JP2015225005A - Wheel bearing with sensor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a wheel bearing with a sensor which can suitably divide a region of an input load evaluation value for evaluating a state of an input load, can be switched to an estimation calculation parameter which can calculate a correlation value with an actual input load value, and can improve the estimation accuracy of a load which acts on a wheel.SOLUTION: Sensor units 20 are arranged in a plurality of points at an external periphery of a fixed-side member out of two members an outer member 1 and an inner member 2 of a wheel bearing. There is arranged load estimation means 31 which switches an estimation calculation parameter by a region of an input load evaluation value, and estimates a load from the estimation calculation parameter and sensor outputs of the sensor units 20. Parameter creation means 32 statically evaluates the input load by a correlation value between a load which is estimated by the estimation calculation parameter at each region and an input load value by the regression analysis of the input load value which is given for teaching, and the sensor outputs of the sensor units 20. The region and the estimation calculation parameter of the region are defined on the basis of the evaluation.

Description

  The present invention relates to a wheel bearing having a built-in load sensor for detecting a load applied to a bearing portion of the wheel.

  As a method for detecting the load applied to each wheel, a wheel bearing has been proposed in which a strain gauge is directly attached to the outer ring outer diameter surface and the load is detected from the distortion of the outer ring outer diameter surface (for example, Patent Document 1).

  In the technique disclosed in Patent Document 1, when detecting the load acting on the wheel bearing, the deformation amount of the fixed-side raceway with respect to the load is small, so the amount of deflection is small, and the load cannot be detected with high accuracy. . In order to solve such a problem, a sensor comprising a strain generating member that comes into contact with the outer peripheral surface of a fixed bearing ring of a wheel bearing at two or more contact fixing portions, and a sensor that detects the strain of the strain generating member Wheel bearings have been proposed by providing units (for example, Patent Documents 1 and 2). In these Patent Documents 1 and 2, the following presentation is further made.

In Patent Document 2, the direction of the axial load Fy is determined based on the difference in the amplitude value (vibration component accompanying the revolution motion of the rolling element) in the sensor signals of the sensor units arranged above and below, and depending on whether the axial load Fy is positive or negative , A configuration for calculating a load using a load estimation parameter suitable for each is presented.
In Patent Document 3, there is provided means for switching load calculation processing according to the rotational speed, load estimation calculation using an average value and amplitude is performed in a normal running state, and load using only an average value signal output at a low speed or a stationary state. A configuration for performing the estimation calculation is presented.

Special table 2003-530565 JP 2010-043901 A JP 2010-181154 A

  As in Patent Documents 2 and 3, for a plurality of regions divided according to the input load state, load estimation calculation is performed by linear approximation, depending on the state of the detected sensor signal and the state of the estimated load. By selecting the optimum one from a plurality of load estimation parameters and calculating the estimated load, it is possible to obtain a detection result with a small error in which the nonlinear characteristics of the sensor are corrected.

However, Patent Documents 2 and 3 have the following problems.
(1) It is difficult to divide the region optimally.
(2) Since the nonlinear characteristics of the sensor signal differ depending on the component force in each direction, it is necessary to set a region for each component force.
(3) A method for solving the problems (1) and (2) and determining a series of load estimation parameters in a short time is desired.

The problem in dividing the area will be specifically described.
<Conventional determination threshold value determination method>
About the input load evaluation value
The input load evaluation value is an evaluation value for switching the estimated calculation parameter.
In the above conventional example, the difference in amplitude between the upper and lower sensor output values is obtained and used as the input load evaluation value E for determining the input load state. The input load evaluation value E in this case is a value that monotonously increases with respect to the axial load Fy as shown in FIG.
As shown in FIG. 1 corresponding to the embodiment, when the contact fixing part of the sensor unit 21 is provided in the vicinity of the rolling surface on the double-row outboard side, the output signal varies as shown in FIG. It can be seen. This is a change in distortion due to the passage of rolling elements. When strain sensors are installed on the upper surface (opposite road surface) and the lower surface (road surface), when the axial load Fy, which is the lateral load of the vehicle, is input from the road surface, depending on the magnitude of the axial load Fy, The amplitude value changes. FIG. 21 shows sensor outputs of the upper surface portion and the lower surface portion.

  Since the sensor is installed on the outboard side, when the axial load Fy is in the + direction (the direction entering the vehicle body side), the individual rolling element loads are reduced at the upper surface portion and increased at the lower surface portion. The opposite is true when the axial load Fy is in the-direction (the opposite direction to the vehicle body). Using this, the direction of the axial load Fy is determined as shown in FIG. 22 and the load estimation calculation is performed, so that the accuracy of the output can be improved.

"Division and discrimination threshold"
An example of the sensor output with respect to the input load evaluation value E is shown in FIG. The sensor output exhibits nonlinear characteristics with respect to the input load evaluation value E. For this reason, it is divided into (n + 1) regions and linearly approximated, and the load estimation parameter Mi (i = n + 1) is calculated using the data in each region. At this time, the input load evaluation value when the area is divided is set to Ci (i = 1 to n) as the area discrimination threshold (FIG. 13).

`` Basic formula for load estimation ''
An estimated load value Fcalc (t) is calculated from sensor output data S (t) at a certain time t using the calculated load estimation parameter.
Fcalc (t) = Mi * S (t)
Here, it is assumed that the load estimation parameter Mi to be used is appropriately selected based on the input load evaluation value E obtained from the input load and the discrimination threshold value Ci. Also, the sensor output data S (t) is assumed to be a vector (S1, S2, S3,..., Sm) consisting of m sensor outputs.

