), the value of the function func_fy_i (= lateral force Fsuby_i) increases as the wheel slip angle βi increases. ) Increases in the positive direction (to the left of the i-th wheel 2-i), the relationship between Cbeta_i (βi) and βi is set. In the relationship shown in FIG. 7A, the coefficient Cbeta_i (βi) has a saturation characteristic with respect to the wheel side slip angle βi. That is, as the absolute value of βi increases, the ratio of the change in coefficient Cbeta_i (βi) to the change in βi (the value obtained by differentiating Cbeta_i (βi) with βi) decreases.

  Further, the relationship between the coefficient Catty_i (κi) and the wheel slip rate κi is set as shown in the graph of FIG. 7B, for example. That is, the relationship is set so that the value of the coefficient Catty_i (κi) approaches “0” from “1” as the absolute value of the wheel slip ratio κi increases from “0”. In other words, the relationship between Cattyx_i (κi) and κi is set so that the magnitude of the lateral force Fsuby_i as the value of the function func_fy_i decreases as the absolute value of the wheel slip ratio κi increases. Yes. This corresponds to the fact that the magnitude of the driving / braking force Fsubx_i generally increases and the magnitude of the lateral force Fsuby_i decreases as the absolute value of the wheel slip ratio κi increases.

  Therefore, the friction characteristic model represented by the expressions 1-9 and 1-9a is such that the lateral force Fsuby_i of the i-th wheel 2-i is proportional to the road surface friction coefficient μ and the ground load Fz_i, and the wheel skid angle βi In contrast, Fsuby_i is a monotonically decreasing function, and the Fsuby_i becomes smaller as the absolute value of the wheel slip ratio κi becomes larger.

  Supplementally, the friction characteristic model represented by the equations 1-9 and 1-9a corresponds to the second model of the friction characteristic models in the present invention.

  In S110, the wheel frictional force estimator 24b obtains the wheel coordinate system upper wheel two-dimensional road surface reaction force estimated value ↑ Fsub_i of each wheel 2-i using the friction characteristic model set as described above. Specifically, for each wheel 2-i, the wheel frictional force estimating unit 24b calculates the wheel slip rate estimated value κi_estm calculated in S106, the wheel side slip angle estimated value βi_estm calculated in S108, and calculated in S102. The ground contact load estimated value Fz_i_estm and the road surface friction coefficient estimated value μ_estm_p calculated in the process of S122 (calculated by the μ estimator 26) described later in the previous calculation processing cycle are respectively represented by the function func_fxi (κi, βi). , Fz_i, μ) and func_fyi (κi, βi, Fz_i, μ) are used as input parameter values to perform the calculation of the right side of the expression 1-8a and the calculation of the right side of the expression 1-9a. Then, the wheel frictional force estimating unit 24b drives the value of the function func_fxi obtained by the calculation of Expression 1-8a as the x-axis direction component estimated value of the wheel two-dimensional road surface reaction force estimated value ↑ Fsub_i. -Estimated braking force Fsubx_i_estm. Further, the wheel frictional force estimation unit 24b uses the value of the function func_fyi obtained by the calculation of Expression 1-9a as the y-axis direction component estimated value of the wheel coordinate system upper wheel two-dimensional road surface reaction force estimated value ↑ Fsub_i. The force estimated value is Fsuby_i_estm. In this case, the value of the coefficient Cslp_i (κi) necessary for the calculation of the right side of the expression 1-8a is determined based on the map representing the relationship shown in FIG. 6A from the wheel slip rate estimated value κi_estm. Further, the value of the coefficient Catty_i (βi) necessary for the calculation of the right side of the expression 1-8a is determined from the estimated wheel side slip angle βi_estm based on the map representing the relationship shown in FIG. Further, the value of Cbeta_i (βi) necessary for the calculation of the right side of the expression 1-9a is determined based on the wheel lateral slip ratio estimated value βi_estm based on the map representing the relationship shown in FIG. Further, the value of the coefficient Cattx_i (κi) necessary for the calculation of the right side of Expression 1-9a is determined from the wheel slip ratio estimated value κi_estm based on the map representing the relationship shown in FIG.

  As described above, of the road surface reaction force acting on each wheel 2-i, the estimated driving / braking force value Fsubx_i_estm and the lateral force estimation as the estimated value of the road surface reaction force (friction force) having dependency on the road surface friction coefficient μ. The value Fsuby_i_estm is calculated using the latest value (previous value μ_estm_p) of the road surface friction coefficient estimated value μ_estm and the friction characteristic model.

  Supplementally, in the present embodiment, the function func_fx_i is set so that the driving / braking force Fsubx_i of each wheel 2-i is proportional to the road surface friction coefficient μ. For example, the function func_fx_i is set by the following expression 1-8b. May be.

func_fx_i (κi, βi, Fz_i, μ) = Cslp2_i (μ, κi) * Cattx_i (βi) * Fz_i
... Formula 1-8b

Cslp2_i (μ, κi) in the expression 1-8b is a coefficient that defines the change characteristics of the driving / braking force Fsubx_i accompanying changes in the road surface friction coefficient μ and the wheel slip ratio κi, and μ * in the expression 1-8a. It is a more generalized version of Cslp_i (κi). In this case, the relationship between the coefficient Cslp2_i (μ, κi), the road surface friction coefficient μ, and the wheel slip ratio κi is set as shown in the graph of FIG. This relationship is set so that the coefficient Cslp2_i (μ, κi) becomes a monotonically decreasing function with respect to the wheel slip ratio κi, and at the same time the absolute value thereof becomes a monotonically increasing function with respect to the road surface friction coefficient μ. FIG. 8 representatively illustrates a graph of Cslp2_i (μ, κi) corresponding to three types of road surface friction coefficient μ. In the relationship shown in FIG. 8, the coefficient Cslp2_i (μ, κi) has a saturation characteristic with respect to the wheel slip ratio κi. That is, when the absolute value of κi increases, the magnitude of the rate of change of the coefficient Cslp_i (μ, κi) with respect to the increase of κi (the value obtained by partial differentiation of Cslp_i (μ, κi) with κi) becomes the absolute value of κi. It becomes small with the increase of.

  When the function func_fx_i is set as described above, a non-linear relationship can be set between the driving / braking force Fsubx_i of each wheel 2-i and the road surface friction coefficient μ.

  As for the function func_fy_i related to the lateral force Fsuby_i of each wheel 2-i, as in the case of the function func_fx_i related to the driving force Fsubx_i, instead of μ * Cbeta_i (βi) in Expression 1-9a, the road surface friction coefficient μ Also, a coefficient Cbeta2_i (μ, βi) that defines the change characteristic of the lateral force Fsuby_i accompanying the change of the wheel side slip angle βi may be used.

  The function func_fy_i related to the lateral force Fsuby_i of each wheel 2-i may be configured with the driving / braking force Fsubx_i as an input parameter instead of the wheel slip rate κi. In this case, as the value of Fsubx_i, the driving / braking force estimated value Fsubx_i_estm obtained as described above by the function func_fx_i of the expression 1-8a or the expression 1-8b may be used. The braking force detection value Fsubx_i_sens may be used. That is, in S100, based on the wheel torque detection value Tq_i_sens of each wheel 2-i generated by the observation target amount detection means 22 and the wheel speed detection value Vw_i_sens, the driving / braking force is detected by the following expression 1-8c. The value Fsubx_i_sens is obtained.

Fsubx_i_sens = Tq_i_sens / Rw_i−Vwdot_i_sens * Iw_i / Rw_i ² Equation 1-8c

Vwdot_i_sens on the right side of Expression 1-8c is a temporal change rate (differential value) of the wheel speed detection value Vw_i_sens. In addition, as the values of the wheel effective radius Rw_i and the wheel inertia moment Iw_i in Expression 1-8c, predetermined values set in advance are used. The second term on the right side of Equation 1-8c may be replaced with the term ωwdot_i_sens * Iw_i / Rw_i that uses ωwdot_i_sens, which is a differential value of the wheel rotational angular velocity detection value ωw_i_sens indicated by the output value of the wheel rotational angular velocity sensor 8-i. Good.

  Returning to the description of FIG. 4, next, in S112, the resultant force calculation unit 24c calculates the total road surface reaction force combined translational force vector estimated value ↑ Fg_total_estm and the total road surface reaction force combined yaw moment estimated value Mgz_total_estm.

  In this case, the resultant force calculation unit 24c calculates the ground load estimated value Fz_i_estm of each wheel 2-i calculated in S102, and the driving / braking force estimated value Fsubx_i_estm and lateral force estimated value of each wheel 2-i calculated in S110. From Fsuby_i_estm and the front wheel rudder angle detection value δf_sens (= δ1_sens = δ2_sens) among the detection values of the observation target amount obtained in S100, the total road surface reaction force combined translational force vector estimated value ↑ Fg_total_estm and the total road surface reaction force combination The yaw moment estimated value Mgz_total_estm is calculated.

Specifically, the resultant force calculation unit 24c first calculates, for each wheel 2-i, a two-dimensional road surface reaction force vector estimated value ↑ Fsub_i_estm (= (Fsubx_i_estm, Fsuby_i_estm) ^T ) on the wheel coordinate system according to the following formula 1- 10 to calculate the two-dimensional road surface reaction force vector estimated value ↑ Fi_estm = (Fx_i_estm, Fy_i_estm) ^T.

↑ Fi_estm = Rot (δi_sens) * ↑ Fsub_i_estm ...... Formula 1-10

In this case, in Formula 1-10, the front wheel steering angle detection value δf_sens is used as the values of δ1_sens and δ2_sens for the front wheels 2-1 and 2-2. In the present embodiment, since the rear wheels 2-3 and 2-4 are non-steering wheels, the values of δ3_sens and δ4_sens in Expression 1-10 are set to “0”. Therefore, for rear wheels 2-3 and 2-4, since ↑ F3_estm = ↑ Fsub_3_estm and ↑ F4_estm = ↑ Fsub_4_estm, the arithmetic processing of Expression 1-10 may be omitted.

Next, the resultant force calculation unit 24c calculates the total road surface reaction force combined translational force vector estimated value ↑ Fg_total_estm (= (Fgx_total_estm, Fgy_total_estm, Fgz_total_estm) ^T ) by the following expression 1-11, and also calculates the total road surface by the following expression 1-12. Reaction force synthetic yaw moment estimated value Mgz_total_estm is calculated.

↑ Fg_total_estm = (ΣFx_i_estm, ΣFy_i_estm, ΣFz_i_estm) ^T ...... Equation 1-11
Mgz_total_estm = Σ (↑ Pi × ↑ Fi_estm) ...... Formula 1-12

Note that “Σ” in Equations 1-11 and 1-12 means the sum of all the wheels 2-i (i = 1, 2, 3, 4). Further, ↑ Pi × ↑ Fi_estm in the right side of Expression 1-12 is the outer product of the wheel position vector ↑ Pi of the i-th wheel 2-i and the two-dimensional road surface reaction force vector estimated value ↑ Fi_estm. By the two-dimensional road surface reaction force vector estimated value ↑ Fi_estm of the wheel 2-i, it means a moment around the yaw axis generated at the center of gravity of the vehicle 1.

  Supplementally, calculation of the yaw axis direction component Fgz_total_estm in ↑ Fg_total_estm may be omitted.

  Next, in S114, the vehicle motion estimation unit 24e causes the vehicle motion center amount estimated value Vgx_estm, the vehicle center of gravity side slip velocity estimated value Vgy_estm, the yaw rate estimated value γ_estm, the vehicle center of gravity longitudinal acceleration estimated value Accx_estm, A vehicle center-of-gravity lateral acceleration estimated value Accy_estm and the like are calculated.

  Here, the vehicle motion estimation unit 24 e includes a vehicle motion model that represents the relationship between the resultant force of the road surface reaction force as an external force acting on the vehicle 1 and the motion of the vehicle 1. In this embodiment, the vehicle motion model is represented by the following expressions 1-13 to 1-15.

Fgx_total = m * (Vgdot_x−Vgy * γ) ...... Formula 1-13
Fgy_total = m * (Vgdot_y + Vgx * γ) ...... Formula 1-14
Mgz_total = Iz * γdot ...... Formula 1-15

Equations 1-13 and 1-14 represent dynamic equations relating to the translational motion of the center of gravity of the vehicle 1 in the X-axis direction and the Y-axis direction of the vehicle body coordinate system, respectively. Equation 1-15 represents a dynamic equation relating to the rotational motion of the vehicle 1 around the yaw axis. Note that the vehicle motion model in the present embodiment is a model on the assumption that the road surface on which the vehicle 1 is traveling is a horizontal plane (the road surface bank angle θbank and the road surface slope angle θslope are both “0”).

  In S114, the vehicle motion estimator 24d, the vehicle motion model represented by the above formulas 1-13 to 1-15, the total road surface reaction force combined translational force vector estimated value ↑ Fg_total_estm and the total road surface reaction force calculated in S112. A vehicle motion state estimated value is calculated using the synthesized yaw moment estimated value Mgz_total_estm. In this case, for some vehicle motion state quantity estimated values, the previous values of the some vehicle motion state quantity estimated values are also used to calculate them. Further, for some vehicle motion state quantity estimated values, the part of vehicle motion state quantity estimated values are calculated so as to approach the detected values obtained in S100 (so as not to deviate from the detected values). .

  Specifically, the vehicle motion estimator 24d uses the following equations 1-13a to 1-15a obtained from the equations 1-13 to 1-15, respectively, to estimate the vehicle center-of-gravity longitudinal velocity change rate Vgdot_x_estm and the vehicle center of gravity. Side slip velocity change rate estimated value Vgdot_y_estm and yaw angular acceleration estimated value γdot_estm are calculated. Further, the vehicle motion estimation unit 24d uses the following expressions 1-16a and 1-17a according to the definitions of the vehicle center-of-gravity longitudinal acceleration Accx and the vehicle center-of-gravity lateral acceleration Accy, respectively, to estimate the vehicle center-of-gravity longitudinal acceleration estimated value Accx_estm, Calculate the value Accy_estm.

Vgdot_x_estm = Fgx_total_estm / m + Vgy_estm_p * γ_estm_p ...... Equation 1-13a
Vgdot_y_estm = Fgy_total_estm / m−Vgx_estm_p * γ_estm_p ...... Equation 1-14a
γdot_estm = Mgz_total_estm / Iz Equation 1-15a
Accx_estm = Vgdot_x_estm−Vgy_estm_p * γ_estm_p ...... Equation 1-16a
Accy_estm = Vgdot_y_estm + Vgx_estm_p * γ_estm_p ...... Equation 1-17a

In this case, Fx_total_estm, Fy_total_estm, and Mgz_total_estm in Equations 1-13a to 1-15a are the values calculated in S112 (current value), and Vgy_estm_p, Vgx_estm_p, and γ_estm_p are determined in S114 in the previous calculation processing cycle, respectively. Value (previous value). Further, Vgdot_x_estm in Expression 1-16a and Vgdot_y_estm in Expression 1-17a are values (current values) calculated by Expressions 1-13a and 1-14a, respectively. Further, predetermined values set in advance are used as the value of the vehicle mass m in the expressions 1-13a and 1-14a and the value of the vehicle yaw inertia moment Iz in the expression 1-15a.

  Supplementally, the yaw rate detection value γ_sens (previous value or current value) may be used instead of the estimated yaw rate value (previous value) γ_estm_p in Equations 1-13a and 1-14a. Further, the vehicle center-of-gravity longitudinal acceleration estimated value Accx_estm and the vehicle center-of-gravity lateral acceleration estimated value Accy_estm may be obtained by the calculation of the first term on the right side of Equation 1-13a and the first term on the right side of Equation 1-14a, respectively. Good. That is, Accx_estm and Accy_estm may be calculated by the following equations 1-16b and 1-17b, respectively.

Accx_estm = Fgx_total_estm / m ...... Equation 1-16b
Accy_estm = Fgy_total_estm / m ...... Formula 1-17b

Next, the vehicle motion estimator 24d calculates the vehicle center-of-gravity longitudinal velocity change rate estimated value Vgdot_x_estm, the vehicle center-of-gravity skid velocity change rate estimated value Vgdot_y_estm, the yaw angular acceleration estimated value γdot_estm, and the vehicle center-of-gravity longitudinal velocity estimated value obtained as described above. The vehicle center-of-gravity longitudinal speed estimation from the previous value Vgx_estm_p, the vehicle center-of-gravity skid speed estimated value Vgy_estm_p, and the yaw rate estimated value previous value γ_estm_p according to the following equations 1-18, 1-19, and 1-20, respectively: Vehicle center-of-gravity forward / backward speed provisional estimated value Vgx_predict, vehicle center-of-gravity skid speed provisional estimated value Vgy_predict, and yaw rate provisional estimated value γ_predict as provisional value of yaw rate estimated value And calculate.

Vgx_predict = Vgx_estm_p + Vgdot_x_estm * ΔT (Formula 1-18)
Vgy_predict = Vgy_estm_p + Vgdot_y_estm * ΔT (Formula 1-19)
γ_predict = γ_estm_p + γdot_estm * ΔT Equation 1-20

In addition, ΔT in Expressions 1-18 to 1-20 is an arithmetic processing cycle of the control device 20. The right sides of these equations 1-18 to 1-20 correspond to the integral operation of Vgdot_x_estm, the integral operation of Vgdot_y_estm, and the integral operation of γdot_estm, respectively.

