HUT78137A - Driving stability control system - Google Patents

Abstract

In a procedure for controlling vehicle stability, the input variables of reference speed and steering angle determined by the required travel curve are converted into a set point for the yaw speed using a mathematical model of the car. This is compared with the actual value and a controller responds to the difference using an additional yaw torque to bring this difference to zero. The car model has a calculation circuit (350) for the set point of yaw speed using the rules for steady arc travel and the measured steering angle is fed in via a low-pass filter (351) of the first or second order. Observation of the long-term vehicle stance allows two correction factors to be applied to the filter.

Description

The present invention generally relates to a system for controlling travel stability. More particularly, the present invention relates to a system for controlling the deflection torque of a four-wheeled vehicle in cornering by means of a device for detecting the current coefficient of friction.

Introducing the general structure of driving stability control.

1. General structure of driving stability control

The term stability control concept encompasses four principles for influencing the behavior of a vehicle with predetermined pressures on each wheel brake and interventions in the drive motor control.

These principles are: braking slip control (anti-lock braking, ABS), which prevents individual wheels from locking during the braking process; drive slip control (ASR), which prevents the drive wheels from spinning; electronic brake force distribution (EBV), which controls the proportion of brake forces between the front and rear axles; as well as deflection torque control (GMR), which ensures that the driving conditions are stable during cornering.

1) Torque around the vehicle's height axis

Thus, in this context, a vehicle is a four-wheeled vehicle equipped with a hydraulic braking system. In the hydraulic braking system ··· ················································ · · · · · · · · · · · · · · · · · · · · Each wheel has a brake to which an inlet valve and an outlet valve are assigned. The wheel brakes are connected to the master brake cylinder via the inlet valve, while the outlet valves lead to a non-pressurized container or a low pressure container. Finally, there is an auxiliary pressure source that can apply pressure to the wheel brakes regardless of the position of the brake pedal. The inlet and outlet valves may be electromagnetically operated to control the pressure in the wheel brakes.

Four dynamic speed sensors (one per wheel), a deflection-speed sensor, a cross-accelerometer and at least one brake-pressure-sensed pressure sensor are used to detect driving dynamics. If the auxiliary pressure source is arranged in such a way that the pressure generated by the driver cannot be distinguished from that of the auxiliary pressure source, the pressure sensor may be replaced by a pedal or pedal force gauge.

With such a large number of sensors, a fallback solution is preferably implemented. This means that when a part of the sensorics fails, only the component of the control that is referenced to that part is turned off. For example, if the deflection-speed sensor fails, it is not possible to control the deflection torque, but anti-lock, drive slip control and electronic brake force distribution are still operational. The control of the driving stability can therefore be limited to these other three functions.

When driving stability control, the driving behavior of the vehicle is influenced in such a way that the driver can better control it in critical situations or that critical situations can be avoided in advance. Critical situation refers to the unstable state of a vehicle. In the extreme case of an unstable condition, the vehicle does not follow the driver's instructions. The function of driving stability control is thus to make the vehicle behave within the physical limits required by the driver.

While braking slip control (anti-lock braking), traction slip control and electronic brake force distribution are of primary importance for tire slip on the road, there are additional quantities involved in steering deflection, such as the deflection angle Ψ '.

Different vehicle reference models can be used to control the deflection torque. Calculation is the simplest based on the one-shot model. In this model, the front wheels and the rear wheels are paired in one pair, which is located in the longitudinal axis of the vehicle. With a two-track model, the calculations are much more complicated. However, since lateral displacements of the center of gravity (swinging movements) can be taken into account in a two-push model, the results are more accurate.

··· · __ 5 - ··· · ♦ ··· · ···

Γ »

For the one-shot model, the following system equations apply in the state space representation:

F 1.1 β '= .Cu (β / ν) - Ψ' + ci2 (Ψ '/ v ² ) + C13 (He / v)

F 1.2

Ψ '' = ο ₂ ιβ + C22 (^ '/ ^v ) + ^c 23 &

The drift angle β and the deflection angle Ψ 'are the state determinants of the system. The input quantity acting on the vehicle is its steering angle, from which the vehicle receives the deflection angle velocity Ψ '. The cü model coefficients are formed as follows:

F 1.3 '11' 12

Ί3 ^C h ^C vm

^C h ~ ^C v

Θ .2 -, 2 ^C h ¹ h ^{+ J} v '23 '21 • · V

Here and c _{v are} the stiffness due to tire flexibility on the rear and front axle, wheel suspension flexibility and steering flexibility. 1 ^ and l _{v is} the distance between the rear axle and the front axle of the center of gravity of the vehicle. Θ the moment of inertia of the vehicle deflection, ie the moment of inertia about the height axis of the vehicle.

This model does not take into account length factors and center of gravity shifts. In addition, this approximation applies only to low angular velocities. Thus, the accuracy of this model decreases with decreasing bending radii and increasing speed, but the computational effort is not too high. This one-stroke model is further discussed by Adam Zomotor

Fahrwerktechnik: Fahrverhalten (Vogel

Buchverlag, Würzburg 1987).

German patent application DE-40 30 704 A1 proposes a more accurate two-wheel vehicle model than the one-stroke model.

Here the state determinants are the deflection angle Ψ 'and the drift angle β. However, when using a two-track model, it should be remembered that enormous computing power is required to provide regulatory intervention in a reasonably short period of time.

As with any regulatory intervention, it is important that the calculated set values to be generated are reasonable when adjusting the deflection torque. This means that in certain driving situations, the ideal behavior of a vehicle cannot be enforced simply because even the most targeted regulatory intervention does not know the physical laws ··········································································· ··· • · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · For example, high force locking is not possible on a smooth road, so that the speed and direction of the vehicle can only be changed slowly. While coarse friction coefficient of the friction of the roadway is sufficient for the anti-lock system, it is desirable to determine the friction coefficient of the road more accurately when adjusting the deflection torque, as it is more complicated in model calculations than in the case of anti-fouling.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a system for controlling the torque of a four-wheeled vehicle in cornering with a device for detecting the current coefficient of friction, which senses the coefficient of friction only when the vehicle is in cornering.

According to the invention, this object is solved by continuously updating the determined friction coefficient, at least during the control of the deflection torque, and calculating it according to the following formula:

int where μ is the coefficient of internal friction, aq _{uer is} the measured cross- _acceleration , ai _on g is the measured or calculated longitudinal acceleration of the vehicle, g is the gravitational constant, and 1 <f.

Since the control of the deflection torque is used to handle cornering, it is sufficient to observe the coefficient of friction only when cornering.

• ·· • ·

A vehicle equipped with a deflection-torque control usually has one or more devices for measuring cross-acceleration. With these measurement data, it is possible to determine the coefficient of friction, which is not only limited to the acceleration of the vehicle, but also takes into account the cross-acceleration.

However, since in most cases they do not reach the force limit during cornering, it is advisable to assign a higher coefficient of friction to the friction coefficient, which is based on the accelerations and reflects the current force utilization, which approximates the maximum possible force utilization.

The algorithms in the sub-claims indicate when and for how long a friction coefficient should be detected in a curve.

The following describes, with reference to Figure 29, how such a stability control system can be constructed. From the figures

Figure 1 is a schematic block diagram of the general design of a stability control system, a

Fig. 2 is a block diagram of the structure of a deflection-torque controller, a

Figure 3 is a flow chart for determining a driving position, such as a bend, a

Figures 4 and 5 are flow diagrams for determining the road friction coefficient, where Figure 5 is to be inserted into Figure 4, with the following: ··············································································· * · · • ·· · «« ·· · * ··

Figures 6 and 8 are schematic block diagrams of a combined method for determining the instantaneous value of the drift angle and the drift angle in different representation modes,

Fig. 7 is a schematic block diagram for directly determining the angular velocity of drift based on kinematic considerations as part of the combined procedure of Fig. 6,

Figure 9 is a control loop for controlling the stability of the vehicle by varying the calculation model for the vehicle in a speed dependent fashion,

Figures 10 and 11 are diagrams depicting a vehicle's ramp angle dependence on the rake angle and the speed vector of each wheel,

12-15. Fig. 4A is a block diagram of a control circuit for controlling travel stability, wherein the quantities compared in the comparison unit are derivatives of the deflection angle velocity;

Fig. 16 is a control circuit for determining travel stability using the pressure gradient and / or the valve brake switching time of the vehicle brake as a control characteristic,

Fig. 17 is a schematic block diagram for describing an additional deflection torque controller, a

Fig. 18 is a schematic block diagram for describing a low pass filter, a

Fig. 19 is a flow chart for calculating the corrected required deflection angle, a

Fig. 20 is a schematic block diagram for calculating the corrected auxiliary deflection torque, a

Figure 21 is a schematic representation of a vehicle, a

Figure 22 is a schematic block diagram for describing a distribution logic unit, a

Figure 23 is a schematic representation of a vehicle and offensive forces with the steering wheel rotated,

Figure 24 is a diagram for describing the lateral force and length factors as a function of wheel slip, a

25A. 25B and 25B. Schematic representation of motor vehicles for describing under-control and over-control behavior;

Figure 26 is a flowchart of the decision logic within the distribution logic unit, a

Figure 27 is a schematic block diagram for calculating switching times for inlet valves and outlet valves,

Figure 28 is a diagram for describing time intervals within a computing cycle, a

Figure 29 is a schematic block diagram for determining the tire pressure.

In the following, with respect to Figure 1, the course of a stability control is generally described.