"Task"
Conventionally, the criterion for determining the discrimination threshold value Ci is not clear, and each discrimination threshold value is determined at the discretion of the operator in accordance with the state of the sensor output, resulting in poor reproducibility. Therefore, it was difficult to judge whether the created load estimation parameters were optimal.
Further, as shown in FIG. 21, the input load evaluation value E is correlated with the axial load Fy and the estimation accuracy of the axial load Fy is high, but the input load evaluation value E and the longitudinal load Fx and the vertical load Fz are high. There is a tendency that the correlation with is lower than the axial load Fy. Therefore, there is a possibility that it is not the optimum determination threshold value or load estimation parameter for the longitudinal load Fx or the vertical load Fz. Also, if thresholds and load estimation parameters are prepared for each component force, the number of divisions in the region increases, making the adjustment process more complicated and increasing the time required for optimization. To do.

The present invention solves the above-mentioned problem, and it is possible to appropriately divide the input load evaluation value area for evaluating the state of the input load, and to estimate the calculation parameter that can perform a calculation with a high correlation value with the actual input load value. An object of the present invention is to provide a sensor-equipped wheel bearing that can be switched and can improve the estimation accuracy of a load acting on the wheel.
Another object of the present invention is to enable setting of a region and a series of load estimation parameters in a short time for each component force in each direction.

The sensor-equipped wheel bearing according to the present invention includes an outer member 1 in which double-row rolling surfaces 3 are formed on the inner periphery, and an inner member in which rolling surfaces 4 facing the rolling surface 3 are formed on the outer periphery. 2 and a double-row rolling element 5 interposed between the opposing rolling surfaces 3 and 4 of both members 1 and 2, and a wheel bearing for rotatably supporting the wheel with respect to the vehicle body,
A plurality of sensor units 20 are provided on the outer diameter surface of the stationary member of the outer member 1 and the inner member 2, and the sensor unit 20 is fixed in contact with the outer diameter surface. A strain generating member 21 having a contact fixing portion 21a and one or more sensors 22 attached to the strain generating member and detecting the strain of the strain generating member 21;
The following load estimation means 31 and the parameter generation means 32 for generating the input load evaluation value E region used in the load estimation means 31 and the estimated calculation parameter Mi for each region from the teaching data.

  The load estimation means 31 includes an input load evaluation value generation unit 34 that calculates an input load evaluation value E for evaluating a state of an input load of the wheel bearing from a sensor output of each sensor unit 20, and the input load evaluation value. The estimated calculation parameter Mi set for each region of E is switched by the input load evaluation value E calculated by the input load evaluation value generation unit 34. From the switched estimated calculation parameter Mi and the sensor output of the sensor unit 20, A load calculating unit for calculating an estimated load value of an input load acting on the wheel bearing or the ground contact point of the tire;

The parameter generation means 32 is provided as teaching data, and the input load value Fin (t) acting on the wheel bearing during vehicle travel, and the sensor unit when the input load value Fin (t) is applied. Twenty sensor outputs S (t) are stored, and the input load evaluation value E is calculated from the stored sensor output S (t), and the input load evaluation value E is temporarily calculated by regression analysis. For each area, an estimated load value of the input load is obtained from the assumed value of the estimated calculation parameter and the sensor output, and the obtained estimated load value of the input load is statistically calculated with a correlation value with the input load value. Evaluate and calculate the estimated load value of the input load by changing the temporary area and the assumed value of the estimated calculation parameter, and repeat the process of performing the evaluation and use it in the load estimating means 31 And a parameter calculating part 42 defining with said estimation calculation parameters Mi of each region and each of these respective regions of the input load evaluation value E.
The input load evaluation value E is a value for evaluating a predetermined physical quantity that affects the load state of the wheel bearing, that is, the detected value of the load acting on the wheel bearing. The speed at which the bearing rotates), the difference between the amplitude values of the upper and lower sensor units 20, or the difference between the amplitude values of the left and right sensor units 20.

The operation of the above configuration will be described. When a load acts between the tire of the wheel and the road surface, the load is applied to the stationary member (for example, the outer member) of the wheel bearing, that is, input, and deformation occurs. Here, since the two or more contact fixing portions 21 a of the strain generating member 21 in the sensor unit 20 are fixed in contact with a fixed side member such as the outer member 1, the distortion of the fixed side member 1 is applied to the strain generating member 21. The distortion is easily transmitted and the distortion is detected with high sensitivity by the sensor.
The detected sensor output varies depending on the state of the load acting on the wheel bearing. If the state of the load is appropriately evaluated and used for the estimation calculation, the load estimation accuracy is improved.

According to this configuration, the load estimating means 31 determines the estimation calculation parameter Mi for each region of the input load evaluation value E that is an evaluation value of the load state acting on the wheel bearing, and uses the input load evaluation value E as a result. The estimated calculation parameter Mi for the corresponding region obtained by the input load value evaluation value generation unit 34 is selected. Using the estimated calculation parameter Mi selected in this way, the input load acting on the wheel bearing or the tire contact point is estimated from the sensor output output from each sensor unit 20 according to a predetermined calculation rule. Therefore, a detection result with a small error in which the nonlinear characteristic of the sensor with respect to the input load evaluation value E of the sensor unit 20 is corrected can be obtained.
In this case, the parameter generation means 32 performs linear approximation by multivariate regression analysis, and statistically evaluates the accuracy of the approximation with the correlation value between the load estimation output and the teaching data, and obtains the optimum value. Therefore, a series of operations can be performed automatically and systematically. That is, the parameter generation means 32 can stably and appropriately obtain a series of estimated calculation parameters in a short time, and the load estimation accuracy in each region is further improved.
In this way, the criteria for determining the threshold value for switching the region is clarified, and the region is determined by a predetermined method, so the reproducibility is excellent, the created estimation calculation parameter is optimal, and accurate load estimation is possible. Yes.

In the present invention, the parameter calculation unit 42 switches the region for determining the estimated calculation parameter Mi, and the threshold value is obtained by calculating the correlation value between the estimated load value and the input load value by the parameter calculation unit 42 in each region. It may be requested to maximize.
In this way, when the correlation value between the actual load value and the estimated load value is the highest, by setting the discrimination threshold for switching the load estimation parameter, the load estimation accuracy in each region is higher than before. improves.