  Here, in the present embodiment, the vehicle motion estimation unit 24d makes the yaw rate estimated value γ_estm close to the yaw rate detected value γ_sens with respect to the yaw rate γ of the motion state amount to be estimated (so as not to deviate from γ_sens). decide. The vehicle motion estimation unit 24d also uses the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm as the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm (i = 1, 2, 3, 4) for the vehicle center-of-gravity longitudinal speed Vgx. ) Is determined so as to approach the recognized vehicle center-of-gravity longitudinal speed (so as not to deviate from the recognized vehicle gravity center longitudinal speed).

  Therefore, the vehicle motion estimation unit 24d obtains the yaw rate deviation γestm_err as the deviation between the yaw rate detection value γ_sens obtained in S100 and the yaw rate provisional estimated value γ_predict calculated by Equation 1-20 as described above with respect to the yaw rate γ. Calculated according to 1-21. Further, the vehicle motion estimation unit 24d detects and detects the wheel speed that is one of the wheel speed detection values Vw_i_sens (i = 1, 2, 3, 4) obtained in S100 with respect to the vehicle center-of-gravity longitudinal speed Vgx. A vehicle speed deviation Vgx_estm_err as a deviation between the value Vw_i_sens_select and the vehicle longitudinal speed temporary estimated value Vgx_predict calculated by the expression 1-18 as described above is calculated by the following expression 1-22. The wheel speed selection detection value Vw_i_sens_select corresponds to the detection value of the actual vehicle speed based on the wheel speed detection value Vw_i_sens (i = 1, 2, 3, 4) (the detection value of the actual vehicle center-of-gravity longitudinal speed Vgx_act). This is a value selected from the speed detection value Vw_i_sens (i = 1, 2, 3, 4). In this case, when the vehicle 1 is accelerated, the slowest wheel speed detection value among the wheel speed detection values Vw_i_sens (i = 1, 2, 3, 4) is selected as the selected wheel speed detection value Vw_i_sens_select. Further, when the vehicle 1 is decelerated, the fastest wheel speed detection value among the wheel speed detection values Vw_i_sens (i = 1, 2, 3, 4) is selected as the selected wheel speed detection value Vw_i_sens_select.

γestm_err = γ_sens−γ_predict ...... Equation 1-21
Vgx_estm_err = Vw_i_sens_select−Vgx_predict (Formula 1-22)

Next, the vehicle motion estimator 24d calculates the final vehicle center-of-gravity longitudinal velocity estimated value Vgx_estm, vehicle center-of-gravity skid velocity estimated value Vgy_estm, and yaw rate estimated value according to the following equations 1-23 to 1-25, respectively. Determine γ_estm.

Vgx_estm = Vgx_predict + Kvx * Vgx_estm_err ...... Equation 1-23
Vgy_estm = Vgy_predict ...... Formula 1-24
γ_estm = γ_predict + Kγ * γestm_err ...... Formula 1-25

Note that Kvx in Equation 1-23 and Kγ in Equation 1-25 are gain coefficients of a predetermined value (<1) set in advance.

  As shown in these formulas 1-23 to 1-25, in this embodiment, the vehicle center-of-gravity longitudinal speed provisional estimated value Vgx_predict (estimated value on the vehicle motion model) calculated by the formula 1-18 is In accordance with the vehicle speed deviation Vgx_estm_err calculated by the equation 1-22, the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm is corrected by a feedback control law (here, a proportional law) so that the vehicle speed deviation Vgx_estm_err approaches 0. Is determined. Further, the vehicle center-of-gravity skid speed provisional estimated value Vgy_predict (estimated value on the vehicle motion model) calculated by Expression 1-19 is directly determined as the vehicle center-of-gravity skid speed estimated value Vgy_estm. Further, the yaw rate provisional estimated value γ_predict (estimated value on the vehicle motion model) calculated by the equation 1-20 is set to the yaw rate deviation γ_estm_err according to the yaw rate deviation γ_estm_err calculated by the equation 1-21. The yaw rate estimated value γ_estm is determined by correcting the feedback control law (here, the proportional law) so as to approach 0 ″. As described above, in this embodiment, the vehicle center longitudinal velocity estimated value Vgx_estm as the vehicle speed of the vehicle 1 on the vehicle motion model is not deviated from the wheel speed selection detection value Vw_i_sens_select as the actual vehicle speed detection value (Vw_i_sens_select and To match or nearly match). Further, the yaw rate estimated value γ_estm as the yaw rate of the vehicle 1 on the vehicle motion model is determined so as not to deviate from the yaw rate detected value γ_sens as the detected value of the actual yaw rate γ_act (so as to coincide with or substantially coincide with γ_sens). .

  The above is the detail of the process of S114 (process of the vehicle motion estimation part 24d).

  Note that the vehicle motion estimation unit 24d in the present embodiment uses the vehicle center-of-gravity longitudinal speed estimation value Vgx_estm and the yaw rate estimation value γ_estm from the wheel speed selection detection value Vw_i_sens_select (actual vehicle speed detection value) and the yaw rate detection value γ_sens, respectively. Although it has been determined not to diverge, either one or both may always match Vw_i_sens_select and γ_sens. In this case, the process for calculating Vgx_estm or γ_estm is not necessary.

  In addition, the vehicle motion estimation unit 24d calculates Vgdot_x_estm, Vgx_estm, Vgdot_y_estm, Vgy_estm, γ_estm, Accx_estm, and Accy_estm as vehicle motion state amount estimated values. May be further requested. For example, when the vehicle gravity center side slip angle βg is controlled using the vehicle motion state estimated value, the vehicle gravity center side slip angle estimated value βg_estm may be calculated. In this case, the vehicle center-of-gravity skid angle estimated value βg_estm can be calculated by the following equation 1-26 from the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm and the vehicle center-of-gravity skid speed estimated value Vgy_estm obtained as described above.

βg_estm = tan ⁻¹ (Vgy_estm / Vgx_estm) ...... Formula 1-26

Of the vehicle motion state quantity estimated values obtained by the vehicle motion estimator 24d, Accx_estm and Accy_estm are used by the gradient angle estimator 30 and the bank angle estimator 28 described later in detail in this embodiment. Yes, it is not used in the processing of the vehicle model calculation means 24. Therefore, the calculation of Accx_estm and Accy_estm may be performed by the gradient angle estimation means 30 and the bank angle estimation unit 28, respectively.

  Next, in S116 of FIG. 4, the wheel motion estimation unit 24f calculates a wheel speed estimated value Vw_i_estm of each wheel 2-i.

  Here, the wheel motion estimation unit 24f represents a wheel motion model that represents the relationship between the force (wheel torque Tq_i and driving / braking force) acting on each wheel 2-i and the rotational motion of the wheel 2-i. I have. In this embodiment, this wheel motion model is a model expressed by the following expression 1-27.

Tq_i−Fsubx_i * Rw_i = Iw_i * (Vwdot_i / Rw_i) (Formula 1-27)

“Vwdot_i” in Expression 1-27 is a temporal change rate (differential value) of the wheel speed Vw_i of the i-th wheel 2-i, and is hereinafter referred to as a wheel speed change rate. Further, the left side of the expression 1-27 indicates the wheel torque Tq_i applied to the i-th wheel 2-i from one or both of the driving system and the braking system of the vehicle 1, and the driving / braking force Fsubx_i of the i-th wheel 2-i. Means the combined torque with the torque applied to the wheel 2-i.

  Then, the wheel motion estimation unit 24f first calculates the wheel speed change rate estimated value Vwdot_i_estm of each wheel 2-i by the following expression 1-27a obtained based on the expression 1-27.

Vwdot_i_estm = Rw_i * (Tq_i_sens−Fsubx_i_estm * Rw_i) / Iw_i
... Formula 1-27a

In this case, Tq_i_sens in Expression 1-27a is the detection value (current value) obtained in S100 for each wheel 2-i, and Fsubx_i_estm is the value (current value) obtained in S110 for each wheel 2-i. . A predetermined value set in advance is used as the value of the wheel effective radius Rw_i and the wheel inertia moment Iw_i of each wheel 2-i.

  Next, the wheel motion estimation unit 24f calculates the wheel speed change rate estimated value Vwdot_i_estm obtained as described above and the previous value Vw_i_estm_p of the wheel speed estimated value for each wheel 2-i by the following formula 1-28. A wheel speed provisional estimated value Vw_i_predict is calculated as a provisional value of the speed estimated value.

Vw_i_predict = Vw_i_estm_p + Vwdot_i_estm * ΔT (Formula 1-28)

Expression 1-28 corresponds to the integral calculation of Vwdot_i_estm.

  Here, in the present embodiment, the wheel motion estimation unit 24f makes the wheel speed estimation value Vw_i_estm close to the wheel speed detection value Vw_i_sens, similarly to the case where the vehicle motion estimation unit 24d calculates the yaw rate estimation value γ_estm and the like. (Do not deviate from Vw_i_sens).

  Therefore, the wheel motion estimator 24f provides, for each wheel 2-i, a deviation between the wheel speed estimated value Vw_i_sens obtained in S110 and the wheel speed provisional estimated value Vw_i_predict calculated by Expression 1-28 as described above. The wheel speed deviation Vw_i_estm_err is calculated by the following equation 1-29.

Vw_i_estm_err = Vw_i_sens−Vw_i_predict …… Equation 1-29

Next, the wheel motion estimation unit 24f determines a final wheel speed estimation value Vw_i_estm in the current calculation processing cycle by the following equation 1-30 for each wheel 2-i.

Vw_i_estm = Vw_i_predict + Kvw * Vw_i_estm_err ...... Equation 1-30

In Equation 1-30, Kvw is a predetermined predetermined gain coefficient (<1).

  Therefore, in the present embodiment, each wheel speed provisional estimated value Vw_i_predict (estimated value on the wheel motion model) calculated by the expression 1-28 is set in accordance with the wheel speed deviation Vw_i_estm_err calculated by the expression 1-29. Thus, each wheel speed estimated value Vw_i_estm is determined by correcting the wheel speed deviation Vw_i_estm_err by a feedback control law (proportional law in this case) so as to approach “0”.

  The processing of S102 to S116 described above is the details of the processing of the vehicle model calculation means 24.

  Next, the control apparatus 20 performs the process of the bank angle estimation means 28 in S118.

  Here, the vehicle motion estimation unit 24d calculates a vehicle motion state quantity estimated value using a vehicle motion model constructed on the assumption that the road surface on which the vehicle 1 is traveling is a horizontal plane. Accordingly, the vehicle center-of-gravity lateral acceleration estimated value Accy_estm is estimated using the vehicle motion model on the premise that the road surface bank angle θbank is “0”. (Y-axis direction) of the motion acceleration value.

  On the other hand, as described above, when the actual bank angle θbank_act of the road surface on which the vehicle 1 is actually traveling is not “0”, the actual acceleration to which the lateral acceleration sensor 15 is sensitive (that is, the actual sensor-sensitive lateral acceleration Accy_sensor_act). Is obtained by superimposing the acceleration component (= g * sin (θbank_act)) of the gravitational acceleration in the direction parallel to the lateral direction of the vehicle 1 on the actual vehicle center-of-gravity lateral acceleration Accy_act (= Vgdot_y_act + Vgx_act * γ). Accordingly, if there is no error in the vehicle center-of-gravity lateral acceleration detection value Accy_sens, the following equation 2-1 is established.

Accy_sens = Accy_sensor_act
= Accy_act + g * sin (θbank_act) ...... Formula 2-1.

Therefore, the deviation (= Accy_sens−) between the vehicle center-of-gravity lateral acceleration detected value Accy_sens (= detected value of sensor-sensitive lateral acceleration Accy_sensor) based on the output of the lateral acceleration sensor 15 and the vehicle center-of-gravity lateral acceleration estimated value Accy_estm based on the vehicle motion model. Accy_estm, hereinafter, the vehicle center-of-gravity lateral acceleration deviation Accy_estm_err) depends on the actual bank angle θbank_act of the road surface, and in a steady state, the acceleration component in the direction parallel to the lateral direction of the vehicle body 1B (= g *) sin (θbank_act)).

  Therefore, in the present embodiment, the bank angle estimation means 28 executes the processing shown in the flowchart of FIG. 9 to obtain the road surface bank angle estimated value θbank_estm.

  In the following, the bank angle estimating means 28 calculates the vehicle center-of-gravity lateral acceleration deviation Accy_estm_err according to the following equation 2-2 in S118-1. That is, Accy_estm_err is calculated by subtracting the vehicle center-of-gravity lateral acceleration estimated value Accy_estm calculated by the vehicle motion estimation unit 24e in S114 from the vehicle center-of-gravity lateral acceleration detected value Accy_sens generated by the lateral acceleration detecting unit 22f in S100 of FIG. Is done.

Accy_estm_err = Accy_sens-Accy_estm ...... Formula 2-2

Next, the bank angle estimation means 28 calculates a road surface bank angle provisional estimated value θbank_pre as a provisional value of the road surface bank angle estimation value in S118-2. In this case, the bank angle estimating means 28 calculates θbank_pre from the vehicle center-of-gravity lateral acceleration deviation Accy_estm_err obtained in S118-1 by the following equation 2-3.

θbank_pre = sin ⁻¹ (Accy_estm_err / g) …… Formula 2-3

Next, in S118-3, the bank angle estimator 28 obtains the road surface bank angle estimated value θbank_estm by passing the road surface bank angle provisional estimated value θbank_pre calculated as described above through a low-pass characteristic (high cut characteristic) filter.

  The above is the details of the processing of S118 (processing of the bank angle estimating means 28).

  Next, the control apparatus 20 performs the process of the gradient angle estimation means 30 in S120.

  Here, as with Accy_estm, the vehicle center-of-gravity longitudinal acceleration estimated value Accx_estm is the value of the center-of-gravity point of the vehicle 1 estimated using the vehicle motion model on the assumption that the road surface gradient angle θslope is “0”. It means the acceleration value of motion in the front-rear direction (X-axis direction of the vehicle body coordinate system).

  On the other hand, as in the case of the road bank angle θbank, when the actual slope angle θslope_act of the road surface on which the vehicle 1 is actually traveling is not “0”, the actual acceleration (that is, the actual acceleration sensor 14 is sensitive). The sensor-sensitive longitudinal acceleration Accx_sensor_act) is the acceleration component (= −g * sin (θslope_act)) of the gravitational acceleration in the direction parallel to the longitudinal direction of the vehicle body 1B to the actual vehicle center-of-gravity longitudinal acceleration Accx_act (= Vgdot_x_act + Vgy_act * γ). Is superimposed. Therefore, if there is no error in the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens, the following equation 3-1 is established.

Accx_sens = Accx_sensor_act
= Accx_act-g * sin (θslope_act) ...... Formula 3-1.

Therefore, the deviation (= Accx_sens−) between the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens (= detection value of sensor-sensitive longitudinal acceleration Accx_sennsor) based on the output of the longitudinal acceleration sensor 14 and the vehicle center-of-gravity longitudinal acceleration estimation value Accx_estm based on the vehicle motion model. Accx_estm, hereinafter, the vehicle center-of-gravity longitudinal acceleration deviation Accx_estm_err) depends on the actual slope angle θslope_act of the road surface, and in a steady state, the acceleration component in the direction parallel to the longitudinal direction of the vehicle body 1B (= −g * Sin (θslope_act)).

  Therefore, the process of the slope angle estimating means 30 calculates the road surface slope angle estimated value θslope_estm by the same process as the bank angle estimating means 28.

  Specifically, the gradient angle estimating means 30 obtains the road surface gradient angle estimated value θslope_estm by executing the processing shown in the flowchart of FIG.

  Explaining below, the gradient angle estimating means 30 calculates the vehicle center-of-gravity longitudinal acceleration deviation Accx_estm_err in S120-1 by the following equation 3-2. That is, Accx_estm_err is calculated by subtracting the vehicle center-of-gravity longitudinal acceleration estimated value Accx_estm calculated by the vehicle motion estimation unit 24e in S114 from the vehicle center-of-gravity longitudinal acceleration detected value Accx_sens generated by the longitudinal acceleration detection means 22e in S100 of FIG. Is done.

Accx_estm_err = Accx_sens−Accx_estm ...... Equation 3-2

Next, in S120-2, the gradient angle estimation means 30 calculates a road surface gradient angle provisional estimated value θslope_pre as a provisional value of the road surface gradient angle estimated value. In this case, the gradient angle estimation means 30 calculates θslope_pre by the following equation 3-3 from the vehicle center-of-gravity longitudinal acceleration deviation Accx_estm_err obtained in S120-1.

θslope_pre = −sin ⁻¹ (Accx_estm_err / g) …… Equation 3-3

Next, in S120-3, the slope angle estimating means 30 obtains the road surface slope angle estimated value θslope_estm by passing the road surface bank angle provisional estimated value θslope_pre calculated as described above through a low-pass characteristic (high cut characteristic) filter.

  The above is the details of the processing of S120 (processing of the gradient angle estimation means 30).

  Next, the control device 30 executes the process of the μ estimation means 26 in S122.

  Before describing the details of this process, first, the estimation principle of the road surface friction coefficient μ in the present embodiment will be described.

  In this case, for convenience of explanation, it is assumed that the actual dynamics of the vehicle 1 is approximately expressed by the following equation 4-1.