The vehicle 1 constitutes the so-called control section. The vehicle 1 is influenced by the quantities specified by the driver. These are the PFahrer brake pressure exerted by the driver and the steering angle δ. The vehicle 1 measures the resulting quantities, such as the actual torque of the engine, Mjyjot, the aquer cross-acceleration, the deflection angle Ψ ', the wheel speeds and the hydraulic signals, such as the wheel offsets ····································································· • · · · · ··· ···························································· ···

- 11 others. To evaluate this data, the stability control device includes four electronic controllers, namely the regulator 7 for the anti-lock braking system (ABS), the regulator 8 for the traction control (ASR), the regulator 9 for the electronic brake force distribution (EBV) and the regulator 10 GMR. The anti-locking regulator 7, the slip control 8 and the electronic brake force control 9 may be adapted without modification.

The wheel speeds are applied to the anti-lock controller 7, the drive slip control 8 and the electronic brake force control 9. In addition, the drive slip control controller 8 receives information about the remaining motor torque, the actual torque Mj ^ ist ist of the motor. This information is also provided to the regulator 10 for deflection torque control. In addition, sensors receive data on the vehicle's cross- _acceleration aq _Uer and deflection angle Ψ '. Since the anti-locking regulator 7 in any case determines the reference speed ^v ref of the vehicle based on the individual wheel revolutions of the vehicle wheels, this reference speed should not be calculated from the anti-locking regulator 7 but from the anti-locking regulator 7. buy. There is only a small difference in the way in which the yaw torque control is calculated, where the reference speed of the vehicle is calculated or when it is taken into account. ·····································································

- We make our own calculations to control the 12 torque. The same applies, for example, to the acceleration aj_ _on g of the vehicle. Its value can also be calculated in the anti-blocking regulator 7 and passed to the regulator 10 of the deflection torque control. This is of limited value for determining the friction coefficient μ of the roadway, since a more precise friction coefficient is required for controlling the deflection torque than for the anti-lock system.

Each of the four electronic stability control devices, i.e. the regulator for deflection torque control, the anti-lock control 7, the drive slip control 8 and the electronic brake force control 9, operate in parallel and are produced independently for each wheel according to its own control strategy. the brake pressure values PqmR ' ^P ABS' ^p ASR 'Pebv to be generated.

In parallel, the drive slip control regulator 8 and the deflection torque control controller 10 additionally calculate the engine torque M ^ sr AND MgtellM to be generated.

The 10 deflection torque control controllers define each of the Pgmr wheel pressure values to be generated as follows:

The deflection torque control regulator 10 first calculates an additional Mg deflection torque, which leads to stabilization of the vehicle within a bend when this torque is generated by proper operation of the brake. This is it

V · · ·

- 13 Mg 2 distribution logic units, which can also be considered as the control part 10 for deflection torque. This distribution logic unit 2 also receives the driver's wish to slow down the vehicle, which is detected on the basis of the PFahrer brake pressure exerted by the driver. The distribution logic unit 2 calculates from the predetermined deflection torque M _g and the brake pressure demanded by the driver the brake pressures P _G MR for the controlled deflection torque, which may be very different for each wheel. Like the other controllers, the anti-lock controller 7, the drive slip control controller 8 and the electronic brake force regulator 9, these brake pressures for the controlled deflection torque P ^ MR are given the priority switches to determine the brake pressure 3 to optimize the function. This 3 prioritáskapcsolás the driver's wishes, taking into account to determine the optimal road stability ap _so'll tire pressures required.

These preset pressures are either the same as those given by one of these four controllers, or they form a superposition.

The same applies to engine torque as to wheel brake pressures. While anti-lock braking and electronic brake force distribution only affect the wheel brakes, differential torque control and drive slip control affect engine torque. In the 10 regulators of the deflection torque control and the 8 regulators of the drive slip control, the engine is powered by the motor.

- MgtellM ^M ASR values to be calculated separately for 14 torques are also evaluated in this case by the 4 priority switches and superimposed on a specified torque. However, this required torque ij_ M _so it can also be set up value computed by one of the two regulators.

On kerékféknyomás p II and _so the value of the engine torque M required _so u achieved by a driving stability control intervention in the braking and the engine. For this purpose, the pressure controller 5 is further provided with hydraulic signals or values corresponding to the actual wheel brake pressure. From these, the pressure controller 5 generates valve signals which are applied to each of the wheel brakes in the vehicle 1. The motor controller 6 controls the drive motor of the vehicle in accordance with M _so ] _] _. This results in a modified M ^ _st motor torque. This then again results in different input quantities for the four electronic controllers 7, 8, 9 and 10 of the stability control device.

2. Structure of the deflection torque regulator (GMR)

Figure 2 shows a schematic block diagram of how the additional Mg deflection torque for the distribution logic unit 2 is determined within the controller 10 of the deflection control. To do this, the input quantities are the steering angle δ, the reference speed v _re f of the vehicle _from the anti-lock controller 7, the measured transverse acceleration aq and the measured deflection angle sz'Μθεε. The vehicle's reference speed v _re f passes through a 17 filter which sets a constant value above zero at low speeds ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ so that the denominator of a fraction does not have zero in further calculations. The unfair value of v _re f is given to only one activating logic unit 11, which recognizes the stationary position of the vehicle.

This direct sensing of the vehicle reference speed v _re f by the actuating logic unit 11 may also be omitted, assuming that when the filtered reference speed v _re fpj _] _ of the vehicle assumes a constant minimum value, the vehicle is stationary.

Control by controlling the yawing moment is built into a 12 vehicle reference model, which is the δ steering angle, the vehicle v _from f _D j _] _ calculates the established value of the deflection angular velocity ΔΨ 'changes on the basis of filtered reference speed and the measured deflection angle speed.

For these calculations, the friction coefficient μ of the roadway is also needed to keep the values to be created within the physically possible limits. The friction coefficient μ is calculated by the friction coefficient and position recognition unit 13 as an estimate of μ ^Λ . If the friction coefficient determined in the anti-blocking control is sufficiently accurate, this can also be applied, or the friction coefficient calculated in the regulator 10 of the deflection torque control is passed to the anti-blocking regulator 7.

For its calculations, the coefficient of friction and position recognition unit 13 uses the filtered reference speed v _re fpj_p, the measured cross-acceleration of the vehicle aq, the measured deflection angle ^ F'Mess, and the steering angle δ.

The position recognition unit distinguishes between different situations, such as straight forward, cornering, reverse, and vehicle parking. It is assumed that the vehicle is stationary when the filtered reference speed v _re fpj_i of the vehicle reaches a constant minimum value. Instead of the non-vintage reference speed of the vehicle, this information can also be applied to the activating logic unit 11 to detect the vehicle's stationary position. To detect reverse, we use the sign of the deflection angle Ψ 'for a given steering angle δ to be opposite to that of the forward direction. For this purpose, the measured deflection angle velocity '^' Mess is compared with the predetermined deflection angle velocity predetermined by the vehicle reference model 12. If the signs are always the opposite, and this is true for the time derivative of the two curves, the vehicle is in reverse because Ψ * 3θ11 is always calculated for forward motion, since conventional speed sensors do not detect information about the direction of rotation of the wheel.

Finally, the vehicle v _from fpj _] _ determining the kinematic drift angular velocity on the basis of filtered reference speed of the vehicle measured by ^the quer keresztgyorsulása measured and the deflection angular velocity ^ Mess, briefly taken 14 β'kinematic determination.

• ··················•••••••••••••••

The calculated value of the drift angle passes through a first-order low-pass filter 15 so that the peaks are cut off when the drift angle changes sharply. The low pass filter 15 provides an estimate of the drift angular velocity β ' ^Α to the activating logic unit 11 and to the program 16 implementing the deflection torque control law. The program 16 also uses the value to be created for the change in deflection angle ΔΨ ', which is the difference between the measured deflection angle Ψ'μθξξ and the required deflection angle velocity Ψ' _3Ο11 calculated from the _{vehicle reference model} . From this we determine the additional Mg deflection torque for the vehicle, which must be achieved by the brake pressures.

The 16 programs operate continuously to ensure that valid regulatory features are always available. It depends on the activating logic unit 11 whether these adjusting moments pass to the distribution logic unit 2 shown in FIG.

The activating logic unit 11 receives not only the value of the vehicle reference velocity v _re f and, as described, the drift angle velocity β ' ^Λ , but also the required deflection angle velocity Ψ' ₅ ο11 from the measured deflection velocity | ΔΨ '| deviation in absolute value as well

and, in the case of reverse, information from the coefficient of friction and position detection unit 13.

If the vehicle is in reverse, Mg transmission is interrupted The same applies when the vehicle is stationary or • · ν ·· »* · <

fe ·> · · · 3 · fe «· * · · · · · · · · · · · · · · · · · · · ·

- 18 if neither the estimated drift angle velocity β ' ^Λ nor the value of the deviation angular velocity change ΔΨ' to be generated reaches a value that requires control. The logic circuit for calculating the MstellM torque of the motor is not shown.

2.1. Friction factor and position recognition

Turning to the recognition of friction factor and position, a

Figures 3, 4 and 5 show flow diagrams of logical processes in the coefficient of friction and position recognition unit 13.

Figure 3 shows position recognition. The process illustrates eight different driving positions. These are the following:

<0> vehicle stationary, <1> steady straight forward, <2> accelerated straight forward, <3> slowed straight forward, <6> reverse, <7> steady corner, <8> accelerated corner, <9> slow motion cornering.

In the flowchart, logical branches are represented by rhombuses.