In this case, a specific example of the calculation of the area and the estimated calculation parameter by the parameter generation unit 32 will be described.
(1) First, the input load value during traveling and the collected data of the sensor outputs of the plurality of sensor units 20 provided on the bearing are prepared. The input load value is measured using, for example, a dedicated six component force measuring wheel, and the sensor output of the sensor unit 20 measured at the same time is used as sensor data. The input load value and sensor output prepared in this way are stored in the teaching data storage unit 41. (2) The parameter calculation unit 42 uses the teaching data recorded in the teaching data storage unit 41 to determine the input load evaluation value E for determining the load region according to the state of the input load and the region. Is determined as follows. The initial value of the discrimination threshold value Ci is calculated from the teaching data according to the rules determined by the parameter calculation unit 42, or is set to a predetermined value.
(3) Using one load area divided by the discrimination threshold as the target area, extract the input load value E data and sensor output data, and perform a linear approximation calculation (regression analysis process) between these data. Thus, the load estimation parameter Mi of the target area is calculated.
(4) The estimated load value is calculated from the calculated load estimation parameter M and the sensor output for the data in the target region, and the error between the input load value and the estimated load value is evaluated. The error is evaluated using, for example, a correlation value between both load values.
(5) While gradually changing the discrimination threshold in the direction of expanding the target area, the processing described in (3) to (4) above was repeated and the error evaluation value was obtained and determined. When the range survey is completed, the obtained error evaluation value is analyzed. As a result, the position where the error is smallest (the position where the error is maximum in the case of the correlation value) is determined as the optimum determination threshold value, and the determination threshold value and the load estimation parameter at that time are determined and stored.
(6) Move to the next area and sequentially perform the processing described in the above (3) to (5) to divide the load area at the optimum position. Then, a series of threshold values indicating the optimum area division position and a series of load estimation parameters obtained from the load areas divided in such a manner are obtained.

  In this invention, the load estimating means 31 estimates each component force in at least two directions among the three directions orthogonal to the vertical direction, the axial direction, and the front-rear direction as the estimated load value of the input load. A threshold value that divides the region for switching calculation parameters may be provided independently for each component force. In this way, the load estimation accuracy is improved by setting an optimum value for each component force as the determination threshold value for switching the load estimation parameter. The “front-rear direction” is a direction orthogonal to the up-down direction and the axial direction, and is a driving direction of the vehicle when going straight, that is, a bearing radial direction that is the front-rear direction of the vehicle.

In the present invention, the load estimating means 31 may use the rotational speed of the wheel bearing as the input load evaluation value E.
In a wheel bearing that is a rolling bearing, the output of the sensor 22 fluctuates as a rolling element passes in the vicinity of the sensor 22. Therefore, the rotational speed of the wheel bearing affects the estimated load value of the load. In this regard, a region can be defined by using the rotational speed of the wheel bearing as the input load evaluation value E, and the estimation calculation parameter Mi corresponding to the region can be used to accurately estimate the load.
In this case, the input load evaluation value generation unit 34 of the load estimation unit 31 obtains the rotation speed of the wheel bearing from the speed of fluctuation of the sensor output that fluctuates when a rolling element passes in the vicinity of the sensor 22. Input load evaluation value E.

In the present invention, the sensor unit 20 may be provided above and below the wheel bearing, and a difference between amplitude values of the upper and lower sensor units may be used as the input load evaluation value E.
The difference between the amplitude values of the upper and lower sensor units 20 has a large influence on the estimated load value of the load in each direction. For this reason, by defining a region using the difference between the amplitude values of the upper and lower sensor units 20 as the input load evaluation value E and using the estimation calculation parameter Mi corresponding to the region, the load can be estimated with high accuracy.

In the present invention, the sensor unit 20 may be provided before and after the wheel bearing, and the difference between the amplitude values of the front and rear sensor units 20 may be used as the input load evaluation value E.
The difference between the amplitude values of the front and rear sensor units 20 has a great influence on the estimated load value of the load in each direction. For this reason, by defining a region using the difference between the amplitude values of the front and rear sensor units 20 as the input load evaluation value E and using the estimation calculation parameter Mi corresponding to the region, the load can be estimated with high accuracy.

In the present invention, three or more sensor units 20 may be provided, and the load estimating unit 31 may calculate loads in three directions, the front-rear direction, the axial direction, and the up-down direction. For example, the sensor units 20 are equally arranged at intervals of 90 degrees in four places, front and rear.
Since the output of the sensor unit 20 is affected by the load in each direction, if the sensor units 20 are provided at three or more locations, the loads in the three axes orthogonal to the front-rear direction, the axial direction, and the vertical direction can be calculated. . Also in this case, it is possible to calculate a load with high accuracy by dividing the region by the input load evaluation value E as described above and setting the corresponding estimation calculation parameter Mi.
In the case where the sensor units 20 are equally distributed at 90 ° intervals in four places, front and rear, the load in each direction can be calculated with higher accuracy. That is, the vertical load Fz and the axial load Fy can be estimated from the output signals of the two sensor units 20 arranged on the upper surface portion and the lower surface portion on the outer diameter surface of the fixed side member, and the right side on the outer diameter surface of the fixed side member. The longitudinal load Fx due to the driving force or the braking force can be estimated from the output signals of the two sensor units 20 arranged on the surface portion and the left surface portion.

  In the present invention, the load estimating means 31 calculates the average value and the amplitude value of the sensor outputs using the sensor outputs of the sensor units 20 within a predetermined time, and calculates the average value, the amplitude value, or the average value. The input load value may be estimated by a calculation process using both the amplitude value and the amplitude value. Thus, the desired input load can be estimated.