  Specifically, the equation 4-1 is the power of a model vehicle that includes the actual side-sliding motion of the vehicle 1 and the rotational motion about the yaw axis, one front wheel as a steering wheel and one rear wheel as a non-steering wheel. It represents a dynamic model called a so-called two-wheel model (linear two-wheel model), which is approximated as a geometrical behavior. The cornering power CPf of the front wheels in this two-wheel model corresponds to the cornering power per wheel of the front wheels 2-1 and 2-2 of the actual vehicle 1 (four-wheel vehicle), and the cornering power CPr of the rear wheels is This corresponds to the cornering power per wheel of the rear wheels 2-3 and 2-4 of the actual vehicle 1 (four-wheel vehicle).

  Here, the cornering power CPf per wheel of the front wheels 2-1 and 2-2 on the reference road surface where the value of the actual road surface friction coefficient μ_act is “1” is CPf0, and the rear wheels 2-3, 2- on the reference road surface. The cornering power CPr of 4 per wheel is set to CPr0. Then, between each of the cornering powers CPf and CPr on the road surface having an actual road surface friction coefficient μ_act of an arbitrary value and the actual road surface friction coefficient μ_act, an approximation is given as in the following equations 4-2a and 4-2b. Proportionally holds.

CPf = CPf0 * μ_act ...... Formula 4-2a
CPr = Cpr0 * μ_act ...... Formula 4-2b

By applying the equations 4-2a and 4-2b to the equation 4-1, the equation 4-1 can be rewritten as the following equation 4-3.

  Hereinafter, on the basis of the equation 4-3 (an equation expressing the linear two-wheel model), a moment around the yaw axis generated at the NSP (neutral steer point) of the vehicle 1 (that is, the NSP yamment Mnsp) is used. A method for estimating the road friction coefficient μ will be described below.

  First, the technical significance of the actual NSP yaw moment Mnsp_act related to the estimation of the road surface friction coefficient μ and the value of the actual NSP yaw moment Mnsp_act are specified from the observed state values of the motion of the vehicle 1 related thereto (estimation) The method will be described.

  The left side of the expression in the first row of Expression 4-3 means the differential value (temporal change rate) of the actual vehicle gravity center skid speed Vgy_act, that is, the actual vehicle gravity center skid speed change rate Vgdot_y_act. −3 in the first row is rewritten to the following expression 4-4.

Vgdot_y_act + Vgx_act * γ_act + g * sin (θbank_act)
= Μ_act * a11 * Vgy_act / Vgx_act
+ Μ_act * a12s * γ_act / Vgx_act + μ_act * b1 * δf_act ...... Equation 4-4

On the other hand, the following expression 4-5 is obtained from the definition (Accy = Vgdot_y + Vgx * γ) of the vehicle center-of-gravity lateral acceleration Accy and the above expression 2-1 relating to the sensor-sensitive lateral acceleration Accy_sensor.

Accy_sensor_act = Vgdot_y_act + Vgx_act * γ_act + g * sin (θbank_act)
... Formula 4-5

From Equation 4-5, it can be seen that the left side of Equation 4-4 matches the actual sensor-sensitive lateral acceleration Accy_sensor_act. Therefore, the following expression 4-6 is obtained from the expressions 4-4 and 4-5.

Accy_sensor_act = μ_act * a11 * Vgy_act / Vgx_act
+ Μ_act * a12s * γ_act / Vgx_act + μ_act * b1 * δf_act
...... Formula 4-6

The right side of the expression 4-6 represents the lateral direction of the vehicle body 1B in the translational force vector acting on the center of gravity of the vehicle 1 by the resultant force of the actual road surface reaction force acting on each wheel 2-i from the road surface. This corresponds to a value obtained by dividing the component (that is, the X-axis direction component Fgy_total_act of the actual total road surface reaction force composite translational force vector ↑ Fg_total_act) by the vehicle mass m. Therefore, Expression 4-6 represents the relationship that Accy_sensor_act (= Accy_act + g * sin (θbank_act)) matches Fgy_total_act / m.

  Further, the left side of the expression in the second row of Expression 4-3 means the differential value (time change rate) of the actual yaw rate γ_act, that is, the actual yaw angular acceleration γdot_act. The two-line formula is rewritten as the following formula 4-7.

γdot_act = μ_act * a21 * Vgy_act / Vgx_act
+ Μ_act * a22 * γ_act / Vgx_act + μ_act * b2 * δf_act
... Formula 4-7

The right side of the equation 4-7 represents the moment around the yaw axis acting on the center of gravity of the vehicle 1 (ie, the actual total road surface reaction) by the resultant force of the actual road reaction force acting on each wheel 2-i from the road surface. This is equivalent to a value obtained by dividing the force composite yaw moment (Mgz_act) by the vehicle yaw inertia moment Iz. Therefore, Expression 4-7 represents a relationship that γdot_act matches Mgz_act / Iz.

  When the above equations 4-6 and 4-7 are used as simultaneous equations and Vy_act is eliminated, the following equation 4-8 is obtained.

γdot_act− (a21 / a11) * Accy_sensor_act
= Μ_act * ((a22− (a21 / a11) * a12s) * γ_act / Vx_act
+ (B2- (a21 / a11) * b1) * δf_act) ...... Formula 4-8

Here, as described above, when the vehicle 1 is traveling with δ1 = δ2 = 0 and the vehicle center-of-gravity skid angle βg is generated, the NSP generates all wheels 2-i (i = 1, 2). , 3, 4) means the point of application (action point) of the resultant force of the lateral force Fsuby_i (i = 1, 2, 3, 4). Therefore, in the dynamic model of the vehicle 1 represented by the equation 4-3, the vehicle center-of-gravity / NSP distance Lnsp, which is the distance between the center of gravity of the vehicle 1 and the NSP, and the cornering power CPf0, The relationship of the following expression 4-9 is established with CPr0.

Lnsp =-(Lf * CPf0-Lr * CPr0) / (CPf0 + CPr0) ...... Formula 4-9

Furthermore, the following expression 4-10 is obtained from the expression 4-9 and the definitions of a11 and a21 shown in the proviso of the expression 4-2.

a21 / a11 = -Lnsp * m / Iz Equation 4-10

Then, by applying the expression 4-10 to the left side of the expression 4-8, the expression 4-8 can be rewritten as the following expression 4-11.

Iz * γdot_act + Lnsp * m * Accy_sensor_act = μ_act * p (γ_act, δf_act, Vx_act)
... Formula 4-11
However,
p (γ_act, δf_act, Vx_act)
= Iz * ((a22− (a21 / a11) * a12s) * γ_act / Vx_act
+ (B2− (a21 / a11) * b1) * δf_act) …… Equation 4-12

Both sides of Expression 4-11 mean the actual moment around the yaw axis at NSP (actual NSP yaw moment Mnsp_act). That is, the actual NSP yaw moment Mnsp_act coincides with the left side and the right side of Equation 4-11 as shown in the following Equations 4-13a and 4-13b.

Mnsp_act = Iz * γdot_act + Lnsp * m * Accy_sensor_act ...... Formula 4-13a
Mnsp_act = μ_act * p (γ_act, δf_act, Vx_act) ...... Formula 4-13b

Equation 4-13a is an external force moment (a moment obtained by inverting the sign of the actual inertial force moment) that balances the actual inertial force moment (moment component of the actual inertial force) around the yaw axis generated by the NSP by the movement of the vehicle 1. ) Represents the actual NSP yaw moment Mnsp_act. The first term on the right side of Equation 4-13a is an external force moment that balances the actual inertial moment around the yaw axis generated at the center of gravity of the vehicle 1 by the movement of the vehicle 1 (that is, the actual total road surface reaction force combined yaw moment). Mgz_total_act). The second term on the right side of Equation 4-13b is the actual translational inertial force in the Y-axis direction of the vehicle body coordinate system generated at the center of gravity of the vehicle 1 due to the movement of the vehicle 1 (the translational force component of the actual inertial force). ) (Ie, the actual total road surface reaction force combined translational force vector ↑ Fg_total_act, Y-axis direction component Fgy_total_act of the vehicle body coordinate system) corresponds to the moment (= Lnsp * Fgy_total_act) generated around the yaw axis in NSP .

  Further, the expression 4-13b is obtained as an actual NSP around the yaw axis acting on the NSP by the resultant force of the actual road surface reaction force acting on each wheel 2-i from the road surface depending on the actual road surface friction coefficient μ_act. This is a representation of the yaw moment Mnsp_act. Note that p (γ_act, δf_act, Vgx_act) defined by Expression 4-12 is the ratio of the increase amount of Mnsp_act to the increase amount of μ_act (differential value of Mnsp_act by μact), as is apparent from Expression 4-13b. In other words, it has a meaning as sensitivity of Mnsp_act (hereinafter referred to as μ sensitivity) with respect to a change of μ_act. In other words, p (γ_act, δf_act, Vgx_act) is the actual NSP yaw moment Mnsp_act when μ_act = 1 (when the actual road surface friction coefficient μ_act matches the friction coefficient of the reference road surface). .

  Here, neither the right side of the equations 4-13a and 4-13b includes the actual vehicle gravity center side slip velocity Vgy_act or the actual road surface bank angle θbank_act. Therefore, it can be seen that the actual NSP yaw moment Mnsp_act is defined without depending directly on the actual vehicle center-of-gravity skid speed Vgy_act or the actual road surface bank angle θbank_act. More specifically, when a change in the actual vehicle center-of-gravity skid speed Vgy_act or a change in the actual road surface bank angle θbank_act occurs, the moment component of the first term on the right side of the equation 4-13a and the Although the moment component of the two terms changes, the change of these moment components basically occurs in opposite directions. For this reason, changes in the moment components of the first term and the second term of Expression 4-13a due to changes in Vgy_act or changes in θbank_act occur so as to cancel each other. As a result, the actual NSP yaw moment Mnsp_act is less likely to be affected by changes in Vgy_act or changes in θbank_act.

  As is clear from Equation 4-13b, the actual NSP yaw moment Mnsp_act does not directly depend on the values of Vgy_act and θbank_act in a situation where μ sensitivity p (γ_act, δf_act, Vx_act) is p ≠ 0. It can be seen that the actual road surface friction coefficient μ_act and μ sensitivity p change depending on the actual road surface friction coefficient μ_act and μ sensitivity p.

  If attention is paid to the expression 4-13a of the expressions 4-13a and 4-13b, the actual yaw angular acceleration γdot_act and the actual sensor-sensitive lateral acceleration Accy_sensor_act as the motion state quantity of the vehicle 1 are observed. Based on these observation values, the actual NSP yaw moment Mnsp_act value generated by the resultant force of the actual road surface reaction force acting on each wheel 2-i from the road surface (which depends on the actual road surface friction coefficient μ_act) is specified. I understand that I can do it. In this case, the right side of Expression 4-13a does not include the actual road surface friction coefficient μ_act, and does not include the actual vehicle center-of-gravity lateral acceleration Vgy_act and the actual road surface bank angle θbank_act. For this reason, the actual yaw angular acceleration γdot_act and the actual sensor-sensitive lateral acceleration Accy_sensor_act can be used without the actual road surface friction coefficient μ_act, actual vehicle center-of-gravity lateral acceleration Vgy_act, and actual road surface bank angle θbank_act. The observed value of the NSP yaw moment Mnsp_act can be obtained.

  Here, the yaw angular acceleration detected value γdot_sens has the meaning as the observed value of the actual yaw angular acceleration γdot_act, and the vehicle center-of-gravity lateral acceleration detected value Accy_sens has the meaning as the observed value of the actual sensor-sensitive lateral acceleration Accy_sensor_act. Therefore, a value calculated by an expression obtained by replacing γdot_act and Accy_sennsor_act on the right side of Expression 4-13a with γdot_sens and Accy_sens as the observed values is hereinafter referred to as an NSP yaw moment detected value Mnsp_sens. That is, Mnsp_sens is defined by the following equation 4-14.

Mnsp_sens = Iz * γdot_sens + Lnsp * m * Accy_sens ...... Formula 4-14

In this case, assuming that the yaw angular acceleration detection value γdot_sens and the vehicle center-of-gravity lateral acceleration detection value Accy_sens match the actual yaw angular acceleration detection value γdot_act and the actual sensor sensitive lateral acceleration Accy_sensor_act with high accuracy, Mnsp_act = Mnsp_sens. Therefore, the NSP yaw moment detected value Mnsp_sens as the observed value of the actual NSP yaw moment Mnsp_act can be calculated from the yaw angular acceleration detected value γdot_sens and the vehicle center-of-gravity lateral acceleration detected value Accy_sens by Equation 4-14. The NSP yaw moment detection value Mnsp_sens calculated in this way does not require an actual external force (actual road surface reaction force) value or an actual road surface friction coefficient μ_act acting on the vehicle 1, and the motion of the vehicle 1. It has a meaning as a value (detected value) of Mnsp estimated based on the observed value of the state quantity.

  Next, apart from the NSP yaw moment detected value Mnsp_sens, an estimated value of the road surface friction coefficient μ is used for the road surface reaction force acting on the wheels of the vehicle 1 on the model, based on an appropriate dynamic model of the vehicle 1. A process for estimating the value of the NSP yaw moment Mnsp generated by the resultant resultant road reaction force will be described.

  Here, in the present embodiment, the road surface reaction force estimated value is actually calculated by the vehicle model calculation means 24 using the friction characteristic model and the vehicle motion model as described above. Then, from the estimated value of the road surface reaction force, the value of the NSP yaw moment Mnsp can be estimated as will be described later. However, in this description, for the purpose of explaining the estimation principle of the road surface friction coefficient μ, for convenience, vehicle model calculation means (hereinafter referred to as explanation vehicle model calculation means) different from the vehicle model calculation means 24 is used. It is assumed that, using the dynamic model of the vehicle 1 represented by Formula 4-3, the estimation calculation process of the motion state amount of the vehicle 1 and the road surface reaction force is sequentially executed at a predetermined calculation process cycle.

  In this case, the explanatory vehicle model calculation means includes a front wheel steering angle detection value δf_sens, a yaw rate detection value γ_sens, a vehicle center of gravity longitudinal speed estimation value Vgx_estm as a vehicle speed observation value, a road surface friction coefficient estimation value μ_estm in each calculation processing cycle. , And the latest value (previous value or current value) of the estimated road bank angle θbank_estm are input as observed values of δf_act, γ_act, Vgx_act, μ_act, and θbank_act on the right side of Equation 4-3, respectively. Here, Vgx_estm, μ_estm, and θbank_estm mean observed values obtained by any appropriate method. In addition, the values of the parameters a11, a12s, a21, a22, b1, and b2 in the equation 4-3 are set in advance.

  And the vehicle model calculation means for description shall perform the estimation calculation process shown below. That is, the explanatory vehicle model calculation means calculates the vehicle center-of-gravity skid speed according to the following equation 5-1, in which an actual value such as γ_act in the first row of the equation 4-3 is replaced with an estimated value or a detected value. A vehicle center-of-gravity skid speed change rate estimated value Vgdot_y_estm, which is an estimated value of the temporal change rate (differential value) of Vgy, is calculated.

Vgdot_y_estm = μ_estm * a11 * Vgy_estm_p / Vgx_estm
+ Μ_estm * a12s * γ_sens / Vgx_estm + μ_estm * b1 * δf_sens
−Vgx_estm * γ_sens−g * sin (θbank_estm)
... Formula 5-1

The vehicle center-of-gravity skid speed estimated value Vgy_estm_p required for the calculation of the first term on the right side of Equation 5-1 is the previous value as the latest value of Vgy_estm that has already been calculated by the vehicle model calculation means for explanation.

  In this case, the fourth term and the fifth term are removed from the right side of Expression 5-1, and the vehicle body of the translational force vector acting on the center of gravity of the vehicle 1 by the resultant force of the road surface reaction force of each wheel 2-i. It has a meaning as a value obtained by dividing the estimated value of the horizontal component of 1B (that is, the estimated value Fgy_total_estm of the Y-axis direction component of the total road surface reaction force combined translational force vector ↑ Fg_total) by the vehicle mass m. The fourth term on the right side is an estimated value of acceleration generated at the center of gravity of the vehicle 1 due to the centrifugal force accompanying the turning motion of the vehicle 1, and the fifth term is the lateral acceleration of the vehicle body 1B in the gravitational acceleration. It means the estimated value of the acceleration component in the direction.

  Therefore, Formula 5-1 calculates Fgy_total_estm / m based on μestm, Vgy_estm_p, Vgx_estm, γ_sens, and δf_sens, and the acceleration of centrifugal force acting on the center of gravity of the vehicle 1 from the calculated value of Fgy_total_estm / m The vehicle center-of-gravity skid speed change rate is subtracted from the estimated value (= Vgx_estm * γ_sens) of the vehicle and the estimated value of the acceleration component in the lateral direction of the vehicle body 1B (= g * sin (θbank_estm)). The process which calculates estimated value Vgdot_y_estm is shown.

  The explanatory vehicle model calculation means then calculates the integral calculation of Vgdot_y_estm from the vehicle center-of-gravity skid speed change rate estimated value Vgdot_y_estm and the previous value Vgy_estm_p of the vehicle center-of-gravity skid speed estimated value obtained as described above. Thus, a new vehicle center-of-gravity skid speed estimated value Vgy_estm (current value) is calculated. Note that ΔT in Expression 5-2 is a calculation processing cycle of the explanatory vehicle model calculation means.

Vgy_estm = Vgy_estm_p + Vgdot_y_estm * ΔT ...... Formula 5-2

Vgy_estm calculated in this way is used to calculate a new vehicle center-of-gravity skid speed change rate Vgdot_y_estm in the next calculation processing cycle.