Starting from a particular position 51 to be determined, the rhombus 52 first determines whether the vehicle is stationary or not. If the filtered reference speed of the vehicle v _re fpj _] _ adds v _m j_ _{n to the} minimum ································································· · · ····· ··

- 19 lis, we assume the <0> position, that is, the vehicle is stationary. If v _re fp-Q exceeds v _m j_ _n , then the rhombus 53 will query the result of the previous position recognition cycle.

If the previously determined position was reverse, ie <6>, then the reverse is still present since the vehicle has not stopped in the meantime. In the meanwhile, in the 52 diamond, the <0> position would have been recognized in the meantime.

If in the previous cycle of position recognition a position other than <6> was detected, then the absolute value of the aq _Uer cross- _{acceleration in rhombus} 54 will be queried. If this is less than a specified threshold aq _uer min, then the vehicle is assumed to be in straight forward _motion , i.e. one of the <1> - <3> positions. The same applies if the absolute value of the measured aq _uer cross _{acceleration is} greater than the threshold aq _{uer m} i _n , but in the next step, the rhombus 55 recognizes that the absolute value of the steering angle δ is less than a threshold of _{m m} - £ _n. . In this case, there is a measurement error at the aq _uer measured cross _acceleration , which results from the fact that the cross _{accelerometers} are generally fixed in the vehicle's transverse axis and tilt with the vehicle in the case of lateral gradient of the driveway, thus exhibiting an actual acceleration.

Thus, if the vehicle is in a straight forward direction, then the value of longitudinal acceleration ay _on g in rhombus 59 is examined.

If its absolute value is less than a threshold of ai _O ngmin, a steady-state linear progression is assumed. However, if the absolute value of the longitudinal acceleration aj_ _on g is greater than this threshold, then the rhombus 60 distinguishes between positive and negative longitudinal acceleration. If the absolute value of ay _on g longitudinal acceleration is above the threshold ai _ori gmin, then the vehicle is in accelerated straight forward position, i.e. <2>. If the value of a] _ _O ng longitudinal acceleration is below ^the longmin threshold, then this means that there is a negative longitudinal acceleration, that is, the vehicle is in the <3> position in slow-motion straight ahead.

If none of the <0> - <3> positions exist, and the diamond 55 recognizes an δ steering angle greater than the threshold of 8 _m £ _n , then the diamond 56 will query whether the vehicle was in reverse. Reverse detection is only required at this point, since the forward angular velocity Ψ 'differs slightly from zero anyway, and thus no regulatory intervention occurs. Reversing must be safely ruled out only when a curve is detected, in which the control of the deflection torque becomes active on its own. This is not possible only on the basis of the signals of the wheel speed sensors, since such sensors transmit only the absolute value of the speed without deducing this direction of travel.

The situation <6>, as mentioned above, is recognized by comparing the measured deflection angle velocity ^ '^ ess with the specified deflection angle velocity Ψ'ξοΙΙ found in the vehicle reference model 12. If the signs are opposite and the same holds for the time derivatives of the two quantities, a

M'Me ss ® ^s Ψ '' soll deflection angle acceleration, the vehicle is in a rearward curve. The signs of deflection angle accelerations are compared to exclude that the opposite signs of deflection angle velocities are not due to phase shift, but are due to a delayed calculation of the required values.

If the conditions for reversing are not met, then a forward curve is involved. In rhombus 57, examine whether this curve is constant speed or not. As before, in the straight forward, rhombus 59 and 60, rhombus 57, the absolute value of ai _on g longitudinal acceleration is first examined. If this is below the threshold ai _O ngmin, it is a constant speed curve, the <7> position. In the case of longitudinal acceleration aiong having an absolute value greater than the threshold ai _on gmin, it is further examined in rhombus 58 whether ay _on g longitudinal acceleration is positive or negative. With positive ay _on g longitudinal acceleration, the vehicle is in accelerated curve, that is <8>, while negative ai _on g longitudinal acceleration is recognized as slow deceleration, which corresponds to <9>.

There are various ways to determine the ag _O ng longitudinal acceleration. It can be determined, for example, _{from the} reference speed v _re f provided by the anti-blocking controller 7. At

It should be noted that such a reference speed v _re f may differ from the actual speed of the vehicle during an anti-lock intervention. Therefore, in the case of blocking, v _re f must be corrected. However, aiong longitudinal acceleration can, under certain circumstances, be obtained directly from the anti-blocking regulator if such calculation is made there.

The position recognition of FIG. 3 is repeatedly and again. The last position found is stored and available in the rhombus.

A possible course of determining the road friction coefficient is shown in Figures 4 and 5. The friction coefficient is determined only when the deflection torque adjuster enters the control. However, since the estimated friction coefficient is not available when entering the regulation, we start with the friction coefficient μ = 1 when starting the regulation.

If the deflection torque control is triggered by an instantaneous driving position, it must be assumed that the vehicle is at least near the boundary of unstable driving positions. Thus, it is possible to deduce the instantaneous friction coefficient of the road from the (actual) measurement values of the vehicle. The friction coefficient determined at the time of entry into regulation further provides the basis for limiting the required deflection angle Ψ'δοίΐ, and thus the ΔΨ 'control difference transmitted to the deflection torque control program 16. The friction coefficient is first determined only upon entry into regulation and is associated with a subsequent update phase which limits the required deflection angle rate to physically reasonable values. Then, based on the initially predefined friction coefficient μ = 1, a maximum friction coefficient μ ^{Λ is} determined upon entry into the control, which is then taken into account when calculating the additional Mg deflection torque.

To do this, first we calculate a friction coefficient p ^A £ nt intern from ^the measured transverse acceleration aq _Uer and a calculated longitudinal acceleration a long, assuming that the force utilization is complete is equal to the instantaneous friction coefficient. However, since it has to be assumed that the maximum force closure has not yet been reached when entering the regulation, a greater friction coefficient μ ^{Α is} assigned to the friction coefficient p ^A int intern by means of a table, a characteristic curve or a constant factor. This friction coefficient μ ^Α is then added to the control so that in the next calculation step, the specified deflection angle Ψ'εοίΐ adapted to the road friction coefficient can be calculated and the control improved. The estimated friction coefficient μ ^Λ needs to be updated even when the control of the deflection torque is applied to the vehicle as the friction coefficient may change during the control. If the control results from the adaptation of the friction coefficient in the vehicle reference model, the angle of deflection is changed, · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·

- 24 ΔΨ 'does not activate it, the friction coefficient μ ^Λ is further updated in TpE _nc j.

If the deflection torque control does not start within this update phase, the estimated friction coefficient μ ^Λ will be reset to 1.

The adaptation or updating of the estimated friction coefficient μ ^Λ may be omitted in certain situations. Such situations are, for example, straight forward, reverse, or vehicle stationary, ie <0> - <4>. These are situations where the deflection torque is not controlled anyway, so there is no need to estimate the coefficient of friction. Updating the friction coefficient may be omitted if the time derivative of the friction coefficient μ ^Λ , ie μ ' ^{Λ, is} negative and the absolute value of the time derivative of the steering angle δ, ie δ' ^0, exceeds a predetermined threshold. In the latter case, it can be assumed that the change in cross- _acceleration aq _{Uer is} due to a change in the steering _angle δ and not, for example, to a change in the friction coefficient.

The friction coefficient calculated in this way is generally valid as it is a medium friction coefficient for all four wheels. Friction coefficients for the wheels individually cannot be determined in this way.

The procedure for determining the friction coefficient will now be discussed with reference to Figure 4. In all driving situations, the vehicle's behavior is influenced by the road friction coefficient over the 61 fields. To determine the corresponding friction coefficient, we first filter, in step 62, the measured transverse _acceleration Sq _Uer . This means that either the measured values are smoothed or the curve passes through a low pass filter so that no extreme peaks occur. Step 63 is the position detection of FIG. The detected driving position is later important in step 74 during the actualization phase. In the rhombus 64, we inquire whether regulatory intervention is needed. In this calculation, we first consider the initial friction coefficient μ = 1. If adjustment is required, the rhombus 65 is queried to determine if this condition was present at the end of the previous cycle of determining the friction coefficient. If this is a control entry, there was no control recognition before, so in step 67 we first determine a P int int internally friction factor. This is calculated using the following equation:

F 2.1 int quer

In the above equation, g is the gravitational constant (acceleration of gravity), g = 9.81 m / s ² .

Then, in step 68, the regQid parameter is set to diamond 65 at 1. In addition, the counting parameter Τμ is set to 1 according to the fact that the first friction coefficient .p ^/ 'i _n ti ⁿ is first determined. In step 69, an estimated friction coefficient μ is assigned to the calculated friction coefficient p ^A j_ _nt intern. This assumes that the existing acceleration components are not based on the full utilization of force. Therefore, the estimated friction coefficient μ ^általában is usually between the observed coefficient p ^A i _n t intern and 1. This completes the determination of the friction coefficient.

Thus, in the next cycle of determining the friction coefficient, assuming the driving position is constant, the decision at rhombus 65 will be reg _o gg = 1. Here again, we define a factor p ^A int that replaces the h ^A int defined in the previous cycle. The parameters defined in step 68 are not updated because μ ^Α £ η ^ was updated during a control. Already in the previous cycle away ^from one was Üold ^cent '®s it remains unchanged. The number of cycles Τμ executed remains 1, since we count only if there is no control. p can then be assigned an updated value of ^the i _n t - as described above - a table using a non-linear change or a constant factor ^μ Λ an estimated friction coefficient.