In the present invention, a temperature sensor 29 may be attached to the sensor unit 20 and temperature correction means 30 may be provided for correcting an average value of the sensor output based on a detection signal of the temperature sensor 29.
If the temperature of the wheel bearing changes due to heat generated by the rotation of the bearing or the surrounding environment, the sensor output of the sensor unit 20 fluctuates due to thermal expansion or the like even if the load does not change. Remain. Therefore, if the temperature correction means 30 for correcting the average value of the sensor output according to the temperature of the wheel bearing or the surrounding temperature is provided, the detection load error due to the temperature can be reduced.

In the present invention, each sensor unit 20 has three or more contact fixing portions 21a, each sensor unit 20 is provided with at least two strain sensors as the sensors 22, and the phase difference between the output signals of the sensors 21 is The interval between the contact fixing portions 21a is set to be n + 1/2 times the pitch of the rolling elements (n = 0, 1, 2,...), And the average value is determined by the sum of the output signals of two or more sensors 22. May be requested.
In the case of this configuration, the output signals of the two sensors 22 have a phase difference of approximately 180 degrees, and the average value is a value obtained by canceling the fluctuation component due to the rolling element passage. In addition, the amplitude value is stable because it is hardly affected by temperature, and the detection accuracy is improved because two signals are used. Thereby, the average value of the output signals of the two sensors 22 becomes a value obtained by canceling the fluctuation component due to passing through the rolling elements, and the amplitude value becomes more stable and accurate.

  The sensor-equipped wheel bearing according to the present invention includes an outer member in which double-row rolling surfaces are formed on the inner periphery, an inner member in which a rolling surface facing the rolling surface is formed on the outer periphery, and both members. In a wheel bearing for supporting a wheel rotatably with respect to a vehicle body, a fixed-side member of the outer member and the inner member is provided. A plurality of sensor units are provided on the outer diameter surface, and the sensor unit has two or more contact fixing portions fixed in contact with the outer diameter surface, and the strain generating member is attached to the strain generating member. An input load evaluation value generator for calculating an input load evaluation value for evaluating a state of an input load of the wheel bearing from a sensor output of each sensor unit; Set for each area of the input load evaluation value The estimated calculation parameter is switched according to the input load evaluation value calculated by the input load evaluation value generator, and the wheel bearing or the tire contact point is acted on from the switched estimated calculation parameter and the sensor output of the sensor unit. A load estimation unit having a load calculation unit that calculates an estimated load value of an input load; and a parameter generation unit that generates the estimation calculation parameter for each region and the region from teaching data. A teaching data storage unit for storing an input load value applied to the wheel bearing during traveling of the vehicle, and a sensor output of the sensor unit when the input load value is applied, which is given as teaching data; The input load evaluation value is calculated from the sensor output, and the input load evaluation value is calculated by regression analysis. For each area, an estimated load value of the input load is obtained from the assumed value of the estimated calculation parameter and the sensor output, and the obtained estimated load value of the input load is statistically calculated with a correlation value with the input load value. The input load evaluation value used in the load estimating means is evaluated, the estimated value of the input load is obtained by changing the temporary region and the estimated value of the estimation calculation parameter, and the process of performing the evaluation is repeated. And a parameter calculation unit that determines the estimated calculation parameters for each of these areas, the area of the input load evaluation value can be appropriately divided, and calculation with a high correlation value with the actual input load value can be performed. It is possible to switch to an estimated calculation parameter that can be performed, and it is possible to improve the estimation accuracy of the load acting on the wheel.

It is a figure showing combining the sectional view of the wheel bearing with a sensor concerning one embodiment of this invention, and the block diagram of the conceptual composition of the detection system. It is a front view of the outward member of the wheel bearing with a sensor. It is an enlarged plan view of a sensor unit in the wheel bearing with sensor. FIG. 4 is a sectional view taken along line IV-IV in FIG. 3. It is sectional drawing which shows the modification of a sensor unit with a part of outer member. It is a front view of the outward member which made the sensor unit different from the example of FIGS. 1-4 in the wheel bearing with a sensor. It is an enlarged plan view of the sensor unit. FIG. 8 is a sectional view taken along line VII-VII in FIG. 7. It is sectional drawing which shows the modification of a sensor unit with a part of outer member. It is explanatory drawing of the influence of a rolling-element position with respect to the output signal of a sensor unit. It is a flowchart of the process of the parameter production | generation means in the wheel bearing with a sensor. It is a graph which shows the sensor output with respect to the input load evaluation value E in the wheel bearing with a sensor. It is an image figure of the linear approximation of the graph of FIG. It is explanatory drawing of the concept which extracts the data for (DELTA) C from the same graph. It is explanatory drawing of the concept which extracts the data for 2x (DELTA) C from the same graph. It is an image figure which determines the threshold value Ci from the correlation value Q. Schematic diagram of threshold Ci data extraction It is a schematic diagram of data numerical value restrictions. It is a graph which shows the relationship between the difference of the amplitude value of the sensor signal of an upper and lower sensor unit, and an input load evaluation value. It is a graph which shows the change of the sensor signal of a prior art example. It is a graph which shows the input load evaluation value of a prior art example. It is a block diagram which shows the conceptual structure of the process of a prior art example.

  An embodiment of the present invention will be described with reference to FIGS. This embodiment is a third generation inner ring rotating type and is applied to a wheel bearing for driving wheel support. In this specification, the side closer to the outer side in the vehicle width direction of the vehicle when attached to the vehicle is referred to as the outboard side, and the side closer to the center of the vehicle is referred to as the inboard side.

  As shown in the sectional view of FIG. 1, the wheel bearing in the sensor-equipped wheel bearing is opposed to the outer member 1 in which a double row rolling surface 3 is formed on the inner periphery, and the respective rolling surfaces 3. The inner member 2 has a rolling surface 4 formed on the outer periphery, and the outer member 1 and the double row rolling elements 5 interposed between the rolling surfaces 3 and 4 of the inner member 2. This wheel bearing is a double-row angular ball bearing type, and the rolling elements 5 are made of balls and are held by a cage 6 for each row. The rolling surfaces 3 and 4 have an arc shape in cross section, and are formed so that the ball contact angle is aligned with the back surface. Both ends of the bearing space between the outer member 1 and the inner member 2 are sealed by a pair of seals 7 and 8, respectively.