  Further, the explanatory vehicle model calculation means calculates a sensor-sensitive lateral acceleration estimated value Accy_sensor_estm, which is an estimated value of the actual acceleration to which the lateral acceleration sensor 15 of the vehicle 1 is sensitive (that is, the actual sensor-sensitive lateral acceleration Accy_sensor_act), using the following equation 5-3: (In other words, by calculation from the first term to the third term on the right side of Equation 5-1).

Accy_sensor_estm = μ_estm * a11 * Vgy_estm_p / Vgx_estm
+ Μ_estm * a12s * γ_sens / Vgx_estm + μ_estm * b1 * δf_sens
... Formula 5-3

Here, supplementing this equation 5-3, as is clear from the equation 4-5, the following equation 5-4 is established.

Accy_sensor_estm = Vgdot_y_estm + Vgx_estm * γ_sens + g * sin (θbank_estm)
... Formula 5-4

As is clear from Expression 5-4 and Expression 5-1, the right side of Expression 5-4 matches the sum of the first to third terms on the right side of Expression 5-1. Therefore, the sensor-sensitive lateral acceleration estimated value Accy_sensor_estm can be calculated by the equation 5-3.

  Further, the right side of the equation 5-3 is an estimated value of the lateral component of the vehicle body 1B in the translational force vector acting on the center of gravity of the vehicle 1 by the resultant force of the road surface reaction force of each wheel 2-i (that is, all It has a meaning as a value obtained by dividing the estimated value Fgy_total_estm of the Y-axis direction component of the road surface reaction force combined translational force vector ↑ Fg_total by the vehicle mass m. Therefore, Expression 5-3 shows processing for calculating Fgy_total_estm / m based on μestm, Vgy_estm_p, Vgx_estm, γ_sens, and δf_sens, and obtaining the calculated Fgy_total_estm / m as Accy_sensor_estm.

  Further, the explanatory vehicle model calculation means calculates the yaw angular acceleration γdot by the following equation 5-5 in which an actual value such as γ_act in the equation in the second row of the equation 4-3 is replaced with an estimated value or a detected value. The yaw angular acceleration estimated value γdot_estm, which is the estimated value of the temporal change rate (differential value) of, is calculated.

γdot_estm = μ_estm * a21 * Vgy_estm_p / Vgx_estm
+ Μ_estm * a22 * γ_sens / Vgx_estm + μ_estm * b2 * δf_sens
... Formula 5-5

The right side of the equation 5-5 represents an estimated value of the moment about the yaw axis acting on the center of gravity of the vehicle 1 by the resultant force of the road surface reaction force of each wheel 2-i (that is, the total road surface reaction force combined yaw moment estimated value Mgz_estm ) Is divided by the vehicle yaw moment of inertia Iz. Therefore, Expression 5-5 indicates a process of calculating Mgz_estm / Iz based on μestm, Vgy_estm_p, Vgx_estm, γ_sens, and δf_sens, and obtaining the calculated Mgz_estm / Iz value as the yaw angular acceleration estimated value γdot_estm.

  Here, Equations 5-3 and 5-5 are used as simultaneous equations, Vgy_estm is eliminated, and Equation 4-10 is applied to obtain Equation 5-6 below.

Iz * γdot_estm + Lnsp * m * Accy_sensor_estm
= Μ_estm * p (γ_sens, δf_sens, Vgx_estm) ...... Formula 5-6
However,
p (γ_sens, δf_sens, Vgx_estm)
= Iz * ((a22− (a21 / a11) * a12s) * γ_sens / Vgx_estm
+ (B2− (a21 / a11) * b1) * δf_sens) ...... Formula 5-7

Note that p (γ_sens, δf_sens, Vgx_estm) defined by Equation 5-7 means the value of μ sensitivity calculated from the observed values γ_sens, δf_sens, Vgx_estm of γ, δf, and Vgx. In the following description, the μ sensitivity p means p (γ_sens, δf_sens, Vgx_estm) defined by the equation 5-7 unless otherwise specified. More generally, the μ sensitivity p defined by Expression 5-7 is a value of μ sensitivity calculated by linear combination of γ_sens and δf_sens. In this case, in this linear combination, if the coefficients applied to γ_sens and δf_sens are set to A1 and A2 (p = A1 * γ_sens + A2 * δf_sens), A1 = Iz * ((a22− (a21 / a11 ) * A12s) / Vgx_estm, A2 = (b2− (a21 / a11) * b1) Therefore, the coefficients A1 and A2 have their ratio A2 / A1 as Vgx_estm as an observed value of the vehicle speed. In other words, it is a linear combination of γ_sens and δf_sens according to Equation 5-7: A2 / A1 is a coefficient set so that it changes in proportion to Vgx_estm. Are the observed values (detected values) γ_sens, δf_sens, Vgx_estm as the values of γ_act, δf_act, Vgx_act when the road surface friction coefficient μ_act is assumed to be a constant value in the linear two-wheel vehicle model of Equation 4-3. The value of μ sensitivity p calculated by the linear combination is compared with the value of the actual NSP yaw moment Mnsp_act specified using It said to be a linear combination constructed so as to be.

  Supplementally, in the present embodiment, as described above, the yaw rate estimated value γ_estm is determined so as to match or substantially match the yaw rate detected value γ_sens. Therefore, an expression obtained by replacing γ_sens on the right side of Expression 5-7 with γ_estm may be used as a defining expression for obtaining the value of μ sensitivity p.

  Both sides of the above equation 5-6 represent an NSP yaw moment estimated value Mnsp_estm that is an estimated value of the moment around the yaw axis at the NSP (moment value on the model based on the equation 4-3). means. That is, the NSP yaw moment estimated value Mnsp_estm coincides with the left side and the right side of Equation 5-6 as shown in Equations 5-8a and 5-8b below.

Mnsp_estm = Iz * γdot_estm + Lnsp * m * Accy_sensor_estm ...... Formula 5-8a
Mnsp_estm = μ_estm * p (γ_sens, δf_sens, Vgx_estm) ...... Formula 5-8b

Expression 5-8a is an NSP yaw as an estimated value of a moment (moment obtained by reversing the sign of the inertial force moment) that balances the inertial force moment generated around the yaw axis at the NSP by the movement of the vehicle 1 on the model. This is a representation of the moment estimated value Mnsp_estm. Equation 5-8b is an NSP as an estimate of the moment around the yaw axis generated by the NSP by the resultant force of the road surface reaction force of each wheel 2-i depending on μ_estm (the resultant force of the road surface reaction force on the model). The yaw moment estimated value Mnsp_estm is expressed.

  In this case, the NSP yaw moment estimated value Mnsp_estm calculated by the expression 5-8b of the expressions 5-8a and 5-8b is calculated depending on the road surface friction coefficient estimated value μ_estm. Mnsp_estm reflects the influence of the error of the road surface friction coefficient estimated value μestm. Further, the right side of Expression 5-8b does not directly include the vehicle center-of-gravity lateral acceleration estimated value Vgy_estm or the road bank angle estimated value θbank_estm. For this reason, the NSP yaw moment estimated value Mnsp_estm calculated by Equation 5-8b is affected by the error of the vehicle center-of-gravity lateral acceleration estimated value Vgy_estm and the road bank angle estimated value θbank_estm, as described for the actual NSP yaw moment Mnsp_act. Will not be directly received.

  Therefore, the vehicle model calculation means for explanation calculates the NSP yaw moment estimated value Mnsp_estm by Expression 5-8b. The NSP yaw moment estimated value Mnsp_estm calculated in this way can be generalized. Based on the dynamic model of the vehicle 1, the estimated value of Mnsp_act calculated depending on μ_estm (more specifically, (Mnsp_act estimated value calculated assuming that μ_estm is accurate).

  The above is the processing of the explanatory vehicle model calculation means. Supplementally, in the vehicle model calculation means for explanation described above, for convenience of explanation of the estimation principle of the road surface friction coefficient μ, the vehicle center-of-gravity skid speed change rate estimated value Vgdot_y_estm, the vehicle center-of-gravity skid speed estimated value Vgy_estm, the yaw angular acceleration estimated value γdot_estm The sensor-sensitive lateral acceleration estimated value Accy_sensor_estm is calculated. On the premise of the dynamic model (linear two-wheel vehicle model) represented by the equation 4-3, the NSP yaw moment estimated value by the equation 5-8b. In calculating Mnsp_estm, Vgdot_y_estm, Vgy_estm, γdot_estm, and Accy_sensor_estm are not necessary, as is apparent from 5-7 and 5-8b. When the dynamic model represented by the equation 4-3 is assumed, the calculation result value on the right side of the equation 5-8a and the calculation result on the right side of the equation 5-8b are the same value. Therefore, Mnsp_estm may be calculated by Expression 5-8a.

  Next, a method for estimating the road surface friction coefficient μ based on the NSP yaw moment detected value Mnsp_sens obtained by the equation 4-14 and the NSP yaw moment estimated value Mnsp_estm calculated by the equation 5-8b will be considered. . As described above, Mnsp_sens is an observed value (γdot_sens, Accy_sens) of the motion state quantity of the vehicle 1 without requiring the value of the road surface reaction force acting on the vehicle 1 as an external force or the value of the road surface friction coefficient μ. It has a meaning as an observed value (detected value) of Mnsp_act obtained based on. Mnsp_estm has a meaning as an observed value (estimated value) of Mnsp_act calculated using μ_estm based on the dynamic model of the vehicle 1. Therefore, the deviation between Mnsp_sens and Mnsp_estm is considered to have a correlation with the error of μ_estm with respect to μ_act.

  Here, the yaw rate detection value γ_sens, the yaw angular acceleration detection value γdor_sens, the front wheel rudder angle detection value δf_sens, the vehicle center-of-gravity longitudinal acceleration estimation value Vgx_estm (vehicle speed estimation value), and the vehicle center-of-gravity lateral acceleration detection value Accy_sens are respectively the actual yaw rate γ_act, It is assumed that the actual yaw angular acceleration γdot_act, the actual front wheel steering angle detection value δf_act, the actual vehicle center-of-gravity longitudinal velocity Vgx_act, and the actual sensor-sensitive lateral acceleration Acc_sensor_act are accurately matched. At this time, the following equation 6-1 is obtained from the equation 4-11.

Iz * γdot_sens + Lnsp * m * Accy_sens = μ_act * p (γ_sens, δf_sens, Vgx_estm)
... Formula 6-1

Then, the following expression 6-2 is obtained from the expression 6-1 and the expressions 4-14, 5-6, and 5-8b.

Mnsp_sens−Mnsp_estm = (Iz * γdot_sens + Lnsp * m * Accy_sens)
-(Iz * γdot_estm + Lnsp * m * Accy_sensor_estm)
= (Μ_act−μ_estm) * p (γ_sens, δf_sens, Vgx_estm)
... Formula 6-2

From Equation 6-2, in order to make the road surface friction coefficient estimated value μ_estm coincide with the actual road surface friction coefficient μ_act, μ_estm is set so that Mnsp_estm coincides with Mnsp_sens in a situation where p (γ_sens, δf_sens, Vgx_estm) ≠ 0. It will be sufficient to determine. More generally, this is an actual NSP calculated using a dynamic model (dynamic model depending on the road surface friction coefficient estimated value μ_estm) including the friction characteristics of each wheel 2-i of the vehicle 1. The estimated value of yaw moment Mnsp_act (NSP yaw moment estimated value Mnsp_estm) is used to detect the detected yaw angular acceleration γdot_sens and the vehicle center-of-gravity lateral acceleration detected value Accy_sens (= sensor-sensitive lateral acceleration Accy_sensor). This means that the road surface friction coefficient estimated value μestm to be applied to the dynamic model should be determined so as to match the estimated value (NSP yaw moment detected value Mnsp_sens) of the actual NSP yaw moment Mnsp_act calculated from To do.

  In this case, p (γ_sens, δf_sens, Vgx_estm) on the right side of Expression 6-2 does not include the vehicle center-of-gravity skid speed estimated value Vgy_estm and the road bank angle estimated value θbank_estm, as is apparent from Expression 5-7. Therefore, in the situation where p (γ_sens, δf_sens, Vgx_estm) ≠ 0, the value of the deviation between Mnsp_sens and Mnsp_estm (the left side of Equation 6-2) is the correlation with the deviation between μ_act and μ_estm (ie, the error of μ_estm). Is considered high. In other words, in a situation where p (γ_sens, δf_sens, Vgx_estm) ≠ 0, the deviation between Mnsp_sens and Mnsp_estm is considered to be mainly caused by the error of μ_estm. Therefore, if the road surface friction coefficient estimated value μ_estm is determined based on the equation 6-2, it is possible to suppress the influence of errors of the vehicle center-of-gravity skid speed estimated value Vgy_estm and the road surface bank angle estimated value θbank_estm. Thus, it is considered that the actual road surface friction coefficient μ_act can be estimated. Therefore, the μ estimation means 26 in the present embodiment calculates the road surface friction coefficient estimated value μ_estm based on the equation 6-2.

  In order to determine the road surface friction coefficient estimated value μ_estm so that Mnsp_estm coincides with Mnsp_sens in the situation where p (γ_sens, δf_sens, Vgx_estm) ≠ 0 based on Equation 6-2 as described above, for example, It is conceivable to determine the road surface friction coefficient estimated value μ_estm so as to satisfy Expression 6-3.

Mnsp_sens = μ_estm * p (γ_sens, δf_sens, Vgx_estm) ...... Formula 6-3

However, in this case, excessive fluctuation of the road surface friction coefficient estimated value μ_estm is likely to occur due to errors in Mnsp_sens, γ_sens, Δf_sens, and Vx_sens. In particular, when the value of p (γ_sens, δf_sens, Vgx_estm) is close to “0”, it is difficult to ensure the reliability and stability of the road surface friction coefficient estimated value μ_estm obtained based on Expression 6-3. It is.

  Therefore, the μ estimation means 26 in the present embodiment is based on the NSP yaw moment detected value Mnsp_sens obtained from the observed value of the motion state quantity of the vehicle 1 and the road surface reaction force estimated depending on the road surface friction coefficient estimated value μ_estm. In order to converge the deviation to “0” (so that Mnsp_estm converges to Mnsp_sens) by feedback calculation processing according to the deviation from the NSP yaw moment estimated value Mnsp_estm required for Then, the value of μestm is updated according to the increase / decrease operation amount. Thus, μestm is sequentially calculated so that the road surface friction coefficient estimated value μestm converges to the actual road surface friction coefficient μ_act (constantly coincides with μ_act). Hereinafter, the deviation (= Mnsp_sens−Mnsp_estm) between Mnsp_sens and Mnsp_estm is referred to as NSP yaw moment estimation error Mnsp_err.

  In this case, as is clear from the equation 6-2, the NSP yaw moment estimation error Mnsp_err is proportional to the μ sensitivity p. As the μ sensitivity p is closer to “0”, the sensitivity of Mnsp_err to the error of μ_estm (the magnitude of the ratio of change in Mnsp_err to the change in error of μ_estm) decreases. Therefore, in this embodiment, in order to ensure the reliability and stability of μ_estm, the gain value that is the ratio of the change in the amount of increase / decrease in μ_estm with respect to the change in Mnsp_err (in other words, Mnsp_err is converged to “0”). The feedback gain of the feedback calculation process) is changed according to the μ sensitivity p.

  The above is the basic estimation principle of the road surface friction coefficient μ in the present embodiment.

  Based on the basic estimation principle of the road surface friction coefficient μ described above, the processing of the μ estimation means 26 in the present embodiment will be described with reference to FIGS. 11 and 12.

  As shown in the block diagram of FIG. 11, the μ estimator 26 functions as the Mnsp_sens calculator 26a that calculates the NSP yaw moment detected value Mnsp_sens, the Mnsp_estm calculator 26b that calculates the NSP yaw moment estimated value Mnsp_estm, Mnsp_err calculation unit 26c for calculating NSP yaw moment estimation error Mnsp_err, μ sensitivity calculation unit 26d for calculating μ sensitivity p (γ_sens, δf_sens, Vgx_estm), NSP yaw moment estimation error Mnsp_err and μ sensitivity p (γ_sens, δf_sens, Vgx_estm), a friction coefficient increase / decrease operation amount determination unit 26e that determines an increase / decrease operation amount Δμ of the road surface friction coefficient μ, and a friction coefficient estimate value update that updates the road surface friction coefficient estimate value μ_estm according to the increase / decrease operation amount Δμ. 26f.

  Then, the μ estimation means 26 sequentially determines the road surface friction coefficient estimated value μ_estm by executing the processing of the flowchart of FIG.

  That is, the μ estimation means 26 executes the processing of the Mnsp_sens calculation unit 26a in S122-1, and calculates the NSP yaw moment detected value Mnsp_sens. Specifically, the Mnsp_sens calculation unit 26a includes the motion state quantity of the vehicle 1 related to the inertial force moment that balances the NSP yaw moment Mnsp among the observation target quantity detection values generated by the observation target quantity detection unit 22 in S100. Mnsp_sens is calculated by performing the calculation of the right side of Equation 4-14 using the yaw angular acceleration detection value γdot_sens as the observed value and the vehicle center-of-gravity lateral acceleration detection value Accy_sens (sensor-sensitive lateral acceleration detection value). In this case, predetermined values set in advance are used as the value of the vehicle yaw inertia moment Iz, the value of the vehicle mass m, and the value of the vehicle center-of-gravity / NSP distance Lnsp necessary for the calculation of Expression 4-14. The first term on the right side of Equation 4-14 corresponds to the total road surface reaction force combined yaw moment detected value Mgz_total_sens, and m * Accy_sens in the second term corresponds to the total road surface reaction force combined lateral force detected value Fgy_total_sens. To do.