If, in a cycle, the rhombus 64 is determined to have no control, then the rhombus 71 is further queried that the reg _o gci parameter was last set to 0 or 1 for control. If it was set to 1 in the last cycle, then the number of cycles Τμ in the 72 rhombus is queried. This Τμ number is 1 if there was control in the last cycle. If control was only in the penultimate cycle then Τμ = 2 and so on. Until Τμ reaches a specific number Τμ in rhombus 72, it is incremented by 1 in step 73, and in step 74, the friction factor p ^z 'i _n t intern is updated again. If, in one of the following cycles, the Τμ _Enc j number is reached without control, the control regoj ^ parameter is reset to 0 (75). The estimated friction coefficient ^μ Λ equated with the friction coefficient μ = 1 base. This completes the update phase of the friction coefficient μ ^Λ .

If, in the next cycle, rhombus 64 again recognizes that no control is required, then rhombus 71 maintains the initial friction coefficient μ ^Λ = 1 in field 76 with reg _o id = 0. The friction coefficient is determined again only when the need for regulatory intervention in the rhombus 64 is recognized.

The criteria for updating the friction coefficient n ^z 'int internally in step 74 are shown in Figure 5. Based on the finding in block 77 that ^the friction coefficient p ^A int intern should be updated, in step 78 we derive the time derivatives of the previously estimated friction coefficients μ ^Λ or μ ^Λ ϊ _η ί and the steering angle δ.

If, in step 79, it is recognized that the vehicle is stationary and is not moving forward, i.e., <6> to <9>, then the results of step 78 are evaluated in step 80. The friction coefficient is determined only if, as already mentioned, the reduction of the friction coefficient is not due to a steering action (steering maneuver). The friction coefficient shall not be updated if the vehicle is either in forward or in reverse, or the vehicle is stationary, or the reduction in the estimated friction coefficient μ ^atra is due to a steering action.

2.2. Determination of β ^Λ and β ′ ^Λ

The degree of stability of the driving state is the dominant drift angle β and its time derivative, the drift angle β '. The determination of these values is described below.

2.2.1. Kinematic β'-determination

The kinematic β'-definition 14 contains nothing but, as distinct from any vehicle model, from the measured quantities or the measured values of the calculated quantities from purely physical considerations, as follows:

The transverse _acceleration aq _{uer of} the vehicle center of gravity is _measured transverse to the longitudinal axis in the plane of motion. The center of gravity of the vehicle relative to an inertia system moves with the velocity vector v:

··:

• · *

F 2.2 cos (Ψ + β) sin (Ψ + β)

Here Ψ is the deflection angle, β is the drift angle. The acceleration vector a is obtained as a derivative of time t:

F 2.3 d COS (Ψ + β) a = - ν = v dt sin (Ψ + β) ν (Ψ + β)

-sin (ψ + β) cos (Ψ + β)

The acceleration sensor measures the projection of the acceleration vector on the transverse axis of the vehicle:

F 2.4

-εΐηΨ cosT

F 2.5 ϊ = a quer - ^d quer ^{= SIR} ^ ^{+ V COS} 3

After linearization of trigonometric functions (sinB = β;

cosfi = 1) the equation can be converted:

F 2.6 $ ₌ Δη ££. ψ. _ϊ_β ν ν

The angular velocity Α β 'is calculated from the above differential equation. As the measurement quantity, beside the aq _Uer cross-acceleration, the deflection angle velocity Ψ ', the scalar velocity of the vehicle v and its time derivative ν' are included, and β 'of the previous calculation can be numerically integrated to determine β.

For the first β'-determination, we take ν '= 0. There is a simplification if the last term is usually neglected so that it is not necessary to determine β.

The advantage of the proposed method is that the angular velocity β is derived directly from the sensor signals and can thus be determined in the nonlinear domain of cross-dynamics. The sensitivity of the process to measurement noise and the integration of measurement errors are adversely affected. As a result, the definition of the drift angle may become very inaccurate.

These drawbacks are avoided by combining kinematic determination with a model-assisted procedure. Figure 6 illustrates how to combine a kinematic determination of the drift velocity β 'with an observer-modeled determination. Figure 6 may be inserted in place of the block 18 shown in dashed line in Figure 2. The input quantity of the method assisted by such a model is additionally the steering angle δ, which is represented by a dashed arrow. THE

- 31 β 'through mutual influence and correction of combined methods of defining ^Λ drift angular velocity of the drift angle β itself with less erroneous calculation is possible, so that as β ^Λ can be made available to the control. This is also represented by an arrow drawn in a dashed line.

2.2.2. Combining kinematic β 'determination with observer vehicle model

Figure 6 replaces the block 18 enclosed in dashed line in Figure 2 so that not only the existing drift angle β 'but also the dominant drift angle β can be determined.

Contrary to the purely kinematic determination of the drift angle β ' ^Λ , here, in addition to the kinematic β' determination 83, an observer vehicle model 84 is involved in determining the driving state. The observer vehicle model 84 receives the steering angle δ as input quantity, as does the vehicle reference model 12 for determining the deflection angle. The filtered reference speed of the vehicle v _re fFj_i is entered as a parameter. The measured output quantities - ^the ®rt quer ^{m m} lateral acceleration and yaw rate deflection reached - the necessary kinematics 83 β'-determination, but not necessary for the observer járműmodeilhez 84, which states in its principle, these quantities. An additional term Y, which in the simplest case is the same as the additional deflection torque calculated by the law of deflection torque, represents the change in the behavior of the vehicle caused by a control intervention32 ··· • «« «. Y serves, therefore, to apply the same conditions to the observer's mapped vehicle as to the real vehicle.

The observer vehicle model 84 also provides a value for the λ'obs deflection angle acceleration in addition to a ½'bias drift angle velocity. The value of the drift angle β ^, ered derived from the kinematic β ′ determination after passing through the low pass filter is multiplied by a weighting factor k, while the value of the drift angular velocity ′ Obs ^{Y from} the observer vehicle model is added after the measured deviation angle and multiplied by a factor h [h}?] multiplied by a weighting factor (1-k), k always between 0 and 1. Without the observer vehicle model, it would be k = 1. After adding the two drift angles, the sum is integrated into the estimated drift angle β ^Λ . This is also made available for control at the kinematic drift velocity β ' ^Λ . In addition, the drift angle β ^{Λ is} transmitted to both the kinematic β-determination 83 and the 84 observer vehicle models. A similar correction quantity is the Ψ''obs deflection acceleration calculated by the 84 observer vehicle model.

This is first integrated into the deflection angle velocity and it returns to the observer vehicle model 84 and subtracts ψ,

Mess from the measured deflection angle. This difference is multiplied by a factor h ₂ , which is the number of the following control steps for the 84 observer vehicle model.

·· «·· ··· ·« ··

- It defines 33 ribbing and has a 1 / s dimension. The dimension of the deflection angle velocity multiplied by this factor Ϊ12 is thus equal to the dimension of deflection angle acceleration ható '' so that the two quantities can be added together and, after further integration, form a retrospective correction quantity for the deflection angle rate. During control of the deflection torque, the term Y takes values other than zero according to the additional deflection torque Mg. Divided by the vehicle inertia of the deflection inertia θ, its Y dimension will also be the deflection angle acceleration dimension and added to the sum of the deflection angle accelerations so that the integral correction quantity also takes into account the control effects.

If the system includes an observer vehicle model 84 of Figure 6 that allows a more reliable determination of the drift angle β than would be possible by purely kinematic determination and integration of the drift angle β ′ ^Λ , then the determined drift angle β ^Λ is the deflection torque. can actually be added to the 10 regulators of its regulation.

The kinematic β 'determination in combination with the observer vehicle model is shown in Figure 7. As already shown in Fig. 6, the measured input quantities of the calculation 91 according to equation F 2.6 are the measured transverse _acceleration aq _uer and the deflection angular velocity.

: * ··.

··· · · »· ·« · · ···

The filtered reference speed vp _e fpj _] _ of the vehicle in field 93 is differentiated by the reference acceleration v * R _e f of the vehicle, divided by the filtered reference speed VRefFil of the vehicle in field 94 which leads to a ίβ factor after 95 nonlinear multiplication. As a result of this 95 nonlinear multiplication, the small value of VR _e f / vR _e fpü will cause the factor ίβ to be zero, so this factor, which is ahead of the drift angle β ^Λ , is negligible. The angle of drift β is considered in the kinematic definition of β'only if the acceleration v'p> _e f of the vehicle reaches a significant value. The β ^Λ used here is the combined β ^Λ that is used for both control and feedback in Figure 6. After calculation 91, the value of the drift angle passes through a low pass filter 92 as described previously and thus gives an estimated drift angle β ' ^Λ .

Figure 8 illustrates the operation of the monitoring vehicle model 84 of Figure 6. Here we have chosen a matrix representation where the arrows drawn with a single line represent scalar formations and the arrows drawn with two lines represent vector formations.

The matrix representation is based on equations F 1.1, F 1.2 and F 1.3. The state quantities β and Ψ 'are summed up to the state vector x (t) so that the following system of equations is obtained:

x '(t) = A (v (t)) x (t) + B (v (t)) u (t)

F 2.7

With the system matrix A (v (t)), the input matrix B (v (t)) _, the state vector x (t), and the input vector u (t):

· * · • · »·« · »·· ♦ 9 · • ··« ··· * · · 99 9 · 99

F 2.8

Ct ♦ c h_v mv (t) ^C hh ^C vm V ² (t) mv {t;

^C h ~ ^c v ¹ v

Θ

Θ v (t) _i

Θ χ {t;

β (t: ψ (t:

u (t)

The input vector u (t) contains as input the steering angle δ and the expression Y, which represents the additional deflection torque generated by the deflection control.