The outer member 1 is a fixed side member, and has a vehicle body mounting flange 1a attached to a knuckle 16 in a suspension device (not shown) of the vehicle body on the outer periphery, and the whole is an integral part. Bolt holes 14 for mounting the vehicle body are provided at a plurality of locations in the circumferential direction on the flange 1a, and knuckle bolts 18 inserted into the bolt insertion holes 17 of the knuckle 16 from the inboard side are screwed into the bolt holes 14. The vehicle body mounting flange 1 a is attached to the knuckle 16.
The inner member 2 is a rotating side member, and includes a hub wheel 9 having a hub flange 9a for wheel mounting, and an inner ring 10 fitted to the outer periphery of the end portion on the inboard side of the shaft portion 9b of the hub wheel 9. And become. The hub wheel 9 and the inner ring 10 are formed with the rolling surfaces 4 of the respective rows. An inner ring fitting surface 12 having a small diameter with a step is provided on the outer periphery of the inboard side end of the hub wheel 9, and the inner ring 10 is fitted to the inner ring fitting surface 12. A through hole 11 is provided at the center of the hub wheel 9. The hub flange 9a is provided with press-fitting holes 15 for hub bolts (not shown) at a plurality of locations in the circumferential direction. In the vicinity of the base portion of the hub flange 9a of the hub wheel 9, a cylindrical pilot portion 13 for guiding a wheel and a braking component (not shown) protrudes toward the outboard side.

  FIG. 2 shows a front view of the outer member 1 of the wheel bearing as viewed from the outboard side. 1 shows a cross-sectional view taken along the line II in FIG. As shown in FIG. 2, the vehicle body mounting flange 1 a is a projecting piece 1 aa in which a circumferential portion provided with each bolt hole 14 protrudes to the outer diameter side from the other portion.

  On the outer diameter surface of the outer member 1 that is a fixed side member, the sensor unit 20 has an upper surface portion, a lower surface portion, a right surface portion, and a left surface of the outer diameter surface that are in a vertical position and a left and right position with respect to the tire ground contact surface. Four equally spaced phase differences of 90 degrees in the circumferential direction are provided on the surface portion. Specifically, one sensor unit 20 is disposed at the center of the upper surface portion of the outer member 1 between the two adjacent projecting pieces 1aa, and the lower surface of the outer member 1 on the outer surface. The other one sensor unit 20 is arranged at the center between two adjacent projecting pieces 1aa, and at the center between the two upper and lower projecting pieces 1aa on the right and left surface parts. A sensor unit 20 is arranged. Note that four sensor units 20 are not necessarily provided, but three or more sensor units 20 are preferably provided.

  As shown in FIGS. 3 and 4 in an enlarged plan view and an enlarged cross-sectional view, these sensor units 20 are a strain generating member 21 and a sensor that is attached to the strain generating member 21 and detects the strain of the strain generating member 21. 22 The strain generating member 21 is made of an elastically deformable metal thin plate material of 2 mm or less such as a steel material, has a planar shape in a strip shape, and has notches 21b on both sides of the center. Further, the strain generating member 21 has two contact fixing portions 21 a that are fixed to the outer diameter surface of the outer member 1 through spacers 23 at both ends. The sensor 22 is affixed to a location where the strain increases with respect to the load in each direction on the strain generating member 21. Here, as the location, the central portion sandwiched between the notch portions 21b on both sides on the outer surface side of the strain generating member 21 is selected, and the sensor 22 detects the strain in the circumferential direction around the notch portion 21b. . Note that the strain generating member 21 is plastically deformed even in a state in which an assumed maximum force is applied as an external force acting on the outer member 1 that is a fixed member or an acting force acting between the tire and the road surface. It is desirable not to do so. This is because when the plastic deformation occurs, the deformation of the outer member 1 is not transmitted to the sensor unit 20 and affects the measurement of strain. The assumed maximum force is, for example, the maximum force in a range where the function as a wheel bearing is not impaired by the application of the force.

  In the sensor unit 20, the two contact fixing portions 21a of the strain generating member 21 are located at the same dimension in the axial direction of the outer member 1, and the two contact fixing portions 21a are separated from each other in the circumferential direction. These contact fixing portions 21a are fixed to the outer diameter surface of the outer member 1 by bolts 24 via spacers 23, respectively. Each bolt 24 is inserted into a bolt insertion hole 26 of the spacer 23 from a bolt insertion hole 25 pierced in the radial direction provided in the contact fixing portion 21 a, and is a female screw bolt provided on the outer peripheral portion of the outer member 1. Screwed into the hole 27. In this way, by fixing the contact fixing portion 21a to the outer diameter surface of the outer member 1 via the spacer 23, the central portion having the notch portion 21b in the strain generating member 21 having a thin plate shape is the outer member 1. It becomes a state away from the outer diameter surface of this, and distortion deformation around the notch 21b becomes easy. Instead of providing the spacer 23, a groove-shaped notch 1c for increasing strain along the axial direction may be provided on the outer diameter surface of the outer member 1 as in the example of FIG.

  As the axial position where the contact fixing portion 21a is disposed, an axial position that is the periphery of the rolling surface 3 (FIG. 1) of the outboard side row of the outer member 1 is selected here. Here, the periphery of the rolling surface 3 of the outboard side row is a range from the intermediate position of the rolling surface 3 of the inboard side row and the outboard side row to the formation portion of the rolling surface 3 of the outboard side row. It is. In order to stably fix the sensor unit 20 to the outer diameter surface of the outer member 1, a flat portion 1 b is formed at a location where the spacer 23 is contacted and fixed on the outer diameter surface of the outer member 1.