  Further, in S122-2, the μ estimating means 26 executes the process of the Mnsp_estm calculating unit 26b to calculate the NSP yaw moment estimated value Mnsp_estm. Specifically, the Mnsp_estm calculating unit 26b calculates the total road surface reaction force combined lateral force estimated value Fgy_total_estm (total road surface reaction force combined translational force vector estimated value ↑ Fg_total_estm in the Y-axis direction) calculated by the vehicle model calculating unit 24 in S112. Component) and the total road surface reaction force composite yaw moment estimated value Mgz_total_estm, Mnsp_estm is calculated by the following equation 7-1.

Mnsp_estm = Mgz_total_estm + Lnsp * Fgy_total_estm ...... Formula 7-1

Next, in S122-3, the μ estimating means 26 executes the process of the Mnsp_err calculating unit 26c to calculate the NSP yaw moment estimating error Mnsp_err. Specifically, the Mnsp_err calculating unit 26c calculates Mnsp_err by subtracting the NSP yaw moment estimated value Mnsp_estm calculated in S122-2 from the NSP yaw moment detected value Mnsp_sens calculated in S122-1.

  Furthermore, the μ estimation means 26 executes the processing of the μ sensitivity calculation unit 26d in S122-4 to calculate the μ sensitivity p. Specifically, the μ sensitivity calculation unit 26d is obtained by the yaw rate detection value γ_sens and the front wheel steering angle detection value δf_sens generated by the observation target amount detection unit 22 in S100, and by the vehicle model calculation unit 24 in S114. The μ sensitivity p (γ_sens, δf_sens, Vgx_estm) is calculated from the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm by calculating the right side of the equation 5-7. In this case, predetermined values set in advance are used as the values of the vehicle inertia yaw moment Iz and the parameters a11, a12s, a21, a22, b1, and b2 necessary for the calculation of Expression 5-7.

  In this case, as is clear from Expression 5-7, the μ sensitivity p is obtained by linear combination of γ_sens and δf_sens. In this linear combination, the ratio between the coefficient applied to γ_sens and the coefficient applied to Δf_sens changes according to Vgx_estm.

  Next, in S122-5, the μ estimating means 26 executes the process of the friction coefficient increasing / decreasing manipulated variable determining unit 26e, and the NSP yaw moment estimating error Mnsp_err calculated in S122-3 and the μ calculated in S122-4. The friction coefficient increasing / decreasing manipulated variable Δμ is determined according to the sensitivity p. In this process, the friction coefficient increasing / decreasing manipulated variable Δμ is determined so that Mnsp_err converges to “0” (that is, Mnsp_estm converges to Mnsp_sens) by the feedback control law. In this case, a proportional law is used as the feedback control law, and Δμ is calculated by multiplying a certain gain value Gmu by Mnsp_err. In this case, Δμ is determined to be proportional to the product of Mnsp_err and the μ sensitivity p. As a result, the gain value Gmu (hereinafter, Gmu is referred to as a friction coefficient operation gain) representing the rate of change in Δμ with respect to the change in Mnsp_err is determined so as to change according to the μ sensitivity p.

  Specifically, in the present embodiment, the friction coefficient increasing / decreasing manipulated variable determiner 26e calculates Δμ by the following equation 7-2. Note that Kmu in Expression 7-2 is a predetermined positive value set in advance.

Δμ = Mnsp_err * Gmu
= Mnsp_err * (p (γ_sens, δf_sens, Vgx_estm) * Kmu)
...... Formula 7-2

That is, the friction coefficient increasing / decreasing manipulated variable determiner 26e is obtained by multiplying the μ sensitivity p calculated in S122-4 by the basic positive gain Kmu that is a preset positive value (= Kmu * p). A friction coefficient increasing / decreasing manipulated variable Δμ is determined by multiplying the friction coefficient operating gain Gmu by the NSP yaw moment estimation error Mnsp_err calculated in S122-3. In this case, the coefficient of friction operation gain Gmu has the same polarity as the μ sensitivity p, and the smaller the magnitude (absolute value) of the μ sensitivity p is, the smaller the magnitude (absolute value) of Gmu is (as a result). , So that the magnitude of Δμ decreases as the magnitude (absolute value) of μ sensitivity p decreases.

  Next, in step S122-6, the μ estimation unit 26 executes the process of the friction coefficient estimated value update unit 26f, and updates the road surface friction coefficient estimated value μ_estm. Specifically, the friction coefficient estimated value update unit 26f adds the friction coefficient increasing / decreasing manipulated variable Δμ calculated in S122-5 to the previous value μ_estm_p of the road surface friction coefficient estimated value, thereby obtaining the road surface friction coefficient estimated value μ_estm. Update from the previous value μ_estm_p to obtain a new road surface friction coefficient estimated value μ_estm (current value μ_estm). In other words, this process is a process of obtaining the road surface friction coefficient estimated value μ_estm by integrating Δμ.

  The above is the details of the processing of the μ estimation means 26 in the present embodiment.

  Supplementally, in the present embodiment, the process of the vehicle model calculation unit 24 (the process of S102 to S116 of FIG. 4) and the process of obtaining the NSP yaw moment estimated value Mnsp_estm by the μ estimation unit 26 (S122-2 of FIG. 12). Thus, the comparison target external force first estimation means in the present invention is realized. In this case, Mnsp_estm corresponds to the first estimated value of the comparison target external force in the present invention. Further, the detected values (δ1_sens, δ2_sens, Vw_i_sens, γ_sens, Accx_sens, Accy_sens, Tq_i_sens) input to the vehicle model calculation means 24 correspond to the observed values of the predetermined types of observation target amounts in the present invention. The detected values (δ1_sens, δ2_sens, Vw_i_sens, γ_sens, Accx_sens, Accy_sens, Tq_i_sens) of the observation target amounts are input parameter values other than the road surface friction coefficient μ (κi, βi, Fz_i) is a detection value of the observation target amount necessary for specifying. Of the processes of the vehicle model calculation means 24, the vehicle motion / road reaction force estimation means in the present invention is realized by the processes of S102 to S116. In this case, the vehicle movement skid speed Vgy_estm corresponds to the state quantity of the vehicle skid movement in the present invention. Further, the relationship represented by the above expression 1-14 corresponds to the dynamic relationship related to the vehicle motion / road surface reaction force estimating means. The process of S122-2 in FIG. 12 may be executed by the vehicle model calculation unit 24.

  Further, in the present embodiment, the comparison target external force second estimating means in the present invention is realized by the process (S122-1 in FIG. 12) for obtaining the NSP yaw moment detected value Mnsp_sens by the μ estimating means 26. In this case, Mnsp_sens corresponds to the second estimated value of the comparison target external force in the present invention. Further, a vehicle in which the yaw angular acceleration detection value γdot_sens and the vehicle center-of-gravity lateral acceleration detection value Accy_sens (sensor-sensitive lateral acceleration detection value Accy_sensor_sens) define the inertia moment around the yaw axis at NSP (inertial force corresponding to the comparison target external force). This corresponds to the observed value of the motion state quantity of 1.

  Further, the process of the μ sensitivity calculation unit 26d of the μ estimation unit 26 (the process of S122-4 in FIG. 12) corresponds to the μ sensitivity calculation unit in the present invention, and the process of the friction coefficient increase / decrease manipulated variable determination unit 26e (FIG. 12). The process of S122-5 of FIG. 12 corresponds to the friction coefficient increase / decrease manipulated variable determining means in the present invention, and the process of the friction coefficient estimated value update unit 26f (the process of S122-6 in FIG. 12) is the friction coefficient estimation in the present invention. This corresponds to value updating means.

  The correspondence relationship between the present embodiment and the present invention is the same in the second to ninth embodiments described later.

  In the present embodiment described above, the friction coefficient increasing / decreasing manipulated variable Δμ is determined so that the NSP yaw moment error Mnsp_err, which is a deviation between the NSP yaw moment detected value Mnsp_sens and the NSP yaw moment estimated value Mnsp_estm, converges to “0”. Is done. For this reason, while suppressing the influence of the estimated value of the state quantity of the skid motion of the vehicle 1 such as the estimated vehicle center of gravity side slip velocity Vgy_estm and the change in the actual road bank angle θ_act on the estimated road friction coefficient μ_estm, μ_estm can be determined. For this reason, highly reliable μ_estm can be obtained stably.

  The road surface friction coefficient estimated value μ_estm is determined so as to be proportional to the product of the NSP yaw moment error Mnsp_err and the μ sensitivity p calculated by linear combination of γ_sens and δf_sens shown in the above equation 5-7. As a result, the friction coefficient operation gain Gmu is set such that the smaller the μ sensitivity p is, the smaller the Gmu is. Accordingly, the situation in which the value of the μ sensitivity p is “0” or a value close to “0” as when the vehicle 1 is traveling straight ahead, in other words, Mnsp_err does not depend on the error of the road surface friction coefficient estimated value μ_estm. It is possible to prevent the road surface friction coefficient estimated value μ_estm from being excessively updated in a situation where a relatively large amount of unnecessary components are likely to be included. Therefore, the robustness of the estimation of the road friction coefficient μ can be improved, and Mnsp_err is reflected in the update of the road friction coefficient estimation value μ_estm by adapting the dependency of Mnsp_err to the error of the road friction coefficient estimation value μ_estm. Can be made. As a result, the estimation accuracy of the road surface friction coefficient μ can be improved, and the robustness of the estimation process can be improved.

[Second Embodiment]
Next, a second embodiment of the present invention will be described with reference to FIG.

  This embodiment is different from the first embodiment only in the process of the friction coefficient increase / decrease manipulated variable determination unit 26e of the μ estimation means 26 (the process of S122-5 in FIG. 12). In this case, in the present embodiment, the μ estimation unit 26 determines that the NSP yaw moment estimation error includes the NSP yaw moment detected value Mnsp_sens and the NSP yaw moment estimated value Mnsp_estm. The updating of the road surface friction coefficient estimated value μ_estm is stopped according to Mnsp_err, and when the update stop condition is not satisfied, the road surface friction coefficient estimated value μ_estm is updated according to Mnsp_err.

  Here, Mnsp_sens and Mnsp_estm have meanings as the values of the same actual NSP yaw moment Mnsp_act estimated by different approaches. The polarity (direction) of the actual NSP yaw moment Mnsp_act can be either positive or negative depending on the traveling state of the vehicle 1. Therefore, in a situation where Mnsp_act ≠ 0, Mnsp_sens and Mnsp_estm should be moments having the same polarity (moments in the same direction). In a situation where Mnsp_sens and Mnsp_estm have different polarities, the error of Mnsp_sens or Mnsp_estm is relatively larger than the absolute value of Mnsp_sens or Mnsp_estm, and the reliability of the value of Mnsp_sens or Mnsp_estm is low (S / N ratio) Is low). Therefore, in this situation, when the road surface friction coefficient estimated value μ_estm is updated according to the NSP yaw moment estimation error Mnsp_err calculated from Mnsp_sens and Mnsp_estm, the absolute value of Mnsp_err further increases, and consequently the road surface friction coefficient estimated value. μ_estm may diverge.

  Therefore, in this embodiment, when at least the polarities of Mnsp_sens and Mnsp_estm are different from each other (opposite polarities), updating the road surface friction coefficient estimated value μ_estm according to Mnsp_err is stopped.

  Specifically, in this embodiment, the friction coefficient increasing / decreasing manipulated variable determiner 26e of the μ estimating means 26 determines the friction coefficient increasing / decreasing manipulated variable Δμ by executing the processing shown in the flowchart of FIG.

  In the following, the friction coefficient increase / decrease manipulated variable determiner 26e first determines whether the condition of Mnsp_estm> Mm and Mnsp_sens> Ms or the condition of Mnsp_estm <−Mm and Mnsp_sens <−Ms is satisfied in S122-5-1. Judge whether or not. Here, Mm and Ms are predetermined non-negative values (positive values near “0” or “0”).

  The determination process of S122-5-1 is a process of determining whether or not the update cancellation condition is satisfied, and that the determination result of S122-5-1 is negative means that the update cancellation condition is satisfied. It means to do. In this case, if the predetermined values Mm and Ms are set to “0”, the determination result in S122-5-1 is negative (the update stop condition is satisfied). And Mnsp_sens are equivalent to having different polarities. On the other hand, when the predetermined values Mm and Ms are positive values, −Mm ≦ Mnsp_estm ≦ Mm or −Ms ≦ Mnsp_sens ≦ Ms is satisfied not only when Mnsp_estm and Mnsp_sens have different polarities. In other words (in other words, when Mnsp_estm or Mnsp_sens is a value in the range near “0”), the determination result in S122-5-1 is negative (the update stop condition is satisfied). Become.

  Next, the friction coefficient increasing / decreasing manipulated variable determiner 26e adjusts the friction coefficient operating gain Gmu in S122-5-2 or S122-5-3 (a value to be multiplied by the NSP yaw moment error Mnsp_err together with the μ sensitivity p). A gain adjustment parameter Kmu_att that is a parameter for changing from the basic gain Kmu) is set according to the determination result of S122-5-1.

  Specifically, the friction coefficient increase / decrease manipulated variable determiner 26e determines the value of Kmu_att in S122-5-2 if the determination result in S122-5-1 is affirmative (if the update stop condition is not satisfied). Is set to “1” and the determination result is negative (when the update stop condition is satisfied), the value of Kmu_att is set to “0” in S122-5-3 (S122-5-5). 3).

  Next, in S122-5-4, the friction coefficient increasing / decreasing manipulated variable determiner 26e calculates the friction coefficient increasing / decreasing manipulated variable Δμ by the following equation 7-2a obtained by adding the gain adjustment parameter Kmu_att to the equation 7-2.

Δμ = Mnsp_err * Gmu
= Mnsp_err * (p (γ_sens, δf_sens, Vgx_estm) * Kmu * Kmu_att)
... Formula 7-2a

Therefore, in this embodiment, a value obtained by further multiplying p * Kmu by Kmu_att is a friction coefficient operation gain Gmu, and by multiplying this Gmu (= p * Kmu * Kmu_att) by Mnsp_err, the friction coefficient increasing / decreasing manipulated variable Δμ is It is determined.

  The present embodiment is the same as the first embodiment except for the processing of the friction coefficient increasing / decreasing manipulated variable determining unit 26e described above. This embodiment has the following effects in addition to the effects exhibited in the first embodiment.

  That is, in the situation where the NSP yaw moment detected value Mnsp_sens and the NSP yaw moment estimated value Mnsp_estm have different polarities by determining the friction coefficient increasing / decreasing manipulated variable Δμ as described above, the determination result of S122-5-1 is Since the result is negative, the value of the gain adjustment parameter Kmu_att is set to “0”. As a result, Δμ is forcibly set to “0”. Therefore, updating the road surface friction coefficient estimated value μ_estm according to Mnsp_err is stopped, and μ_estm is held at a value immediately before the determination result of S122-5-1 becomes negative.

  Thereby, it is possible to prevent the road surface friction coefficient estimated value μ_estm from diverging in a situation where Mnsp_sens and Mnsp_estm have different polarities.

  If the determination result in S122-5-1 is affirmative, that is, if the update stop condition is not satisfied, the value of Kmu_att is set to “1”. Therefore, the road surface friction coefficient estimated value μ_estm is updated according to the NSP yaw moment estimation error Mnsp_err.

  Supplementally, when the predetermined values Mm and Ms in the determination process of S122-5-1 are set to positive values, as described above, Mnsp_sens or Mnsp_estm is a value within the range near “0”. In addition, since the determination result in S122-5-1 is negative, Δμ is forcibly set to “0”. Therefore, not only when Mnsp_estm and Mnsp_sens have different polarities, but also when the error of Mnsp_sens or Mnsp_estm tends to be relatively large compared to the magnitude of the actual NSP yaw moment Mnsp_act, μ_estm depends on Mnsp_err. Can be canceled.

  In the present embodiment, the value of the gain adjustment parameter Kmu_att is determined according to the determination result of S122-5-1. However, the determination result of S122-5-1 is positive without using Kmu_att. In some cases, Δμ may be determined by the equation 7-2, and Δμ may be set to “0” when the determination result is negative.

[Third Embodiment]
Next, a third embodiment of the present invention will be described with reference to FIGS. This embodiment is different from the second embodiment only in how to set the gain adjustment parameter Kmu_att in the friction coefficient increasing / decreasing manipulated variable determiner 26e.

  In the second embodiment, Kmu_att is always set to “1” when the determination result in S122-5-1 is affirmative (when the update cancellation condition is not satisfied). On the other hand, in the present embodiment, when the determination result in S122-5-1 is positive, for example, as shown in FIG. To change.

  FIG. 14 visually shows a set value of Kmu_att corresponding to a set of the value of Mnsp_estm and the value of Mnsp_sens on the coordinate plane with Mnsp_estm as the abscissa axis and Mnsp_sens as the ordinate axis. “0” (excluding “0” at the intersection (origin) of the ordinate and abscissa axes), “0.5”, and “1” represent typical examples of the set value of Kmu_att. In the example shown in FIG. 14, the point (Mnsp_estm, Mnsp_sens) as a set of the value of Mnsp_estm and the value of Mnsp_sens is the second quadrant (region where Mnsp_estm <0 and Mnsp_sens> 0) or the fourth quadrant (Mnsp_estm). > 0 and Mnsp_sens <0), that is, when Mnsp_estm and Mnsp_sens have different polarities, Kmu_att is always set to “0”.