Instead of weighting factors used in a weighting matrix and a weighting factor k ₂ weighted summing of the volume set.

F 2.9

1-k 0 • write k 0 1 r 0th

·· • · · «« • ·· • · · «•« ··· · ·

To exclude state quantities, we introduce two vectors, cr and οΨ>, which delete each component of the state vector x (t):

F 2.10

Cfi = [1.0]; £ ψ- = [0, 1]

The dynamics of the observer's vehicle model, that is, the magnitude of the correction steps, is determined by the vector h, the first component of which is dimensionless h] _ and the dimension of its other component h2 ^_ [1 / s]:

F 2.11

Starting from the vehicle model in the state space description (F 1.1 and F 1.2), the following arrangement is used to determine the drift angle β by an observer, as shown in Figure 8.

In Figure 8, the vehicle 101 is depicted solely for the purpose of distinguishing between input quantities and output quantities.

The vehicle 101 is not part of a combined method for the β ^'Λ drift angular velocity determination.

444 • ·· • ♦ · »* · *

4 4 4 • 44 «· * 4 * · 4

444 44 * 4

The addition unit 104 forms the system equations according to F 2.7. To do this, it multiplies the system matrix A by the state vector x and multiplies the input matrix d by the input quantities δ and Y, that is, the input vector u.

Both the system matrix A and the input matrix B enter as the only variable parameter the current reference speed of the vehicle. The time derivative x 'of the state vector x, as added by the addition unit 104, is multiplied by the weighting matrix Κχ of equation F 2.9, and is output to another addition unit 105.

In parallel with these processes, we estimate the drift angle β ' ^{ban in} the direct process 103. To do this, we apply the filtered reference speed VRefpi] of the vehicle, its time derivative v * Ref deriving from the difference unit 102 in the field 93 of Figure 7, the measured transverse acceleration of the aquer, and the measured offset rate ^ P'Mess. equation. In the first step, the last part of the equation is neglected because there is no value for the drift angle β. Once the drift angle is determined, this amount, as already shown in Figure 7, passes through the low pass filter 92. The estimated drift angle velocity β ' ^Λ obtained there is available for further calculations. This β ' ^{Λ is the} same as β' ^Λ -νο1, which we extracted from the dotted line in Figure 2. The scalar quantity β ' ^Λ is multiplied by the weighting vector k2, so that a vector has the first component of the vector that has an angular velocity dimension of 38, and its second component is zero. This vector is also added to the addition unit 105. The vector resulting from the sum of the time derivative x 'of the state vector x formed by equation F 2.7 and the vector obtained by multiplying by k i is integrated in the integrating unit 106 into the state vector x. Multiplied scalarially by the vector cr and οΨ », one of the β and Ψ 'components of the state vector is scalar excluded and further processed. While the excluded 5 ' ^A is introduced on the one hand to the program 16 of the hijacking torque control law and on the other hand to the direct procedure 103, the calculated Ψ.' Is used in the combined procedure only as the amount of state within the observer and the estimated error. To this end, the addition unit 107 is the difference between the deflection angle Ψ 'and the deflection angle Ψ'Μθδε ^m obtained from the observer's vehicle model. This difference is multiplied by a vector h whose first component is dimensionless and gives the magnitude of the correction steps for the drift angle β ', the dimension of the other component s ~ l and determines the magnitude of the control steps when correcting the deflection angle or'.

As a correction amount, we also refer the β ' ^Λ drift angle velocity] to the direct process of the kinematic β'determination as shown in Figure 7, so that * · * · ♦>>>>> · • · «In the next control step, you can assign a value to the last term of equation F 2.6.

The mutual adjustment of the two calculation methods, ie the calculation based on the vehicle model and the calculation based on the kinematic considerations, makes it possible to determine the drift angle β ^, [[velocity?] Much more precisely, so that it can also be applied to 16 programs of

2.3. Vehicle Reference Models

9-15. Fig. 4 is a description of the vehicle reference model.

Figure 9 is a simplified representation of the control loop for controlling the vehicle stability as shown in Figures 1 and 2. The controller 7-9 of FIG. 1, the associated priority switch 3 and the motor controller 6, and the distributor logic unit 2 are illustrated, united with the pressure controller 5. Within the control loop, an additional Mg deflection torque about the vehicle's height axis is calculated and adjusted to maintain the bend desired by the driver. The additional Mg deflection torque is created by targeted braking processes on each wheel. The distribution logic unit 2 determines the flow of braking processes and the selection of the wheels to be braked. The desired direction of travel is given by the driver with the correct angle of the steering wheel. The steering wheel is connected to the steered wheels with a constant gear ratio (steering change), thus adjusting a certain steering angle δ.

2.3.1. Dynamic single track model

The regulator 10 for deflection torque control includes a so-called vehicle reference model 12. The vehicle reference model 12 shown in Figure 2 and the vehicle reference model 302 shown in Figure 9 are identical. The vehicle reference model receives input data (velocity v _re represented by f, rudder angle δ). The vehicle reference model 302 calculates, based on the input data, how large the deflection angle change must be during the time unit (the deflection angle Ψ'βοΙΙ). The comparator unit 303 connected after the vehicle reference model 302 compares the required deflection angle Ψ _'50 ΐ] _ with the

Mess' Mess with measured deflection angle. The comparator unit 303 provides an output quantity ΔΨ 'as the output value, which is the difference between the Ψ-salt and Ψ' Mess. The difference value thus determined goes to the 16 programs implementing the control law for controlling the deflection torque. The program 16 calculates an additional Mg deflection torque based on ΔΨ ', which is applied to the distribution logic unit 2. The distribution logic unit 2 is the base Mg deflection torque and, if applicable, the PFahrer pressure that the driver will produce in the brakes41.:. · '.:: ···, ··.

:: ··. : ··>. ··· «·· ·. ..... "· outputs quantities. These quantities may be brake pressure values or valve actuation times.

The optimal mode of operation of the vehicle reference model 302 is also important in the low speed range. For this purpose, the vehicle reference model 302 may additionally be provided with a steady-state process model 306 in addition to the linear dynamic single-track model 311 described above.

For a steady procession:

F 2.12 soll v 'korr ch

F 2.13 (1 fold

F 2.14 ch

1 * ch c * c. * (I. * 1 vhhv 'm * (c, * 1. - c * 1)' hh \ r \ r · where v = front, front; h = rear, rear; m = weight; 1 = distance from center of gravity 'P'corr' for ^the corrective term Ψ 'or β, respectively.

• · · · · · ··· ····· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·

For the linear dynamic single-flow model, the system equations FI.1 and FI.2 apply.

The switching between the calculation models - the steady-state passage model 306 and the one-pass model 311 - is performed automatically by the switch shown in the vehicle reference model 302, depending on the speed of the vehicle. There is a hysteresis of a few km / h in the switching processes from one model to another. Below the switching threshold, the required deflection angle velocity Ψ'soll is calculated according to the steady-state process model 306. If the velocity increases from a low velocity above the threshold in this direction, then the required value of the deflection angle Ψ'βοΐΐ is calculated using the dynamic 311 single-flow model. This allows dynamic processes of particular importance for higher speed control to be incorporated into the model.

When moving from the steady-state processor model 306 to the one-pass model 311, the values in the one-pass model, such as Ψ _'50 ΐ] _-1 and β, are used as starting values. This reduces the wobble process during switching. Further calculations are made using the 311 single-pass model until the lower speed threshold is reached when the speed decreases. In order to keep the warm-up processes small here, we calculate the correction factors T ' ^, ] ] _r, and Fc required for the procession model for Y' _so n and β, respectively, by the one of 43 · ·· · ······················ ·· · · · · · · · · · · · · · · · • • · · · · · · · · · · · · · ·· ·· calculated value · Overriding model, as well as _from f av 8 speed and steering angle based on input volume.

The corrective values are as follows:

F 2.15

A = Ψ ,, - δ rope soll eh

F 2.16 (1 c, * U aj h 'v h'

The influence of these corrective factors decreases exponentially over time as follows:

F 2.17 Cor (n + 1) = Cor (η) * λ where λ is Ο.,. Μΐ. With n and (n + 1) respectively, counting cycles are counted.

This avoids leaps and bounds, since in constant cases the two calculation methods give different results, so changing the calculation model allows

the required values for control are determined very accurately up to v = 0 km / h.

In Figure 9, we discussed that different models can be used as vehicle calculation models. A steady procession may be a beneficial model. According to the model, the

The required deflection angle ΐ'εοίΐ can be calculated using the formula given above. If we now want to depict such a vehicle calculation model, it is possible to give the measured values of λ and V _re f for a computational coupling, and output the required value of the deflection angle T'soll.

2.3.3. Simplified model

Following is a very simple model for determining the required deflection angle velocity. This is an alternative to the combined model described above. It has the advantage of providing an acceptable result with little computation.

According to the model, the required deflection angle speed Ψ ' ₅ ο11 is calculated as follows:

F 2.18

Ψ'3Ο11 = (δ * V) / 1 • · ·

This equation with equations F 2.14 and F 2.15 is derived from equation F 2.12, assuming that the stiffnesses c _v and c ^ are very high.

This assumption is based on the following considerations.

For the vehicle reference model described above, the required deflection angle velocity Ψ ' ₃ ο11 was calculated either with a dynamic vehicle model (such as a one-pass model) or with a static model (called a steady-state process value) and compared with the deflection angle velocity ^ f ^ Mess. In each case, the value to be generated (and thus the regulatory intervention) directly depends on the goodness of the vehicle model. As these are linear replacement models, the model may in some cases differ significantly from the actual vehicle behavior.