  Various sensors can be used as the sensor 22. For example, the sensor 22 can be composed of a metal foil strain gauge. In that case, the distortion generating member 21 is usually fixed by adhesion. The sensor 22 can also be formed on the strain generating member 21 with a thick film resistor.

  The sensor unit 20 is provided with a temperature sensor 29 (FIG. 3) in addition to the sensor 22 which is a strain sensor, and the temperature correction means 30 (FIG. 1) can correct the output of the strain sensor 22 according to the observed temperature. preferable. The temperature correction means 30 is a means for correcting the output signal of the sensor 22 to a value at a constant temperature such as 20 ° C., for example. This temperature correction can reduce the detected load error. As the temperature sensor 29, for example, a thermistor or a platinum resistance element can be used. The temperature sensor 29 may be provided in the vicinity of the installation portion of each sensor unit 20 on the outer diameter surface of the outer member 1 instead of being mounted on the sensor unit 20.

  6 to 10 show modified examples of the sensor unit 20. In the sensor unit 20 of this example, the strain generating member 21 is fixed to the outer diameter surface of the outer member 1 by the three contact fixing portions 21a, and one strain signal is obtained by averaging the outputs a and b of the two strain sensors 22. Is output. The distortion generating member 21 is provided with notches 21b similar to the above on two sides in the longitudinal direction, and a sensor 22 is attached to a portion narrowed by the notches 21b. Also in this example, instead of providing the spacer 23, a groove-shaped notch 1c for increasing strain along the axial direction may be provided on the outer diameter surface of the outer member 1 as in the example of FIG.

  As shown in FIG. 10, in the sensor unit 20 having these three contact fixing portions 21 a, the signals of the two strain sensors 21 are obtained by arranging the strain sensors 22 at intervals of ½ pitch of the pitch P of the rolling elements 5. Is shifted by 1/2 pitch. For this reason, since the amplitude component by the rolling element 5 is canceled, it becomes possible to calculate | require an average value more correctly. Therefore, there is an effect that the estimation accuracy of the drift amount when the drift occurs increases and the correction is stabilized.

  A means for estimating the load from the sensor output of the sensor unit 20 will be described with reference to FIG. This wheel bearing with sensor is a wheel bearing provided with the sensor unit 20 having the above-described configuration, and the following load estimation means 31, the region of the input load evaluation value E used in the load estimation device 31, and the estimation for each region. Parameter generating means 32 for generating calculation parameters Mi from teaching data.

  The load estimating means 31 and the parameter generating means 32 are configured by an electronic circuit including, for example, a mylocomputer, and are installed on the outer member 1. The load estimating means 31 and the parameter generating means 32 may be installed in a vehicle ECU (electronic control unit) or the like at a position away from the wheel bearing, and only the parameter generating means 32 is separated from the wheel bearing. For example, the load estimating means 31 and the parameter generating means 32 may be separated from each other.

  The load estimator 31 is switched to an input load evaluation value generator 34 for obtaining the input load evaluation value E, and an estimation calculation parameter Mi for each region corresponding to the obtained input load evaluation value E. And a load calculation unit 35 for calculating the estimated load value. In addition to this, the load estimation means 31 includes a parameter storage unit 33 and a preprocessing unit 36.

  The parameter storage unit 33 is means for storing an estimated calculation parameter Mi for each region in which the input load evaluation value E is set. Although only one parameter storage unit 33 is shown in FIG. 1, three parameter storage units 33 corresponding to the loads Fx, Fy, and Fz in the respective directions are provided.

  For example, the pre-processing unit 36 calculates the load in each axial direction to be estimated, that is, the longitudinal load Fx, the axial load Fy, and the vertical load Fz from the sensor output output from the sensor 22 of each sensor unit 20. This is a means for calculating the sensor output S (t) for each of the axial directions X, Y, and Z processed as described above. The sensor outputs S (t) in the respective axial directions X, Y, and Z to be calculated are values processed so as to become an estimated load value by multiplying the estimated calculation parameter Mi set for each axial direction. That is, the preprocessing unit 36 extracts an average value and an amplitude value from the sensor output, and generates a sensor output vector S (t).

The input load evaluation value generation unit 34 is a means for calculating an input load evaluation value E for evaluating the state of the input load of the wheel bearing from the sensor output of each sensor unit 20.
When the input load evaluation value E is the wheel speed, the input load evaluation value generation unit 34 obtains the bearing rotation speed from the fluctuation of the sensor output of the sensor unit 20 due to the passage of the rolling element 5 of the wheel bearing, for example. The value is set as an input load evaluation value E.
When the input load evaluation value E is the difference between the amplitude values of the upper and lower sensor units 20 or the difference between the amplitude values of the left and right sensor units 20, the input load evaluation value generation unit 34 obtains the difference from the sensor output. The input load evaluation value. In this case, the difference may be obtained directly from the sensor 22 of the sensor unit 20 or may be obtained via the preprocessing unit 36.

The load calculation unit 35 uses the estimated calculation parameter Mi set for each area of the input load evaluation value E in the parameter storage 33, and switches according to the input load evaluation value E calculated by the input load evaluation value generation unit 34. This is a means for calculating the estimated load value of the input load acting on the wheel bearing or the tire contact point from the estimated calculation parameter Mi and the sensor output of the sensor unit 20.
The load calculation unit 35 calculates the estimated load value Fcalc (t) of the input load, and calculates the estimated calculation parameter Mi (i = 1, 2, 2) for each set region of the input load evaluation value E that evaluates the load state of the wheel bearing. …) Using the following basic formula for load estimation:
Fcalc (t) = Mi * S (t)
To do. S (t) is a sensor output at a certain time t. The sensor output S (t) is a value obtained by subjecting the output of the sensor 22 of each sensor unit 20 to a predetermined process by the preprocessing unit as described above.