  Further, when the point (Mnsp_estm, Mnsp_sens) is in the first quadrant (region where Mnsp_estm> 0 and Mnsp_sens> 0), the point (Mnsp_estm, Mnsp_sens) is on the half line L02a or L04a in the figure. Kmu_att is set to “0”, “0.5”, and “1”, respectively, when they exist on the half line L12a or L14a, or on the half line L22a or L24a. The In the area between the half lines L02a and L04a and the ordinate and abscissa axes in the first quadrant, Kmu_att is always set to “0”. In the first quadrant, Kmu_att is always set to “1” in the region above the half line L24a (the side where Mnsp_sens is larger) and the right side of the half line L22a (the side where Mnsp_estm is larger). The Further, in the region between the half lines L02a and L22a, when the value of Mnsp_sens is constant, Kmu_att is set so that Kmu_att continuously changes between “0” and “1” according to Mnsp_estm. Is set. Similarly, in the region between the half lines L04a and L24a, when the value of Mnsp_estm is constant, Kmu_att changes so that Kmu_att continuously changes between “0” and “1” according to Mnsp_sens. Is set.

  Further, when the point (Mnsp_estm, Mnsp_sens) exists in the third quadrant (region where Mnsp_estm <0 and Mnsp_sens <0), Kmu_att is symmetric with respect to Kmu_att set in the first quadrant. Is set to be That is, when Kmu_att in the first quadrant is expressed as Kmu_att = f_kmuatt (Mnsp_estm, Mnsp_sens) as a function of Mnsp_estm and Mnsp_sens, Kmu_att in the third quadrant is set to be Kmu_att = f_kmuatt (−Mnsp_estm, −Mnsp_sens) Is done. In this case, the half lines L02b, L12b, L22b, L04b, L14b, and L24b in the third quadrant in FIG. 14 correspond to the half lines L02a, L12a, L22a, L04a, L14a, and L24a in the first quadrant, respectively.

  The friction coefficient increasing / decreasing manipulated variable determiner 26e in this embodiment determines the road surface friction coefficient increasing / decreasing manipulated variable Δμ while setting Kmu_att as described above, for example, by executing the processing shown in the flowchart of FIG. In FIG. 15, the same reference numerals as those in FIG. 13 are used for the same processes as those in the flowchart of FIG. 13 in the second embodiment.

  In the following, the friction coefficient increasing / decreasing manipulated variable determiner 26e first executes the same determination process as in the second embodiment in S122-5-1. If the determination result in S122-5-1 is affirmative (when the update suspension condition is not satisfied), the friction coefficient increase / decrease manipulated variable determiner 26e performs steps S122-5-6 to S122-5-10. By executing the process, the value of the gain adjustment parameter Kmu_att is set. When the determination result is negative (when the update stop condition is satisfied), in S122-5-3, the gain adjustment parameter Kmu_att is set. Set the value to “0”.

  In the processing of S122-5-6 to S122-5-10, the friction coefficient increasing / decreasing manipulated variable determiner 26e first sets the value of the parameter w1 to the value of Mnsp_estm in S122-5-6 according to the equation shown in the figure. It is determined according to the absolute value (abs (Mnsp_estm)). This parameter w1 is a parameter that defines the form of change in Kmu_att according to the absolute value of Mnsp_estm when the value of Mnsp_sens is constant. In this case, in the example of the present embodiment in which Kmu_att is set as shown in FIG. 14, C1 and C2 in the expression in S122-5-6 are set to positive predetermined values in advance.

  Next, in S122-5-7, the friction coefficient increase / decrease manipulated variable determiner 26e determines the value of the parameter w2 as the first candidate value of Kmu_att, the value of w1 and the absolute value of Mnsp_sens ( abs (Mnsp_sens)). In this case, in the example of the present embodiment in which Kmu_att is set as shown in FIG. 14, C3 in the expression in S122-5-7 is set to a predetermined negative value in advance. It should be noted that the relationship between Mnsp_estm and Mnsp_sens when the value of w2 is “0”, “0.5”, or “1” in the equation in S122-5-7 is the half line L02a, This is a relationship between Mnsp_estm and Mnsp_sens on L02b, or L12a, L12b, or L22a, L22b.

  Next, in S122-5-8, the friction coefficient increasing / decreasing manipulated variable determiner 26e determines the value of the parameter w3 according to the absolute value of Mnsp_sens (abs (Mnsp_sens)) using the formula shown in the figure. This parameter w3 is a parameter that defines the form of change in Kmu_att according to the absolute value of Mnsp_sens when the value of Mnsp_estm is constant. In this case, in the example of the present embodiment in which Kmu_att is set as shown in FIG. 14, C4 and C5 in the expression in S122-5-8 are set to positive predetermined values in advance.

  Next, in S122-5-9, the friction coefficient increase / decrease manipulated variable determiner 26e sets the value of the parameter w4 as the second candidate value of Kmu_att to the value of w3 and the absolute value of Mnsp_estm ( abs (Mnsp_estm)). In this case, in the example of the present embodiment in which Kmu_att is set as shown in FIG. 14, C6 in the expression in S122-5-9 is set to a predetermined negative value in advance. Note that the relationship between Mnsp_estm and Mnsp_sens when the value of w4 is “0”, “0.5”, or “1” in the expression in S122-5-9 is the half line L04a, This is the relationship between Mnsp_estm and Mnsp_sens on L14b, or L14a, L14b, or L24a, L24b.

  Next, the friction coefficient increasing / decreasing manipulated variable determiner 26e determines Kmu_att according to the formula in the figure in S122-5-10.

  As described above, by performing the processing of S122-5-6 to S122-5-10, the value of Kmu_att in the first quadrant and the third quadrant of the coordinate plane shown in FIG. 14 is set as shown in FIG. Will be.

  This embodiment is the same as the second embodiment except for the matters described above. In this embodiment, as described above, by setting the value of Kmu_att, when the road surface friction coefficient estimated value μ_estm is updated according to Mnsp_err (when the update stop condition is not satisfied), Mnsp_sens or Mnsp_estm is compared. When the target is close to “0”, the magnitude of the friction coefficient operation gain Gmu decreases as the Mnsp_sens or Mnsp_estm approaches “0”. As a result, the absolute value of the update amount (friction coefficient increasing / decreasing manipulated variable Δμ) of the road surface friction coefficient estimated value μ_estm is suppressed to be small. Accordingly, the Mnsp_sens or Mnsp_estm is close to “0” and the error of the Mnsp_sens or Mnsp_estm tends to become relatively large with respect to the magnitude of the actual NSP yaw moment Mnsp_act. Can be suppressed.

[Fourth Embodiment]
Next, a fourth embodiment of the present invention will be described with reference to FIG. This embodiment is different from the third embodiment only in how to set the gain adjustment parameter Kmu_att in the friction coefficient increasing / decreasing manipulated variable determiner 26e.

  That is, in the present embodiment, not only the polarities of the NSP yaw moment estimated value Mnsp_estm and the NSP yaw moment detected value Mnsp_sens but also the polarity of the μ sensitivity p (γ_sens, δf_sens, Vgx_estm) are considered, and at least the conditions regarding these polarities are set. When the predetermined update stop condition including is satisfied, updating the road surface friction coefficient estimated value μ_estm according to the yaw moment estimation error Mnsp_err is stopped. Here, as is apparent from the equation 4-13b, the μ sensitivity p as the ratio of the increase amount of the actual NSP yaw moment Mnsp to the increase amount of the actual road surface friction coefficient μ_act has the same polarity as the actual NSP yaw moment Mnsp_act. It should be. Therefore, in this embodiment, when any one of Mnsp_estm, Mnsp_sens, and p is different from the other two polarities, it is assumed that the update stop condition is satisfied, and μ_estm is updated according to Mnsp_err. Discontinue.

  Specifically, in the present embodiment, the friction coefficient increasing / decreasing manipulated variable determiner 26e determines the friction coefficient increasing / decreasing manipulated variable Δμ by executing the processing shown in the flowchart of FIG. In FIG. 16, the same reference numerals as those in FIG. 15 are used for the same processes as those in the flowchart of FIG. 15 in the third embodiment.

  In the process shown in the flowchart of FIG. 16, the determination process of S122-5-20 is performed instead of the determination process of S122-5-1 of FIG. 15 in the third embodiment, and the other processes are the same as those of the third embodiment. The form is the same.

  In this case, in the determination process of S122-5-20, is the condition that Mnsp_estm> Mm and Mnsp_sens> Ms and p> p0, or the condition that Mnsp_estm <−Mm, Mnsp_sens <−Ms, and p <−p0 holds? Judge whether or not. Here, Mm, Ms, and p0 are predetermined non-negative values (positive values near “0” or “0”).

  In the present embodiment, the negative determination result in S122-5-20 means that the update cancellation condition is satisfied. In this case, if the values of the predetermined values Mm, Ms, and p0 are set to “0”, the determination result in S122-5-20 is negative (the update stop condition is satisfied). , Mnsp_estm, Mnsp_sens, and p are equivalent to the polarity different from the other two polarities. On the other hand, when the predetermined values Mm, Ms, and p0 are set to positive values, the polarity of any one of Mnsp_estm, Mnsp_sens, and p is different from the other two polarities. Not only −Mm ≦ Mnsp_estm ≦ Mm or −Ms ≦ Mnsp_sens ≦ Ms or −p0 ≦ p ≦ p0 (in other words, any of Mnsp_estm, Mnsp_sens, and p is within the range of “0”) The determination result in S122-5-20 is negative (the update stop condition is satisfied) also in this case.

  This embodiment is the same as the third embodiment except for the matters described above.

  In this embodiment, when any one of Mnsp_estm, Mnsp_sens, and p has a different polarity from the other two polarities, the updating of μ_estm according to Mnsp_err is stopped, so that it is more reliable. In addition, it is possible to prevent μ_estm from diverging.

[Fifth Embodiment]
Next, a fifth embodiment of the present invention will be described with reference to FIG. This embodiment is different from the fourth embodiment only in how to set the gain adjustment parameter Kmu_att in the friction coefficient increasing / decreasing manipulated variable determiner 26e.

  In the fourth embodiment, when the determination result of S122-5-20 is affirmative (when the update stop condition is not satisfied), the road surface friction coefficient estimated value μ_estm is always updated according to the NSP yaw moment error Mnsp_err. I tried to do it. On the other hand, in this embodiment, when the determination result of S122-5-20 becomes negative (when the update cancellation condition is satisfied), the determination result of S122-5-20 is positive after that. The update of μ_estm is executed in accordance with Mnsp_err only when the state (the state where the update cancellation condition is not satisfied) continues for a predetermined time or longer. In other words, in the present embodiment, once the update stop condition is satisfied, the state in which the update stop condition is not satisfied continues for a predetermined time or more as a necessary condition for updating μ_estm according to Mnsp_err. To do.

  Specifically, in the present embodiment, the friction coefficient increasing / decreasing manipulated variable determiner 26e determines the friction coefficient increasing / decreasing manipulated variable Δμ by executing the processing shown in the flowchart of FIG. In FIG. 17, the same reference numerals as those in FIG. 16 are used for the same processes as those in the flowchart of FIG. 16 in the fourth embodiment.

  In the following, the friction coefficient increasing / decreasing manipulated variable determiner 26e first performs the same determination process (determination process as to whether or not the update stop condition is satisfied) in S122-5-20 as in the fourth embodiment. Then, when the determination result in S122-5-20 is negative (when the update stop condition is satisfied), the friction coefficient increase / decrease manipulated variable determiner 26e sets the value of the countdown timer TM in S122-5-21. After setting to a predetermined initial value Twait, the determination process of S122-5-22 is executed. The friction coefficient increase / decrease manipulated variable determiner 26e executes the determination process of S122-5-22 as it is when the determination result of S122-5-20 is affirmative (when the update stop condition is not satisfied). .

  In the determination process of S122-5-22, the friction coefficient increase / decrease manipulated variable determiner 26e determines whether or not the current value of the countdown timer TM is equal to or less than “0” (the time measurement for the initial value Twait has been completed). Whether or not).

  If the determination result in S122-5-22 is affirmative, the friction coefficient increase / decrease manipulated variable determiner 26e executes the processes of S122-5-6 to S122-5-14 described in the third embodiment. By doing so, the value of the gain adjustment parameter Kmu_att is set.

  On the other hand, if the determination result in S122-5-22 is negative, the friction coefficient increase / decrease manipulated variable determiner 26e sets the value of the countdown timer TM to the time corresponding to the calculation processing period ΔT in S122-5-23. Only decrease. Further, the friction coefficient increasing / decreasing manipulated variable determiner 26e sets the value of the gain adjustment parameter Kmu_att to “0” in S122-5-24.

  Next, the friction coefficient increasing / decreasing manipulated variable determiner 26e executes the same processing as in the second embodiment in S122-5-4 to determine the friction coefficient increasing / decreasing manipulated variable Δμ.

  As a result of the above-described processing, once the determination result in S122-5-20 becomes negative (when the update stop condition is satisfied), the determination result in S122-5-20 becomes positive (update stop). Even if the determination result in S122-5-20 is affirmative until the condition does not hold) for a predetermined time or more defined by the initial value Twait of the countdown timer TM, the value of the gain adjustment parameter Kmu_att Is set to “0”. Therefore, a state in which updating of μ_estm according to Mnsp_err is stopped is maintained. If the state in which the determination result in S122-5-20 is affirmative (the state in which the update stop condition is not satisfied) continues for a predetermined time or more defined by the initial value Twait of the countdown timer TM, the response is based on Mnsp_err Then, updating μ_estm is resumed.

  The present embodiment is the same as the fourth embodiment except for the matters described above. According to the present embodiment, when the state in which the update stop condition is satisfied is shifted to the state in which the update stop condition is not satisfied, Mnsp_err is set to Mnsp_err in the period immediately after the transition (the period corresponding to the time of the initial value Twait). Accordingly, updating μ_estm is prohibited. For this reason, it is possible to prevent the road surface friction coefficient estimated value μ_estm from being updated to an inappropriate value when the update stop condition is temporarily not satisfied due to disturbance or the like.

  In the fourth embodiment and the fifth embodiment, when the determination result of S122-5-20 is affirmative (when the update stop condition is not satisfied), the process of S122-5-6 of the third embodiment is performed. Although the gain adjustment parameter Kmu_att is determined by the process of S122-5-14, when the determination result of S122-5-20 becomes affirmative (when the update stop condition is not satisfied), the value of Kmu_att is set to the second embodiment. Similarly to the above, it may be set to “1”.

  In the fifth embodiment, the update cancellation condition is the same as that in the fourth embodiment. However, the same update cancellation condition as in the second and third embodiments may be used. That is, in the fifth embodiment, the determination process of S122-5-1 may be performed instead of the determination process of S122-5-20.

  In the second to fifth embodiments, the value of the gain adjustment parameter Kmu_att is used when the update cancellation condition is satisfied (when the determination result of S122-5-1 or S122-5-20 is negative). Instead, the friction coefficient increasing / decreasing manipulated variable Δμ may be set to “0”. Alternatively, instead of setting the friction coefficient increasing / decreasing manipulated variable Δμ to “0”, the value of Δμ is set to a predetermined positive value, and the road surface friction coefficient estimated value μ_estm is constant in a state where the update stop condition is satisfied. You may make it increase gradually with the time increase rate of.

[Sixth Embodiment]
Next, a sixth embodiment of the present invention will be described with reference to FIG. This embodiment is different from the first embodiment only in part of the processing of the μ estimation means 26. Specifically, in this embodiment, the μ estimation means 26 has a saturation characteristic element 26g for inputting the μ sensitivity p (γ_sens, δf_sens, Vgx_estm) calculated by the μ sensitivity calculation unit 26d. The saturation characteristic element 26g generates an output having a saturation characteristic with respect to the input μ sensitivity p (function value of the μ sensitivity p), and the output is hereinafter referred to as a μ sensitivity dependent value p_a. In this case, in the saturation characteristic element 26g, the relationship between the μ sensitivity p and the μ sensitivity dependent value p_a is set in advance in the form of map data or an arithmetic expression. Specifically, the relationship between the μ sensitivity p and the μ sensitivity dependent value p_a is such that when p is “0”, p_a becomes “0”, and p_a monotonously increases as p increases, As the absolute value of p increases, the rate of change in p_a relative to the increase in p (the value obtained by differentiating p_a by p) decreases as the absolute value of p increases (the value of p_a is saturated). To be set).

  Then, the friction coefficient increase / decrease manipulated variable determination unit 26e of the μ estimation means 26 in the present embodiment performs the calculation of the right side of the equation 7-2 using the μ sensitivity dependent value p_a instead of the μ sensitivity p. The friction coefficient increasing / decreasing manipulated variable Δμ is determined. In other words, Δμ is determined by the calculation of the following equation 7-2b.

Δμ = Mnsp_err * Gmu
= Mnsp_err * (p_a * Kmu) ...... Equation 7-2b

The present embodiment is the same as the first embodiment except for the matters described above. Therefore, in this embodiment, Δμ is determined so as to be proportional to the product of Mnsp_err and μ sensitivity-dependent value p_a.