If the actual behavior of the vehicle is additionally altered, for example by loading the vehicle or wearing parts of the vehicle, then the model is not described satisfactorily by the model. Consequently, continuous parameter estimation requires model adaptation. This causes the following issues:

For estimation, excitation must exist, ie the driver should excite the vehicle in the linear range (<0.4 g) with steering control. This is hardly the case with normal running.

In addition, it is not possible to directly estimate all parameters of the linear one-pass model. Therefore, certain parameters must be pre-selected.

• ············ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·urgurg

Thus, regulation based on model assumptions can always provide a satisfactory solution only for the values predefined by the model. In many cases, it may therefore be satisfactory to follow a simpler regulatory approach.

An important goal of driving stability control is to shape the driving behavior so that the vehicle's response to the values that the driver sets on the steering wheel, brake and accelerator is always predictable and well controlled.

Therefore, the vehicle's under- and over-controlled operating conditions must be recognized and corrected to neutral with appropriate brake or motor control intervention.

The basic idea of the simplified control principle is that direct control is used as a control characteristic for under-control and over-control behavior. For this purpose, we define a vehicle control behavior by comparing the mean tilt angle of the front axle and the rear axle (Uy, an). · ^{If the} tilt angle is higher at the front, there is undercontrolled behavior. Conversely, there is overcontrolled behavior. By definition, neutral behavior occurs when the tilt angle is equal to the front and rear.

By:

F 2.19> 0: undercontrolled = 0: neutral <0: overcontrolled.

«V ^{- a} H • · · ···· ··

• · · · · · ··· ·· ·· β «· * ·

Thus, the current driving status of the vehicle can be directly determined based on the difference in the rake angles. Considering the one-stroke vehicle model (Fig. 10), the inclination angle from the steering angle δ, the drift angle β,

Ψ 'can be derived depending on the angle of deflection and the speed v of the vehicle, as follows:

F 2.20a ψΊ,

F 2.20b

Because the drift angle cannot be directly measured or simply calculated, it is not possible to explicitly calculate the individual tilt angles. However, if this is the difference, this quantity can be calculated from the existing measured quantities (rudder angle, deflection angle speed), vehicle reference speed _{re re} f known _from the anti-lock governor and constant wheelbase 1:

··

F 2.21 (Formula on page 46) thus provides an amount that can be used as a measure of under-control and over-control.

Furthermore, by examining the known relationship between the instantaneous radius R of the curve of the center of gravity of the vehicle and the inclination of the ramp,

F 2.22

- {a _v -a _h , it can be seen that F 2.19 is a neutral state, that is

F 2.23 ^V "θ ^{= α Η} assuming a radius R of the curve is determined only by the steering angle δ, i.e.

F 2.24

R = 1 / δ • ···· ································································•

Therefore, it is possible to control which directly uses the calculated angular misalignment as a control characteristic. A prerequisite for this control is that the control characteristic is kept at a low absolute value, so as to achieve approximately neutral behavior. Optionally, this tolerance threshold may be made asymmetric so that tolerance in the direction of overcontrolled behavior may be reduced.

Based on these considerations, the required deflection angle Ψ ' _3Ο ιι can be calculated (F2.18). This required deflection angle velocity Ψ ' _8Ο ιι is then compared to the measured deflection angle velocity and taken as the basis for control in FIG.

2.3.5. Limit the required value

Controlling the driving behavior of a vehicle only makes sense as long as the adhesion of the vehicle wheels on the road allows the calculated additional torque to be applied to the vehicle.

For example, it is undesirable for the control to force the vehicle into the curve defined by the steering angle δ in all cases if the steering wheel is turned too hard or too fast in relation to the current speed.

Therefore, it must be prevented that the prescribed deflection angle velocity ^ 'soll becomes a predetermined value according to the selected vehicle reference model.

This, if only the reference model is followed, can, under unfavorable circumstances, lead to an actual deflection angle Ψ 'at an unintentionally too high steering wheel _angle and at high speeds, so that in extreme cases the vehicle it rotates about its own axis, while its center of gravity is essentially straight. This condition is much less favorable for the driver than the condition in which the vehicle, due to poor frictional conditions, is unable to follow the driver's wish and is pushed in a highly under-controlled straight line. In the latter case, the vehicle moves at least straight and at the same time does not rotate about its axis. The vehicle reference model has additional computational algorithms to avoid these adverse consequences in special cases. These algorithms allow to determine the actual measured speed is valid for ^ 'sollmax maximum deflection angle velocity using the friction coefficient ^μ Λ. The coefficient of friction μ ^Λ is determined by the coefficient of friction and position detection 13. The computational algorithms are based on the theory of a steady-state procession, for which Ψ '= aq _Ue r / ^v (F2.18).

The maximum allowable cross-acceleration aq] _i _{m is} essentially a function of the coefficient of friction, the velocity v, the longitudinal acceleration ai _on g and optionally other parameters. so ·· «• ·· · · ·

- 51 F 2.25 ^a qlim ⁼ f (mu, v, a] _ong '· · ·) ·

The maximum deflection angle is:

F 2.26 ^ 'sollmax ^{= a} qlim / ^v

Therefore, a limit deviation angle can be set which does not directly take into account the driver's preference but contributes to preventing the vehicle from rotating even around its elevation axis in the event of an outbreak.

For details on the appropriate determination of μ, see section 2.1. are discussed in detail below.

It is also feasible that regulatory intervention occurs only under certain boundary conditions. This can, for example may be that activation logic unit 11 of Figure 2 does not give more value to the current Mq two distribution logic unit, if it is too large drift angle β ^Λ have been identified, which may occur depending on the currently prevailing rate.

2.4. Regulatory Law

In the following, the structure of the program of control law in the 16 programs of the deflection torque regulator 10 will be described. The program calculates from the four inputs the additional Mg deflection trace around the vehicle's height axis, which is required for the vehicle's stable behavior, especially when cornering. The calculated Mg deflection torque is the basis for calculating the pressures to be set in the wheel brakes.

Available as input to the Regulatory Act (see Figure 17):

the 500 input: ΔΨ ' the 501 input: ΔΨ ' the 502 input: β ' ^Λ the 503 input: β ^Λ

In the case of using the slope angle as the basis, the input 500 is Δλ and the input 501

Will be Δλ '.

Input 503 is optional. This is particularly the case when the so-called 84 observer vehicle model is included in the total calculation system.

The value at 500 inputs is the difference between the measured deflection angle ^ F '^ ess and the required deflection angle T'oll calculated using a vehicle reference model 12.

The change in the value at input 501 or the amount at input 500 divided by the time interval Tq between a calculation cycle and the next calculation cycle, or the time derivative of the measured deflection angle and · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·

A calculation cycle is a calculation cycle for the stability control device of Figure 1. Such a cycle uses the real time defined by its structure, the cycle time Tq. This should be kept low enough for the effectiveness of the regulation.

The values at inputs 500 and 501, i.e., ΔΨ 'and ΔΨ' ', are first applied to the low pass filter 510 and 511, respectively.

In principle, the two low-pass filters are of the same design. Their structure is shown in Figure 18.

The inlet amount 520 of the low-pass filter shown in Figure 18 is denoted u and the outgoing amount 521 is denoted y. The output quantity 521 is added to register 522 and is available as the previous value y (k-1) at the next calculation. The output quantity 521 for the calculation cycle is calculated using the following formula:

F 2.27 y (k) = λ * y (kl) + (1 - X) * u * k _p , in which you can enter values between 0 and 1. λ describes the value of the low pass filter. At the limit of λ = 0, the recursive function is eliminated: the previous values of y (kl) are not significant for the calculation of the new 521 outgoing quantity ··· ······················································································

- 54 of them. The closer the λ to the value 1 is, the more powerful the previous values are, so that the instantaneous input volume 520 is slowly acting as an output quantity 521.

k is a linear evaluation factor.

The low pass filtering described above takes place at the two inputs at its two inputs 500 and 501 and leads to a filtered value of 515,516.

The same low-pass filter 512 passes through the amount of input 502, i.e. β ' ^Λ . The filtered value 517, like the non-filtered value at input 503, is applied to nonlinear filters. The function of these filters is to set the output value to 0 for low input values and to pass the input value reduced by the limit for input values above a specified limit. The delimitation takes place in both the negative and positive ranges. The β'th ® ^s ftfh limit may be a quantity that is fixedly implemented in the program or an amount that depends on additional parameters, such as the coefficient of friction between the tires and the road. The limit values in this case are calculated separately as a linear function of the friction coefficient.

Each of the four quantities, i.e. the filtered values 515, 516, 517 and 518, are weighted by an additional linear factor in steps 530, 531, 532, and 533, respectively.

These factors are fixedly implemented in the computing system. These factors can be calculated on the basis of the appropriate vehicle models, but generally require fine-tuning by driving tests. This way, for all vehicles, ille · ··· ············································································ ♦ ··

- determine the appropriate set of linear factors for 55 vehicle types. The weighted input quantities at the inputs 500, 501, 502, 503 are then added (540 addition tags). Thus, the additional Mg deflection torque is obtained, which is taken as the basis for further computation of the program.

However, in practice it has been found that the calculated deflection torque needs to be modified.

There are two ways to do this:

1. modification of the input quantities, in particular ΔΨ ',

2. filtering the calculated Mg deflection torque.

In both methods, we try to control not only the deflection angle velocity but also the drift angle.