The parameter generation unit 41 includes a teaching data storage 41 and a parameter calculation unit 42.
The teaching data storage 41 is given as teaching data. The input load value Fin (t) acting on the wheel bearing during vehicle travel, and the sensor unit 20 when the input load value Fin (t) is applied. The sensor output S (t) is stored.
The parameter calculation unit 42 divides the region from the teaching data stored in the teaching data storage 41, generates the estimated calculation parameter Mi for each region, and stores it in the parameter storage unit 33 in the load estimation unit 31. It is a means of recording.

The parameter calculation unit 42 calculates the input load evaluation value E from the sensor output S (t) stored in the teaching data storage unit 41, and performs the estimation for each temporary region of the input load evaluation value E by regression analysis. The estimated load value of the input load is obtained from the assumed value of the calculation parameter E and the sensor output, and the estimated load value of the obtained input load is statistically evaluated with a correlation value with the input load value. Each region of the input load evaluation value E used in the load estimation means 31 and these are repeatedly obtained by changing the region and the assumed value of the estimation calculation parameter to obtain an estimated load value of the input load and repeating the evaluation. The estimation calculation parameter Mi for each region is determined.
More specifically, the parameter calculation unit 42 performs the above-described processing in the steps (1) to (6) in [Means for Solving the Problem].

More specifically, the parameter calculation unit 42 calculates according to the flowchart of FIG. By performing the processing of the flow in the figure, it is possible to automatically calculate the region division based on the optimum discrimination threshold and the load estimation parameter Mi with good reproducibility.
(1) First, the collected data of the input load value Fin (t) and the sensor output S (t) during the traveling is facilitated by the test traveling, and this collected data is stored in the teaching data storage unit 41 (step S1).

(2) A data upper limit value Cmax (initial value of the discrimination threshold) is set (step S3).
In this case, the input load evaluation value E is calculated from the data of all sensor outputs stored in the teaching data storage unit 41, and the upper limit value and the lower limit value of the input load evaluation value E that can be obtained from all the sensor outputs are respectively calculated. Let Emax and Emin.
The input load evaluation value E is any of the wheel speed, the difference between the amplitude values of the upper and lower sensor units 20, and the difference between the amplitude values of the left and right sensor units 20.
Assuming that the input load evaluation value E and the sensor output have the relationship shown in FIG. 12, for example, the upper limit value and lower limit value that can be taken by the input load evaluation value E in FIG.
(3) The calculation is performed as described above, and the data upper limit value Cmax is set as Cmax = Emax as the initial value of the discrimination threshold C.
(4) Before step S3, the counter i is set to i = 1 (step S2).

(5) A point that is decreased from the initial value Cmax of the discrimination threshold C by the set value ΔC is set as Cp1 as shown in FIG. 14 (step S4). Data of the input load Fin (t) and sensor output S (t) included in the region between the initial value Cmax and the decrease point Cp1 is extracted from the teaching data storage unit 41 (step S5).
(6) Using the extracted data, the load estimation parameter M (Cp1) for the corresponding region is calculated (step S7).

The process of calculating the load estimation parameter M (Cp1) using the extracted data at this time is performed as follows, for example.

Input load value Fin (t) is a 3x1 matrix.
The load estimation calculation parameter Mi is a 3 × N matrix.
The sensor output S (t) is an N × 1 matrix.
* At this time, N is the number of average or amplitude values.

The estimated load value Fcalc (t) of the extracted data portion is calculated using the load estimation calculation parameter M (Cp1) calculated in this way (step S8).
However, if the number of extracted data is less than the set number N (N is arbitrarily set), the process returns to step S4, and further decreases by ΔC to widen the corresponding area (step S4), and data extraction of the corresponding area is performed. (Step S5) is repeated. The number of data is counted as one set of the input load value Fin (t) and the output S (t) of each sensor unit 22 at each time t1, t2,.

(7) A correlation value Q (Cp) between the input load value Fin (t) during travel and the estimated load value Fcalc (t) obtained in step S8 is calculated (step S9). This correlation value is set as an input load evaluation value Q (Cp) for evaluating the quality of the load estimation parameter Mi.

A normalized value may be used as follows.

(8) The reduction point Cp is compared with the lower limit value Emin of the input load evaluation value E (step S10). If the reduction point Cp is larger than the lower limit value Emin, the process returns to step S4 and is further decreased by the set value ΔC. The decrease point Cp2 is set as the decrease point Cp2, and the data of the input load Fin (t) and the sensor output S (t) between the data upper limit Cmax (initial value of the discrimination threshold) and the decrease point Cp2 are extracted (FIG. 15). ), An input load evaluation value Q (Cp2) is calculated.
(9) The process of the above (8), that is, the processes of steps S4 to S10 are repeated, and a series of input load evaluation values Q (Cpn) is calculated until Cpn ≦ Emin (n = 1, 2,...).

  (10) A decrease point Cpi when the obtained series of input load evaluation values Q (Cpn) becomes the highest is obtained as shown in FIG. 16, and the decrease point Cpn is determined as a discrimination threshold value Ci (first region). In this case, i = 1) is determined (step S11). In this way, the determination threshold value Ci is determined, and the estimated calculation parameter Mi is determined (step S12).

(11) It is compared whether or not the remaining number of data is less than or equal to the set number (N) (step S13), and i is increased by 1 until the set number (N) or less (step S14). The process returns to step S3.
(12) When determining the determination threshold value C2 or later, the determination threshold value (Ci-1) obtained immediately before is set to Cmax (steps S14 and S3), and the processing after the above (5), that is, step The procedure from S3 to S14 is repeated (FIG. 17).
(13) When the number of extracted data cannot satisfy the set number N (step S13), the series of operations is terminated.