  In this embodiment, when the friction coefficient increasing / decreasing manipulated variable Δμ is determined according to Mnsp_err, the magnitude of the friction coefficient operating gain Gmu (feedback gain) becomes excessive when the absolute value of μ sensitivity p is large. Is suppressed. As a result, it is possible to prevent the road surface friction coefficient estimated value μ_estm calculated by the μ estimating means 26 from fluctuating in an unstable manner or vibrating.

  Supplementally, the determination of the friction coefficient increasing / decreasing manipulated variable Δμ using the μ sensitivity dependent value p_a as in the present embodiment can also be applied to the second to fifth embodiments. In this case, in the process of S122-5-4 in the second to fifth embodiments (the process of calculating Δμ when the update stop condition is not satisfied), the μ sensitivity dependent value p_a is used instead of the μ sensitivity p. Is used to determine the friction coefficient increasing / decreasing manipulated variable Δμ by performing the calculation of the right side of the expression 7-2a. In other words, Δμ may be determined by the following equation 7-2c.

Δμ = Mnsp_err * Gmu
= Mnsp_err * (p_a * Kmu * Kmu_att) ...... Equation 7-2c

When the μ sensitivity dependent value p_a is used in the process of S122-5-4 in the fourth embodiment or the fifth embodiment as described above, the determination process of S122-5-20 (the polarity of the μ sensitivity p) When the update cancellation condition determination process is added), the μ sensitivity p is used as it is, or the μ sensitivity p is used instead of the μ sensitivity p. A determination process may be performed.

[Seventh Embodiment]
Next, a seventh embodiment of the present invention will be described with reference to FIG. This embodiment is different from the first embodiment only in part of the processing of the μ estimation means 26.

  Specifically, in the present embodiment, the μ estimation unit 26 includes the NSP yaw moment estimated value Mnsp_estm calculated by the Mnsp_estm calculator 26b, the NSP yaw moment detected value Mnsp_sens calculated by the Mnsp_sens calculator 26a, Frequency component adjusting filters 26ba, 26aa, 26da for inputting the μ sensitivity p (γ_sens, δf_sens, Vgx_estm) calculated by the μ sensitivity calculation unit 26d, and the saturation characteristic element 26g described in the sixth embodiment. And have.

  In this example, the filters 26ba, 26aa, and 26da all have low cut characteristics (characteristics that block low frequency components below a predetermined frequency). More specifically, each of the filters 26ba, 26aa, and 26da has a transfer function expressed by, for example, Ta * S / (1 + Ta * S), and the frequency characteristics of the filters 26ba, 26aa, and 26da are the same target characteristics (low cut characteristics). (The time constants Ta of the transfer functions are set to be the same). Note that Mnsp_err in a state in which μ_estm accurately matches μ_act due to, for example, the difference in frequency characteristics of the sensors used for generating the input values of the filters 26ba, 26aa, and 26da, for example. When a phase shift from p or a phase shift between Mnsp_sens and Mnsp_estm occurs, the frequency characteristics of the filters 26ba, 26aa, and 26da are shifted from each other so as to eliminate the phase shift. It may be.

  In this embodiment, the μ estimator 26 detects the NSP yaw moment filtering detection value Mnsp_sens_f, which is the output of the filter 26aa that inputs Mnsp_sens, instead of the deviation (NSP yaw moment estimation error) between Mnsp_sens and Mnsp_estm, and Mnsp_estm. The NSP yaw moment filtering estimated error Mnsp_err_f (= Mnsp_sens_f−Mnsp_estm_f), which is a deviation from the NSP yaw moment filtering estimated value Mnsp_estm_f, which is the output of the filter 26ba is input by the Mnsp_err calculating unit 26c. In the present embodiment, since the frequency characteristics of the filters 26aa and 26ba are the same, obtaining the deviation Mnp_err_f between Mnsp_sens_f and Mnsp_estm_f as described above is an NSP yaw moment error that is a deviation between Mnsp_sens and Mnsp_estm. This is equivalent to obtaining Mnsp_err_f by passing Mnsp_err through a filter having the same frequency characteristics as the filters 26aa and 26ba. Therefore, instead of the filters 26aa and 26ba, a filter for inputting Mnsp_err (a filter having the same frequency characteristics as the filters 26aa and 26ba) may be provided, and Mnsp_err_f may be obtained by passing Mnsp_err through this filter.

  Further, the μ estimation means 26 inputs the μ sensitivity filtering value p_f, which is the output of the filter 26da that inputs the μ sensitivity p, into the saturation characteristic element 26g instead of the μ sensitivity p, so that the function value of p_f is obtained. The μ sensitivity dependent value p_fa is obtained. In this case, the relationship between p_f and p_fa is the same as the relationship between the input (p) and the output (p_a) of the saturation characteristic element 26g described in the sixth embodiment. Note that the μ sensitivity dependent value p_fa may be obtained by inputting the value (p_a) obtained by passing the μ sensitivity p through the saturation characteristic element 26g to the filter 26da.

  In this embodiment, the friction coefficient increase / decrease manipulated variable determination unit 26e of the μ estimation means 26 replaces the NSP yaw moment estimation error Mnsp_err and μ sensitivity p with the NSP yaw moment filtering estimation error Mnsp_err_f and μ sensitivity dependent value, respectively. The friction coefficient increasing / decreasing manipulated variable Δμ is determined by calculating the right side of Expression 7-2 using p_fa. In other words, Δμ is determined by the following equation 7-2d.

Δμ = Mnsp_err_f * Gmu
= Mnsp_err_f * (p_fa * Kmu) ...... Equation 7-2d

The present embodiment is the same as the first embodiment except for the matters described above. Therefore, in this embodiment, Δμ is determined so as to be proportional to the product of Mnsp_err_f and μ sensitivity-dependent value p_fa.

  In the present embodiment, the filters 26ba, 26aa, and 26da correspond to the first filter, the second filter, and the third filter in the present invention, respectively. The NSP yaw moment filtering estimated value Mnsp_estm_f corresponds to the first estimated filtering value in the present invention, and the NSP yaw moment filtering detected value Mnsp_sens_f corresponds to the second estimated filtering value in the present invention.

  In this embodiment, the friction coefficient increasing / decreasing manipulated variable Δμ is determined using the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa obtained by using the filters 26ba, 26aa, 26da having low cut characteristics. Thus, unnecessary components included in Mnsp_sens, Mnsp_estm, and p due to steady offsets and drifts in the output of sensors such as the yaw rate sensor 13 and the lateral acceleration sensor 15 or the actual road surface bank angle θbank_act are removed. Thus, the road surface friction coefficient estimated value μ_estm can be obtained. As a result, the accuracy of μ_estm can be increased.

  Supplementally, the determination of the friction coefficient increasing / decreasing manipulated variable Δμ using the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa as in the present embodiment is also applied to the second to fifth embodiments. be able to. In this case, in the processing of S122-5-4 in the second to fifth embodiments (processing for calculating Δμ when the update stop condition is not satisfied), the NSP yaw moment estimation error Mnsp_err and μ sensitivity p Instead, the friction coefficient increasing / decreasing manipulated variable Δμ may be determined by calculating the right side of the equation 7-2a using the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa. That is, Δμ may be determined by the following equation 7-2e.

Δμ = Mnsp_err_f * Gmu
= Mnsp_err_f * (p_fa * Kmu * Kmu_att) ...... Equation 7-2e

Further, regarding the determination process regarding the update cancellation condition, when performing the determination process of S122-5-1, the determination process may be performed using Mnsp_sens_f and Mnsp_estm_f instead of Mnsp_sens and Mnsp_estm, respectively. Further, when performing the determination process of S122-5-20, the determination process may be performed using Mnsp_sens_f, Mnsp_estm_f, and p_fa (or p_f) instead of Mnsp_sens, Mnsp_estm, and p, respectively. .

  In the present embodiment, the μ sensitivity p is passed through both the filter 26da and the saturation characteristic element 26g. However, if the μ sensitivity p does not become so large, the saturation characteristic element 26g is omitted. May be. In this case, the μ sensitivity filtering value p_f may be used in place of the μ sensitivity p in the process of calculating the friction coefficient increasing / decreasing manipulated variable Δμ.

  The filters 26ba and 26aa related to the estimated value and the detected value of the NSP yaw moment among the filters 26ba, 26aa and 26da may be omitted, or the filter 26da related to the μ sensitivity p may be omitted.

[Eighth Embodiment]
Next, an eighth embodiment of the present invention will be described with reference to FIG. This embodiment is different from the first embodiment only in part of the processing of the μ estimation means 26.

  Specifically, the μ estimation means 26 in this embodiment includes filters 26bb, 26ab, 26db having frequency characteristics different from those of the filters 26ba, 26aa, 26da described in the seventh embodiment, and the filters 26bb, 26ab, 26db. The NSP yaw moment estimated value Mnsp_estm, the NSP yaw moment detected value Mnsp_sens calculated by the Mnsp_estm calculating unit 26b, the NSP yaw moment detected value Mnsp_sens calculated by the Mnsp_sens calculating unit 26a, and the μ sensitivity calculating unit 26d, respectively. The μ sensitivity p (γ_sens, δf_sens, Vgx_estm) is input.

  In this example, the filters 26bb, 26ab, and 26db all have high cut characteristics (characteristics that cut off high frequency components of a predetermined frequency or higher). More specifically, each of the filters 26bb, 26ab, and 26db has a transfer function expressed by, for example, 1 / (1 + Tb * S), and the frequency characteristics thereof are the same target characteristics (high cut characteristics). (The time constants Tb of the transfer functions are set to be the same). In this case, the frequency characteristics of the filters 26bb, 26ab, and 26db are low-pass characteristics in other words. Note that Mnsp_err in a state where μ_estm accurately matches μ_act due to, for example, a difference in frequency characteristics of each sensor used to generate each input value of each filter 26bb, 26ab, 26db. When a phase shift from p or a phase shift between Mnsp_sens and Mnsp_estm occurs, the frequency characteristics of the filters 26bb, 26ab, and 26db are shifted from each other so as to eliminate the phase shift. It may be. Further, each of the filters 26bb, 26ab, and 26db is not limited to the low-pass characteristic and may be a band-pass characteristic as long as it has a high cut characteristic.

  In the present embodiment, the μ estimation means 26, instead of the deviation (NSP yaw moment estimation error Mnsp_err) between Mnsp_sens and Mnsp_estm, NSP yaw moment filtering detection value Mnsp_sens_f that is the output of the filter 26ab that inputs Mnsp_sens, An NSP yaw moment filtering estimation error Mnsp_err_f (= Mnsp_sens_f−Mnsp_estm_f) that is a deviation from the NSP yaw moment filtering estimated value Mnsp_estm_f that is an output of the filter 26bb that receives Mnsp_estm is calculated by the Mnsp_err calculating unit 26c. As in the case described in the seventh embodiment, the NSP yaw moment estimation error Mnsp_err, which is a deviation between Mnsp_sens and Msnp_estm, is passed through a filter having the same frequency characteristics (high cut characteristics) as the filters 26ba and 26aa. Mnsp_err_f may be obtained.

  Further, the μ estimation means 26 of the present embodiment has the saturation characteristic element 26g described in the sixth embodiment. Then, the μ estimation means 26 passes the μ sensitivity filtering value p_f, which is the output of the filter 26db to which the μ sensitivity p is input, through the saturation characteristic element 26g instead of the μ sensitivity p as in the seventh embodiment. , Μ sensitivity dependent value p_fa as a function value of p_f is obtained. Note that the μ sensitivity dependent value p_fa may be obtained by passing the value (p_a) obtained by passing the μ sensitivity p through the saturation characteristic element 26g through the filter 26db.

  The friction coefficient increase / decrease manipulated variable determiner 26e of the μ estimation means 26 uses the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent filtering value p_fa, respectively, instead of the NSP yaw moment estimation error Mnsp_err and μ sensitivity p. Then, the provisional value Δμ_a of the friction coefficient increasing / decreasing manipulated variable Δμ is determined by calculating the right side of the expression 7-2. That is, the value obtained by the calculation of the right side of the expression 7-2d is obtained as the provisional value Δμ_a.

  In this embodiment, since the filters 26bb, 26ab, and 26db have high cut characteristics, the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa (or the μ sensitivity filtering value p_f) in a relatively high frequency range are used. A phase lag is generated, and as a result, a phase lag of the provisional value Δμ_a obtained by the calculation on the right side of the equation 7-2d is likely to occur. For this reason, if Δμ_a is used as it is and the road surface friction coefficient estimated value μ_estm is updated, vibration of μ_estm is likely to occur. Therefore, in the present embodiment, the friction coefficient increasing / decreasing manipulated variable determiner 26e is provided with a phase compensation element 26ea for advancing the phase of the provisional value Δμ_a (releasing the phase delay). The phase compensation element 26ea has a transfer function expressed by (1 + Tb * s) / (1 + Tc * S), and the time constant Tb of the numerator is a transfer function expressing the filters 26bb, 26ab, and 26db. Is the same as the time constant Tb of the denominator.

  In this embodiment, the friction coefficient increase / decrease manipulated variable determiner 26e passes the provisional value Δμ_a obtained as described above through the phase compensation element 26ea, thereby obtaining the final friction coefficient increase / decrease manipulated variable Δμ (current value). To decide. The product of the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa is passed through the phase compensation element 26ea, and then the output of the phase compensation element 26ea is multiplied by the basic gain Kmu to increase or decrease the friction coefficient. The amount Δμ (current value) may be determined.

  The present embodiment is the same as the first embodiment except for the matters described above.

  In the present embodiment, the filters 26bb, 26ab, and 26db correspond to the first filter, the second filter, and the third filter in the present invention, respectively. The NSP yaw moment filtering estimated value Mnsp_estm_f corresponds to the first estimated filtering value in the present invention, and the NSP yaw moment filtering detected value Mnsp_sens_f corresponds to the second estimated filtering value in the present invention.

  In this embodiment, the friction coefficient increasing / decreasing manipulated variable Δμ is determined using the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa obtained using the high-cut characteristic filters 26bb, 26ab, 26db. Thus, an unnecessary component included in Mnsp_sens, Mnsp_estm, and p due to high frequency noise included in the output of each sensor such as the yaw rate sensor 13 and the lateral acceleration sensor 15 is removed, and the road surface friction coefficient estimated value μ_estm is obtained. be able to. As a result, the accuracy of μ_estm can be increased. Further, by eliminating the phase delay of Δμ by the phase compensation element 26ea, the μ_estm determined by the μ estimation means 26 is prevented from being vibrated, and the robustness of the estimation process of the road surface friction coefficient μ is improved. Can be increased.

  Supplementally, the determination of the friction coefficient increasing / decreasing manipulated variable Δμ using the NSP yaw moment filtering estimation error Mnsp_err_f and the μ sensitivity dependent value p_fa as in the present embodiment is also applied to the second to fifth embodiments. be able to. In this case, similarly to the case described with respect to the seventh embodiment, in the processing of S122-5-4 in the second to fifth embodiments (processing for calculating Δμ when the update stop condition is not satisfied). As described above, the friction coefficient increase / decrease manipulated variable Δμ may be determined using the phase compensation element 26ea.

  Further, when the determination process of S122-5-1 is performed, the determination process may be performed using Mnsp_sens_f and Mnsp_estm_f, respectively, instead of Mnsp_sens and Mnsp_estm. Further, when performing the determination process of S122-5-20, the determination process may be performed using Mnsp_sens_f, Mnsp_estm_f, and p_fa (or p_f) instead of Mnsp_sens, Mnsp_estm, and p, respectively. .

  In this embodiment, the μ sensitivity p is passed through both the filter 26db and the saturation characteristic element 26g. However, if the μ sensitivity p is not so large, the saturation characteristic element 26g is omitted. May be.

[Ninth Embodiment]
Next, a ninth embodiment of the present invention will be described with reference to FIG. This embodiment is different from the eighth embodiment only in part of the processing of the μ estimation means 26.

  Specifically, in the present embodiment, the μ estimation means 26 includes a dead band processing unit 26ha that receives the NSP yaw moment filtering estimation error Msnp_err_f calculated by the Mnsp_err calculation unit 26c, a μ sensitivity p, the filter 26db, and a saturation. A dead zone processing unit 26hb for inputting a μ sensitivity dependent value p_fa that is passed through the characteristic element 26g is further provided.

  The dead zone processing unit 26 ha outputs “0” when the input value corresponding to this is a value within a predetermined dead zone in the vicinity of “0” set in advance, and the input value is the upper limit value of the dead zone ( > 0) and a value obtained by subtracting the upper limit value and the lower limit value from the input value, respectively, when it is smaller than the lower limit value (= −upper limit value) of the dead zone. The same applies to the dead zone processing unit 26hb. Note that the dead zones of the dead zone processing units 26ha and 26hb need not be in the same range.

  In this embodiment, the NSP yaw moment filtering estimation error Mnsp_err_fa that is the output of the dead zone processing unit 26ha that receives Mnsp_err_f and the μ sensitivity-dependent value p_fb that is the output of the dead zone processing unit 26hb that receives p_fa are Mnsp_err, Instead of μ, it is input to the friction coefficient increasing / decreasing manipulated variable determiner 26e. In this embodiment, the frictional increase / decrease manipulated variable determiner 26e determines the friction coefficient increase / decrease manipulated variable Δμ by the same process as in the eighth embodiment using the input Mnsp_err_fa and p_fb.

  Matters other than those described above are the same as those in the eighth embodiment.