2.4.1. Adjusting the input quantities

Using a vehicle reference model, a required value of the yaw rate is calculated as discussed above. As the applied vehicle reference model may not fully match the actual conditions, the model calculation results generally need to be corrected once more. In the reference model, essentially the values provided by the deflection angle sensor and a steering angle sensor are evaluated. The calculated target deviation angle velocity can be corrected by taking into account the values provided by a cross-acceleration sensor.

· * ·· ··

Evaluation can take place in different ways. In the following, we propose a method in which the measured transverse acceleration is first converted to an angular velocity β ' ^Λ . This value is used to correct the required value for the deflection angle.

β ^'Λ example, by calculating the kinematic 14 β'-definition and low-pass filter 15 (Figure 2).

The procedure is carried out according to the scheme shown in Figure 19. The estimated value of the drift angle β ' ^Λ , after optional low-pass filtering, is compared with a first thj threshold (400 rhombus). The significance of this comparison becomes clear only after correcting the required value of the deflection angle Ψ ' _Ξο , and is therefore explained in more detail below.

If ³ β ' ^Λ3 > thx, then the absolute value of β ^{, Λ} is compared to a second th2 threshold (step 401), where the second threshold is greater than the first th}. If this threshold is also exceeded, then the time integration of the drift angular velocity β ' ^Λ is first performed in step 402. For this purpose, the drift angular velocity β ' ^{Λ is} multiplied by the cycle time Tq and added to the Intgj__x pre-integration result. The integration steps are counted by n, so that the number n is increased by 1 after integration (step 403). The integration time is thus given by n the number of integration steps that have occurred. The Intgn (β ' ^Λ ) integration result is compared with a fis threshold (step 404). The size of the threshold gives ···· ··· the maximum allowable deviation from a theoretically acceptable drift angle. AE _s threshold approx. It is around 5 degrees.

If this threshold is exceeded, the required deflection rate Ψ'3θ11 is re-evaluated by an additive constant S (step 405), which depends on the instantaneous drift angle β ' ^Λ and the number of integration steps n. This means that each new cycle, in which the threshold is exceeded u _s, reducing the deflection angle required speed again. The additive constant S is either added to or subtracted from the sign of Ψ ' _5Ο 11, so that the absolute value of the required deflection angle is always reduced. If Intg _n is below the β ₃ threshold, then Ψ 'is not limited (step 407).

In a new cycle, it is again checked that the estimated drift angle velocity is less than the threshold value. If this is the case, it is understood that the vehicle has re-stabilized. As a result, in step 406, n is again set to 0, and for further calculation, in step 407, a predetermined deflection angle rate is used that is not corrected, i.e., equal to the value available as a result of the vehicle reference model. In addition, the integration value Intg _n _i is set to zero.

If the absolute value of the drift angle velocity is greater than th ^ but not th ₂ , the old value of Intg _n remains unchanged, meaning that integration in one cycle is omitted. The previous restriction remains. When the th ₂ threshold is again exceeded, the integration continues with the integration of «« · ♦ ♦ · · · · · · ·.

2.4.2. Correction of Mg

Another possibility is the manipulation of the Mg deflection torque calculated by 16 programs (regulation law). To do this, we form the difference between the previous value M 1 - (k-1) and the current value of Mg (k). Index 1 indicates that these values are direct results of the deflection torque controller and are not calculated from the following correction. This difference is related to the cycle time Tq to give Δ Mi. To this gradient Δ M] _ we add a correction gradient given by a correction factor β ^{, Λ} .

The corrected gradient is then multiplied by the cycle time Tq and added to the deflection torque M (k-1) of the previous calculation. This gives the current Mg torque, which is the basis for further calculation.

This calculation is performed by the logic unit shown in Figure 20. The calculated torques from the 16 program (control law) subprogram are entered into the shift register 420. The first register 421 of the shift register 420 always has the current value Mi (k). The second register 422 of the shift register 420 has the previous value Mi (k-1). As soon as there is a new value Mi, the value in register 421 is shifted to register 422 and the value in register 421 is replaced by the new value. The values in registers 421 and 422 are given to the computational logic unit 430, which is given by the following formula ···

- 59 counts for one MFA:

F 2.28

ΔΜ = M _x (k) - Mi (kl) + a * fi '* T ₀ .

The PC logic unit 430 also is applied to the estimated β ^'Λ drift angular velocity kinematic β'-determinations. In addition, a correction factor is stored in memory to convert the drift angle velocity into a torque change. The new torque M (k) shall be calculated according to the following formula:

F 2.29

M (k) = M (k-l) + ΔΜ.

Register 431 stores the current value of the corrected torque, and register 432 stores the value obtained from the previous calculation. The value in register 431 is taken as the basis for further calculation.

3. The distribution logic unit

3.1. Additional deflection torque by application of braking forces

In order to keep the vehicle steady even when cornering, the steering angle must first be detected. The steering angle defines the curve of the vehicle as desired by the driver. With a steady, steady curve, the vehicle must run at approximately constant drift angle and constant deflection angle. The driver must compensate for deviations from this drift angle or from this deflection angle by counter-steering. However, this is not always possible, and especially not when the driver is cornering at a cornering speed. In such situations, the vehicle must be braked braked and additional torques applied around the height axis which must result in the actual yaw rate being adapted to the desired yaw rate. The control algorithms that describe these relationships have been previously described, so there is no need to discuss them in detail here.

In any case, the problem remains that the additional Mg deflection torque calculated by the control algorithm is implemented in a suitable manner by targeted application of the braking forces.

In the case of hydraulic brakes, therefore, the task is essentially to determine the brake pressure for each wheel brake. In addition, the torque to be applied around the height axis must be designed so that the minimum pressure is applied to each brake. Therefore, it is proposed that a coefficient be determined for each wheel and the brake pressures determined from the deflection torque of the vehicle to be generated and the respective weighted coefficient.

As mentioned above, the coefficients favorable for hydraulic braking systems, in particular, are determined by the coefficients 61 · W · · · · · · · · · · · · · · · ··· · ·· »·· ··· ft · ·· so that the brake pressure can be directly determined for each wheel brake. The coefficients are weighted in this way. dividing each coefficient by the sum of the squares of all the coefficients.

Each coefficient then determines the relationship between the tire pressure and the proportion of the individual wheel forces thus generated in the vehicle's deflection torque. Quantities that vary as the vehicle travels are included as parameters in the determination of each coefficient. These are primarily

- the steering angle δ,

- coefficient of friction μ between tire and roadway,

- mass of vehicle m,

- N _z axle load distribution.

Quantities that influence the calculation of the coefficients and that are vehicle-specific or brake-specific, for example in the case of disc brake systems, are:

- surface A of the brake pistons,

- number of pistons per wheel brake,

- coefficient of friction pr between disc and brake pad,

- ratio of effective friction radius to dynamic tire radius s,

- and the efficiency of the brake η.

The advantage of the proposed calculation method is that the predetermined additional deflection torque can be used to quickly calculate the appropriate brake pressures. If the above parameters are threaded · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · terms ·

- 62, this is taken into account by changing the coefficients in the calculation of the brake pressure.

While some influencing quantities are linear in the calculation of the coefficients, above all the dependence of the coefficients on the steering angle 8 is non-linear.

However, it has been proved that the linearized estimation of the relationship between the individual coefficients and the steering angle gives satisfactory results.

Figure 21 is a schematic representation of a vehicle in a straight line. The vehicle has four wheels 601, 602, 603, 604. Each wheel is assigned a 605, 606, 607, or 608 wheel brake. The wheel brakes can be independently controlled. The brake torques exerted by the wheel brakes exert braking forces on the contact surfaces of the wheels on the road surface. For example, when the wheel brake 605 is controlled, a force F is applied to the wheel 601, which generates a torque M (positive torque in the example) around the elevation axis.

Such torques around the height axis of the vehicle can be used specifically to keep the vehicle stable on the path desired by the driver.

The vehicle also has sensors. Such sensors are wheel sensors which detect the angular velocity of the wheels 601, 602, 603, 604. In addition, we detect the steering wheel angle with the steering sensor 612. There is also a sensor 613 for detecting the deflection angle velocity.

Based on these sensors, which detect both driver preference and vehicle behavior, • · ·

63 calculates a deflection torque to be realized which, when applied, adjusts the deflection angle and the drift angle of the vehicle to the driver's preference. For this purpose, the wheel brakes 605, 606, 607, 608 are independently controlled. For this purpose, a control device is provided which forms part of the complex program for controlling the stability of the ride.

The theoretical situation is shown in Figure 22. Program 16 is a program module that calculates the Mg deflection torque. The 22.

FIG. 2B illustrates a control device that calculates the pressures p _{xx to be} generated in each of the wheel brakes 605, 606, 607, 608. The determined pressure values 622, 623, 624, 625 can be further evaluated and converted into control signals suitable for wheel brakes 605, 606, 607, 608.

The control device itself consists of two parts. The first computing part 630 of the control unit calculates coefficients c _xx for each wheel. The coefficients c _xx establish a linear relationship between the pressure in the wheel brake and the part of the deflection torque generated by the braking force on the appropriate wheel. The second computing part 631 of the control unit calculates each p _{xx by} weighting each coefficient and taking into account the deflection torque Mg to be realized,

622, 623, 624, 625.

Pressure values and coefficients are distinguished by indices.

• · · ·· ····· ··

- 64 The indices are as follows:

v: front h: rear

1: left r: right x: stands for either v / 1 or h / r.