When the number of data is extremely small, a large number of threshold values or load estimation parameters are created, and there is a risk that the load estimation accuracy decreases near the switching of the load estimation parameters. In order to prevent this, a threshold value or a load estimation parameter is not created unless there is a certain number of data (N) above a certain value (FIG. 18).
That is, the condition that the process (6) is advanced is that the number of extracted data exceeds the set number N. If this condition is not satisfied, the process returns to (5), and a process of expanding the corresponding area by reducing by ΔC is performed. Even when the area is expanded, the determination threshold value determination is terminated when the set number N is not exceeded.

  In the above procedure, linear approximation was performed by multivariate regression analysis, and the accuracy of the approximation was statistically evaluated with the correlation value between the load estimation output and the teaching data (input data), and the optimum value was determined. A series of operations can now be performed automatically and systematically.

The effects of this embodiment are summarized below.
In the method used in this embodiment, the discrimination threshold value Ci for switching the load estimation parameter Mi can be set where the correlation value between the actual input load value and the estimated load value is the highest. Load estimation accuracy is improved more than ever.
The load estimation accuracy is improved by setting an optimum value for each of the component forces (the loads Fx, Fy, and Fz in each direction) as the determination threshold value Ci for switching the load estimation parameter Mi.
The load estimation parameter Mi and the determination threshold value Ci for switching the region can be calculated in a short time, and a stable load estimation parameter Mi can be obtained.

DESCRIPTION OF SYMBOLS 1 ... Outer member 2 ... Inner member 3, 4 ... Rolling surface 5 ... Rolling body 20 ... Sensor unit 21 ... Strain generation member 22 ... Strain sensor 31 ... Load estimation means 32 ... Parameter generation means 33 ... Parameter storage part 34 ... Input load evaluation value generation unit 35 ... Load calculation unit 41 ... Teaching data storage unit 42 ... Parameter calculation unit

Claims (9)

An outer member having a double-row rolling surface formed on the inner periphery, an inner member having a rolling surface facing the rolling surface formed on the outer periphery, and interposed between the opposing rolling surfaces of both members. In a wheel bearing comprising a double row rolling element, and rotatably supporting the wheel with respect to the vehicle body,
A plurality of sensor units are provided on the outer diameter surface of the fixed side member of the outer member and the inner member, and the sensor unit has two or more contact fixing portions fixed in contact with the outer diameter surface. A strain generating member having one or more sensors attached to the strain generating member and detecting the strain of the strain generating member;
An input load evaluation value generator for calculating an input load evaluation value for evaluating the state of the input load of the wheel bearing from the sensor output of each sensor unit, and an estimated calculation parameter set for each area of the input load evaluation value Is switched according to the input load evaluation value calculated by the input load evaluation value generation unit, and the input load acting on the wheel bearing or tire contact point is estimated from the switched estimation calculation parameter and the sensor output of the sensor unit. Load estimation means having a load calculation unit for calculating a load value;
Parameter generating means for generating the region and the estimated calculation parameter for each region from teaching data;
The parameter generation means is a teaching data storage for storing an input load value acting on the wheel bearing during traveling of the vehicle, and a sensor output of the sensor unit when the input load value is applied, which is given as teaching data. The input load evaluation value is calculated from the stored sensor output, and the input load is calculated from the assumed value of the estimated calculation parameter and the sensor output for each temporary region of the input load evaluation value by regression analysis. The estimated load value of the obtained input load is statistically evaluated with a correlation value with the input load value, and the temporary region and the assumed value of the estimated calculation parameter are changed to change the estimated load value. Obtain the estimated load value of the input load and repeat the process of evaluating it, and each region of the input load evaluation value used in the load estimating means and the estimation calculation parameter for each of these regions And a parameter calculating unit for determining an over data,
The wheel bearing with a sensor characterized by the above-mentioned.   2. The wheel bearing with sensor according to claim 1, wherein the parameter calculation unit is configured such that a threshold value for switching the region for determining the estimation calculation parameter is an estimated load value and the input load by the parameter calculation unit in each region. Bearing for sensor wheel that seeks to maximize the correlation value.   3. The wheel bearing with sensor according to claim 1, wherein the load estimating means has at least one of an orthogonal load, an axial direction, and an orthogonal three-axis direction in the front-rear direction as the estimated load value of the input load. A sensor-equipped wheel bearing provided with thresholds for estimating each component force in two directions and dividing the region for switching the estimated calculation parameter independently for each component force.   4. The sensor-equipped wheel bearing according to claim 1, wherein the load estimating means uses the rotation speed of the wheel bearing as the input load evaluation value. 5.   The sensor-equipped wheel bearing according to any one of claims 1 to 3, wherein the sensor units are provided above and below the wheel bearing, and a difference between amplitude values of sensor outputs of the upper and lower sensor units. A sensor-equipped wheel bearing with the input load evaluation value.   The sensor-equipped wheel bearing according to any one of claims 1 to 3, wherein the sensor unit is provided before and after the wheel bearing in a vehicle traveling direction, and sensor outputs of the front and rear sensor units. A sensor-equipped wheel bearing having a difference in amplitude value as the input load evaluation value.   The wheel bearing with sensor according to any one of claims 1 to 6, comprising three or more sensor units, wherein the load estimating means is orthogonal to the vehicle longitudinal direction, the axial direction, and the vertical direction. A sensor-equipped wheel bearing that calculates the load in three axes.   9. The wheel bearing with sensor according to claim 1, wherein the load estimating means uses an average value and an amplitude of the sensor outputs using sensor outputs of the sensor units within a predetermined time. A sensor-equipped wheel bearing that calculates a value and estimates the input load value by an arithmetic process using an average value, an amplitude value, or both an average value and an amplitude value.   The wheel bearing with a sensor according to any one of claims 1 to 8, wherein each sensor unit has three or more contact fixing portions, and each sensor unit has at least two strains as the sensor. Sensors are provided, and the interval between the contact fixing portions is set so that the phase difference between the output signals of the sensors is n + 1/2 times the pitch of the rolling elements (n = 0, 1, 2,...). A wheel bearing with sensor for obtaining an average value from the sum of the output signals of the above sensors.

Images