  In this embodiment, in addition to the same effects as those of the eighth embodiment, steady offsets and drifts in the output of sensors such as the yaw rate sensor 13 and the lateral acceleration sensor 15, or the actual road surface bank angle θbank_act Due to this, the stationary unnecessary components included in Mnsp_sens, Mnsp_estm, and p can be removed by the dead zone processing units 26ha and 26hb. As a result, the accuracy of μ_estm can be further increased.

  Supplementally, the determination of the friction coefficient increasing / decreasing manipulated variable Δμ using the dead zone processing units 26 ha and 26 hb as in the present embodiment can also be applied to the second to fifth embodiments. In this case, similarly to the case described with respect to the seventh embodiment, in the processing of S122-5-4 in the second to fifth embodiments (processing for calculating Δμ when the update stop condition is not satisfied). As described above, the friction coefficient increasing / decreasing manipulated variable Δμ may be determined using Mnsp_err_fa and p_fb.

  Further, when the determination process of S122-5-1 is performed, the determination process may be performed using the respective filtering values Mnsp_sens_f and Mnsp_estm_f instead of Mnsp_sens and Mnsp_estm. Further, when performing the determination process of S122-5-20, the determination process may be performed using Mnsp_sens_f, Mnsp_estm_f, and p_fb (or p_f) instead of Mnsp_sens, Mnsp_estm, and p, respectively. .

  Further, in the present embodiment, the dead zone processing unit 26ha that inputs Mnsp_err_f and the dead zone processing unit 26hb that inputs p_fa are provided, but either one of the dead zone processing units 26ha or 26hb may be omitted.

  In the first to ninth embodiments described above, in order to obtain the NSP yaw moment estimated value Mnsp_estm, the driving / braking force estimated value Fsubx_i_estm and the lateral force estimated value Fsuby_i_estm of each wheel 2-i are obtained, Although Mnsp_estm is calculated based on the estimated value, the actual NSP yaw moment Mnsp_act is generally highly dependent on the lateral force of the lateral force and driving / braking force of each wheel 2-i, and driving・ Low dependence on braking force. Therefore, the determination of the driving / braking force estimated value Fsubx_i_estm of each wheel 2-i may be omitted. In this case, for example, the lateral translation force acting on the center of gravity of the vehicle 1 by the resultant force of the lateral force estimated value Fsuby_i_estm of the wheel 2-i (i = 1, 2, 3, 4) and the lateral force estimated value Fsuby_i_estm. Are obtained as the total road surface reaction force combined lateral force estimated value Fgy_total_estm and the total road surface reaction force yaw moment Mgz_total_estm, respectively, and from the obtained Fgy_total_estm and Mgz_total_estm, respectively. What is necessary is just to obtain | require NSP yaw moment estimated value Mnsp_estm by said Formula 7-1.

  Also, the skid motion of the vehicle 1 is highly dependent on the lateral force of the wheels 2-i and the driving / braking force. Therefore, when the state quantity of the side-slip motion of the vehicle 1 is estimated by the vehicle motion estimation unit 24d, the vehicle 1 is determined by the resultant force of the lateral force estimated value Fsuby_i_estm of the wheels 2-i (i = 1, 2, 3, 4). Only the lateral translational force acting on the center of gravity point may be regarded as the entire lateral external force (translational force) acting on the center of gravity of the vehicle 1, and the amount of state of the skid motion of the vehicle 1 may be estimated. . For example, the value of the lateral translation force acting on the center of gravity of the vehicle 1 due to the resultant force of the lateral force estimated value Fsuby_i_estm of the wheel 2-i (i = 1, 2, 3, 4) is calculated as the total road surface reaction force combined lateral force. The estimated value Fgy_total_estm is obtained, and the calculation of the equation 1-14a is performed using the Fgy_total_estm, thereby obtaining the estimated vehicle center-of-gravity skid velocity change rate Vgdot_y_estm and integrating the Vgdot_y_estm to obtain the estimated vehicle center-of-gravity skid velocity estimated value Vgy_estm. You may make it ask.

  Further, in the first to ninth embodiments described above, the μ sensitivity p is calculated by the expression 5-7 in the process of S122-4 in FIG. 12, but the μ sensitivity is calculated using the friction characteristic model. It is also possible to calculate. For example, instead of the latest value (previous value) μ_estm_p of the estimated value of the road surface friction coefficient, the value of the road surface friction coefficient (= μ_estm_p + dμ) changed from μ_estm_p by a predetermined small amount dμ is used as shown in FIG. By executing the same processing as S110 and S112, the total road surface reaction force combined lateral force Fgy_total_estm (total road surface reaction force translational force vector estimated value ↑ Fg_total_estm's lateral component when the road surface friction coefficient value is assumed to be μ_estm_p + dμ. ) And the total road surface reaction force yaw moment Mgz_total_estm. Then, by calculating the right side of the equation 7-1 using these Fgy_total_estm and Mgz_total_estm, the NSP yaw moment estimated value when the road surface friction coefficient value is assumed to be μ_estm_p + dμ (hereinafter referred to as Mnsp_estm2) Represent). Then, the NSP yaw moment estimated value Mnsp_estm2 calculated in this way, the NSP yaw moment estimated value Mnsp_estm calculated in S122-2 (the NSP yaw moment estimated value when the road surface friction coefficient value is assumed to be μ_estm_p), and Μ sensitivity p is calculated by dividing the deviation (= Mnsp_estm2−Mnsp_estm) by dμ.

  In the first to ninth embodiments, the NSP yaw moment is used as the comparison target external force in the present invention. However, the comparison target external force in the present invention is not limited to the NSP yaw moment. For example, the lateral component of the vehicle 1 (the Y-axis direction component of the vehicle body coordinate system) of the resultant force of the road surface reaction forces (more specifically, driving / braking force and lateral force) of the front wheels 2-1 and 2-2 is to be compared. It may be used as an external force.

  In the first to ninth embodiments, as the μ sensitivity, the value of p calculated by the equation 5-7 is used. For example, the value of p is the latest value of the road surface friction coefficient estimated value μ_estm. The value obtained by dividing by (previous value) (ratio of p to the latest value of μ_estm) is defined as μ sensitivity, and the friction coefficient increase / decrease manipulated variable Δμ is obtained using μ sensitivity based on this definition instead of p. Also good.

  In the first to ninth embodiments, the bank angle estimation unit 28 and the gradient angle estimation unit 30 are provided. However, in the process of estimating the road surface friction coefficient μ in the first to ninth embodiments, the road surface bank estimation is performed. The value θbank_estm and the road surface slope angle estimated value θslope_estm are unnecessary. Therefore, the bank angle estimation unit 28 and the gradient angle estimation unit 30 may be omitted.

  Further, in the processing of the vehicle model calculation means 24, a vehicle motion model on the premise that the road surface is a horizontal plane is used. However, a vehicle motion model in consideration of the road surface bank angle θbank and the road surface gradient angle θslope may be used. . For example, a vehicle motion model in which the expressions 1-13 and 1-14 are replaced with the following expressions 1-13b and 1-14b, respectively, may be used.

Fgx_total = m * (Vgdot_x−Vgy * γ−g * sin (θslope)) ...... Formula 1-13b
Fgy_total = m * (Vgdot_y + Vgx * γ + g * sin (θbank)) ...... Formula 1-14b

In this case, the vehicle model calculation means 24 estimates the road surface bank angle θbank and the road surface gradient angle θslope while obtaining the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm and the vehicle center-of-gravity skid speed estimated value Vgy_estm, for example, as follows. be able to.

  Specifically, in this case, the vehicle model calculating means 24 estimates the vehicle center-of-gravity longitudinal speed change rate by the following equations 1-13c and 1-14c, respectively, instead of the equations 1-13a and 1-14a. A value Vgdot_x_estm and a vehicle center-of-gravity skid speed change rate estimated value Vgdot_y_estm are calculated.

Vgdot_x_estm = Fgx_total_estm / m + Vgy_estm_p * γ_estm_p
+ G * sin (θslope_estm_p) ...... Equation 1-13c
Vgdot_y_estm = Fgy_total_estm / m-Vgx_estm_p * γ_estm_p
-G * sin (θbank_estm_p) ...... Formula 1-14c

Then, the vehicle model calculation means 24 uses these Vgdot_x_estm and Vgdot_y_estm to determine the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm and the vehicle center-of-gravity skid speed estimated value Vgy_estm, as in the first embodiment. The vehicle center-of-gravity longitudinal speed estimated value Vgx_estm may be matched with the wheel speed selection detection value Vw_i_sens_select.

  Further, the vehicle model calculation means 24 includes a sensor-sensitive longitudinal acceleration estimated value Accx_sensor_estm that is an estimated value of the acceleration to which the longitudinal acceleration sensor 14 is sensitive, and a sensor-sensitive lateral acceleration estimation that is an estimated value of the acceleration to which the lateral acceleration sensor 15 is sensitive. The values Accy_sensor_estm are calculated by the following equations 1-31 and 1-32, respectively.

Accx_sensor_estm = Vgdot_x_estm−Vgy_estm_p * γ_estm_p
-G * sin (θslope_estm_p) ...... Equation 1-31
Accy_sensor_estm = Vgdot_y_estm + Vgx_estm_p * γ_estm_p
+ G * sin (θbank_estm_p) ...... Equation 1-32

It should be noted that Accx_sensor_estm and Accy_sensor_estm are obtained by the calculation of the first term on the right side of Formula 1-13c and the calculation of the first term on the right side of Formula 1-14c, respectively, instead of Formulas 1-31 and 1-32. It may be.

  Here, Accx_sensor_estm obtained as described above means a sensor-sensitive longitudinal acceleration estimated value obtained on the assumption that the previous value (latest value) θslope_estm_p of the road surface slope angle estimated value is accurate. Similarly, Accy_sensor_estm obtained as described above means a sensor-sensitive lateral acceleration estimated value obtained on the assumption that the previous value (latest value) θbank_estm_p of the road surface bank angle estimated value is accurate. Therefore, the deviation between the vehicle center-of-gravity longitudinal acceleration detected value Accx_sens (= sensor-sensitive longitudinal acceleration detected value) based on the output of the longitudinal acceleration sensor 14 and the sensor-sensitive longitudinal acceleration estimated value Accx_sensor_estm is considered to be in accordance with the error of θslope_estm_p. It is done. Similarly, the deviation between the vehicle center-of-gravity lateral acceleration detected value Accy_sens (= sensor-sensitive lateral acceleration detected value) based on the output of the lateral acceleration sensor 15 and the sensor-sensitive lateral acceleration estimated value Accy_sensor_estm corresponds to the error of θbank_estm_p. Conceivable.

  Therefore, the vehicle model calculation means 24 adjusts the road surface gradient angle according to the feedback control law according to the deviation so that the deviation between the vehicle center-of-gravity longitudinal acceleration detection value Accx_sens and the sensor-sensitive longitudinal acceleration estimation value Accx_sensor_estm converges to “0”. A new road surface slope angle estimated value θslope_estm is obtained by updating the estimated value θslope_estm. Similarly, the vehicle model calculation unit 24 sets θbank_estm according to the feedback control law according to the deviation so that the deviation between the vehicle center-of-gravity lateral acceleration detected value Accy_sens and the sensor-sensitive lateral acceleration estimated value Accy_sensor_estm converges to “0”. By updating, a new road surface bank angle estimated value θbank_estm is obtained.

  For example, the vehicle model calculation unit 24 obtains a new road surface slope angle estimated value θslope_estm and a road surface bank angle estimated value θbank_estm by the following equations 1-33 and 1-34, respectively.

θslope_estm = θslope_estm_p + Kslope * (Accx_sens−Accx_sensor_estm)
... Formula 1-33
θbank_estm = θbank_estm_p + Kbank * (Accy_sens−Accy_sensor_estm)
... Formula 1-34

Note that Kslope in Expression 1-33 and Kbank in Expression 1-34 are predetermined values (proportional gain) set in advance, respectively. In this example, θslope_estm and θbank_estm are calculated by integration of deviations (Accx_sens−Accx_sensor_estm) and (Accy_sens−Accy_sensor_estm), respectively.

  As described above, the road surface bank angle θbank and the road surface slope angle θslope can be estimated while obtaining the vehicle center-of-gravity longitudinal speed estimated value Vgx_estm and the vehicle center-of-gravity skid speed estimated value Vgy_estm.

  In this case, it is not necessary to calculate the vehicle center-of-gravity longitudinal acceleration estimated value Accx_estm and the vehicle center-of-gravity lateral acceleration estimated value Accy_estm.

  DESCRIPTION OF SYMBOLS 1 ... Vehicle, 2-1, 2-2, 2-3, 2-4 ... Wheel, S102-S116, S122-2 ... Comparison target external force 1st estimation means, S122-1 ... Comparison target external force 2nd estimation means, 26d (S122-4): μ sensitivity calculation section (μ sensitivity calculation means), 26e (S122-5): Friction coefficient increase / decrease operation amount determination section (friction coefficient increase / decrease operation amount determination means), 26f (S122-6): Friction Coefficient estimated value update unit (friction coefficient estimated value update means).

Claims (8)

A road surface friction coefficient estimating device that estimates while updating a friction coefficient of a road surface on which a vehicle is traveling,
A predetermined type of external force component acting on the vehicle by the resultant force of the road surface reaction force acting on each wheel of the vehicle from the road surface is set as a comparison target external force, and the first estimated value of the comparison target external force is set between the vehicle wheel and the road surface. First comparison target external force estimation obtained using a friction characteristic model representing the relationship between slip and road reaction force, an estimated value of the already determined friction coefficient, and an observation value of a predetermined type of observation target quantity related to vehicle behavior Means,
From the observed value of the motion state quantity of the vehicle that defines the inertial force corresponding to the external force to be compared among the inertial force generated by the motion of the vehicle, the value of the external force component commensurate with the inertial force is obtained, and the obtained external force component A comparison target external force second estimation unit that obtains a value of a second estimation value of the comparison target external force;
Μ sensitivity calculation means for obtaining a value of μ sensitivity, which is a ratio of the increase amount of the external force to be compared to the increase amount of the friction coefficient of the road surface, or a value obtained by dividing the ratio by the value of the friction coefficient of the road surface;
The deviation between the first estimated value and the second estimated value, or the first estimated filtered value and the second estimated value obtained by passing the first estimated value through the first filter for adjusting the frequency component are used for adjusting the frequency component. At least the deviation, the value of the μ sensitivity, or the value of the μ sensitivity, and a third filter for frequency component adjustment so that the deviation of the second estimated filtering value passed through the two filters converges to “0”. Friction coefficient increasing / decreasing manipulated variable determining means for determining an increasing / decreasing manipulated variable of the estimated value of the friction coefficient of the road surface according to the μ sensitivity dependent value passing through one or both of the saturation characteristic elements
Friction coefficient estimated value update means for determining a new estimated value of the friction coefficient by updating the estimated value of the friction coefficient of the road surface according to the increase / decrease operation amount,
The friction coefficient increasing / decreasing manipulated variable determining means determines the increasing / decreasing manipulated variable so as to decrease the increasing / decreasing manipulated variable as the μ sensitivity value or μ sensitivity dependent value decreases. Road friction coefficient estimation device. The road surface friction coefficient estimation device according to claim 1, wherein the friction coefficient increase / decrease manipulated variable determining means determines the increase / decrease according to a deviation / μ sensitivity dependent value which is a product of the deviation and the μ sensitivity value or μ sensitivity dependent value. A road surface friction coefficient estimating apparatus characterized by determining an operation amount. In the road surface friction coefficient estimating device according to claim 2,
The friction coefficient increasing / decreasing manipulated variable determining means determines the increasing / decreasing manipulated variable according to the deviation / μ sensitivity product so as to be proportional to the deviation / μ sensitivity product. In the road surface friction coefficient estimating device according to any one of claims 1 to 3,
The friction coefficient increasing / decreasing manipulated variable determining means according to a deviation between the first estimated filtering value and the second estimated filtering value and a μ sensitivity dependent value obtained by passing the μ sensitivity value through at least the third filter. Means for determining the increase / decrease operation amount;
The road friction coefficient estimating device, wherein the first filter, the second filter, and the third filter are filters having low cut characteristics. In the road surface friction coefficient estimating device according to any one of claims 1 to 3,
The friction coefficient increasing / decreasing manipulated variable determining means according to a deviation between the first estimated filtering value and the second estimated filtering value and a μ sensitivity dependent value obtained by passing the μ sensitivity value through at least the third filter. Means for determining the increase / decrease operation amount;
The road friction coefficient estimating device, wherein the first filter, the second filter, and the third filter are filters having high cut characteristics. In the road surface friction coefficient estimating device according to claim 5,
The friction coefficient increasing / decreasing manipulated variable determining means determines a provisional value of the increasing / decreasing manipulated variable according to the deviation and the μ sensitivity dependent value, and a phase compensation filter having a function of advancing the phase of the provisional value. A road surface friction coefficient estimating apparatus, wherein the increase / decrease operation amount is determined by passing the provisional value. In the road surface friction coefficient estimating device according to any one of claims 4 to 6,
The friction coefficient increasing / decreasing manipulated variable determining means determines that the increasing / decreasing manipulated variable is determined to be “0” when the deviation is within a predetermined range as a range near “0”. A characteristic road surface friction coefficient estimating device. In the road surface friction coefficient estimating device according to any one of claims 4 to 7,
The friction coefficient increasing / decreasing manipulated variable determining means determines that the increasing / decreasing manipulated variable is assumed to be “0” when the μ sensitivity dependent value is within a predetermined range as a range near “0”. A road surface friction coefficient estimating apparatus characterized by determining.

Images