The first computing part 630 takes into account the steering angle provided by the evaluation unit 632 of the steering sensor 612 to the calculation process. In calculating the coefficients, account shall be taken of the coefficient of friction μ, which is derived from the estimation unit 633 from the rotational behavior of the wheels (see also point 2.1). The rotational behavior of the wheels is determined by a sign of the wheel sensors on each wheel. The weight of the vehicle as well as the load distribution N _z are also included in the calculation. These are defined by the evaluation unit 634, which analyzes the behavior of the vehicle in different situations. The first computing part (program part) 630 has access to a memory 635 containing the vehicle-specific and wheel-specific values expanded above.

From these values, a coefficient c _{xx is} calculated for each wheel. The values of the coefficients 640, 641, 642, .643 can be calculated in parallel or sequentially. The calculation is based on a function that is implemented in the program. This function takes into account known relationships between braking pressure and braking force. The relationship is usually linear. Only the steering angle δ has to be considered separately. Here's how to take the steering angle into account.

In the second step, in section 631, each of the 640,

641, 642, 643 in parallel or sequentially

calculate the pressure values for each wheel brake using the formula:

F 3.1a 'xl xl ^C vl ^{+ C} vr ^{+ C} hl ^{+ C} hr

F 3.1b xr xr • vr 'hr

Calculating the individual pressures according to this formula has the advantage that only relatively small pressures on the wheel brakes are required to achieve the calculated brake torque, and the control of the brake pressure reacts very sensitively and quickly to changes, in particular steering angle and friction factors. changes.

His steering angle is taken into account when calculating the factors: Figure 23 shows a schematic view of a vehicle having its front wheels 601, 602 rotated. The distance S between the front wheels and the distance l _v from the center of gravity 610 to the front axle are shown.

The wheel plane 650, 651 engages steering angle 652, 653 with the longitudinal axis of the vehicle. For the sake of simplicity, we assume that the angles δ and rudder 652, 653 are the same. Based on approximation considerations, the effective lever lever h] _ and h _r , respectively, acting on the braking force F, acting on the wheel planes 650, 651, is calculated as follows:

F 3.2ah _r = s / 2 + δ * l _v

F 3.2b hg = s / 2 - δ * CurveAs the small steering angle approximation is not always satisfied, the following formula was found to be favorable where appropriate:

F 3.3a

S

F 3.3b

If the calculated levers are less than zero, we make them equal to zero.

• · ·

- 67 The coefficients c _{xx are} calculated as follows:

F 3.4 ^c xx ^c hydxx * ^ Ι, Γ 'where Cftygxx takes into account all parameters except the steering angle δ.

In this way, the coefficients can be expressed as the product of two terms. One term defines the effective lever arm and the other term is independent of the steering angle.

3.2. Additional deflection torque by reducing lateral forces

One method of generating unilaterally applied braking forces consists in controlling the wheel brakes so that the wheels are braked with different strengths. In the previous paragraph, we have described a procedure that accomplishes this.

This procedure has its limits when it comes to adjusting the stability of the pedal during braking, when a certain amount of brake pressure has already been applied to the wheel brakes due to the driver's braking. The procedure described above can in principle be applied to this case as well. Instead of absolute pressures, changes in pre-set brake pressures are determined.

In any case, the following problems occur. If a wheel brake is already subjected to very high pressure so very high braking forces are applied, then increasing the brake pressure ···

The 68 does not necessarily increase the braking force, since the traction limit between the tire and the road has been reached. In this case, the linear relationship between brake pressure and braking force, based on the model mentioned above, no longer exists.

The limit of braking force on one side of the vehicle may be offset by reducing the braking force on the other side of the vehicle in accordance with the control of the deflection torque.

However, this has the disadvantage that as the braking force decreases, the vehicle deceleration is also reduced. This is not always acceptable as the vehicle must be stopped within the shortest possible distance when the driver initiates a brake application. However, reducing excessively the actual deceleration of the vehicle to the driver's wishes is generally unacceptable. Here's how to solve this problem.

The wheel brakes of at least one of the wheels are controlled such that the wheel slip is adjusted to be greater than the slip at which maximum force is achieved. In this procedure, we take advantage of the fact that the maximum braking force transmitted, that is to say the longitudinal force on the tires, is about 10%. It achieves 20% slip (0% free wheel, 100% locked wheel), and values over 20% offer only a slight reduction in transferable braking, so 20% to 100% slower vehicle deceleration decrease.

However, at the same time, when examining the transmissible lateral force, i.e. the force acting perpendicular to the wheel plane, there is a strong dependence on the wheel slip. This is manifested in the fact that as the slip increases, the transferable lateral force is greatly reduced. The • ·

- 69 In the slip range above 50%, the wheel behaves like a blocked wheel. This means that by this time, lateral forces can hardly be exerted.

Cleverly selecting the wheels on which a large longitudinal slip is set can result in controlled vehicle alignment, whereby the variation of the deflection angle caused by the alignment must correspond to the desired change. Since the lengths in this method are substantially maintained, but the lateral forces are significantly reduced, the deflection angle velocity can be controlled without significantly reducing vehicle deceleration.

The selection of the wheel to be extended for at least a short time with an increased slip is done according to the following rules. To do this, let's look at a right-hand corner designed by the driver. For left turns, the corresponding mirrored rules apply. This may result in the vehicle not turning as sharply as expected. In other words, the vehicle is under-controlled. In this case, the inner rear wheel acts with increased slip relative to the curve. However, if the vehicle enters the bend too strongly, this is called overclocking, and the outer front wheel, with respect to the bend, operates at high slip rates.

In addition, pressure reduction on one of the front wheels can be prevented. This is done according to the following rules. In a driving position where the vehicle is under-controlled, a reduction in brake pressure relative to the curve on the outer front wheel • · *

- We'll stop it. In a situation where the vehicle behaves in an over-controlled position, the reduction of brake pressure relative to the curve is prevented by an inner front wheel.

The actual control of the brake pressure is as follows. As discussed above, the brake pressure in each wheel brake is determined depending on the deflection torque to be achieved and the weighted wheel coefficient coefficients.

In calculating the coefficients, a brake slip-dependent factor can be introduced, which is then adjusted to obtain the desired brake slip described above. Limiting the pressure drop on one wheel can be achieved by setting a lower threshold for the appropriate coefficient.

The procedure implemented in the control program of the brake system is described in more detail below.

The control program calculates, on the basis of weighted coefficients, the brake pressure to be generated for each wheel brake. The calculation becomes more problematic when the vehicle is braked, and especially when it is slowed down by utilizing the adhesion barrier between the tire and the driveway. In such cases, it is possible that the anti-lock regulation first occurs before a superimposed traction control is required.

In such cases, the considerations of the unbraked vehicle cannot be taken over, for example, when the pressure is increased on one of the wheels, the corresponding braking force increases non-linearly since the adhesive sealing limit has been reached. THE

therefore, the increase in pressure in this wheel brake does not generate additional braking force and thus no additional torque.

Although the same effect, that is, the creation of an additional deflection torque, can be obtained by reducing the brake pressure of the other wheels of a given axle, it would reduce the braking force altogether, contrary to the requirement to stop the vehicle in the shortest possible time.

Therefore, the behavior of the vehicle wheels shown in Figure 24 is utilized. This diagram shows the λ slip values on the X-axis between 0 and 100%. 0% for the free wheel,

100% corresponds to a blocked wheel. The friction coefficients μρ _{and the} lateral force values p _s are displayed on the Y-axis in the range of 0 to 1. The solid lines represent the dependence of the coefficient of friction on the slip at different angles of inclination. Particularly at small angles of inclination, it is recognized that the curve has a maximum around λ = 20% slip. Towards 100%, the coefficient of friction slightly decreases. 2 °, the maximum friction coefficient is approx. 0.98 and 0.93 for λ = 100%. On the other hand, when looking at the lateral force values, they are greatly reduced in the slip range, especially at higher angles of inclination. With a 10 ° tilt angle and a 0% slip value, the lateral force is 0.85 and drops to 0.17 at a near 100% slip value.

From the curves in Fig. 24, it can thus be seen that, at slip values of 40 to 80%, relatively high braking forces and only slight lateral forces can be transmitted.

· ♦ • · · · · · · · · · · · · · · · · · · · · · · ···

- 72 This wheel behavior can be used to purposefully reduce the lateral force of a specific wheel on a vehicle. The wheel selection is shown in FIG. and 25b. The diagram below is described in more detail in connection with FIG.

25a. and 25b. FIG. 6A is a schematic representation of a vehicle in a right bend. Depending on the curve radius and the speed of the vehicle, the vehicle should rotate about its height axis, i.e., at a certain rate of deflection in a clockwise direction.

The vehicle, as already mentioned, has a deflection angle sensor. If the deflection angle ^ 'wess ^m ®rt deviates from the Ψ'soll to be achieved, an additional Mg torque shall be applied around the vehicle's height axis.

If the measured deflection angle deviates from that to be achieved without the vehicle rotating enough, this is called undercontrolled behavior. An additional torque has to be expressed, which is negatively counted in this situation. This causes the vehicle to turn around in a bend. This can be achieved in this case by increasing the brake pressure in the right-hand vehicle wheels.

However, if the vehicle is already braked by the driver, these wheels may already be transmitting the maximum braking power. If the rating electronics determines this, the pressure in the right rear wheel brake increases so that the wheel runs at 40 to 80% slip. Wheel 604 is therefore denoted by λ. As mentioned above, the page force is significant 9 9 «··· ··· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·· · · ··

- 73 decrease. Only small lateral forces develop on the right rear wheel, causing the vehicle to spin out to the left, so it starts to rotate clockwise. The side _{force is minimized} until the actual deflection angle ^ '^ ess is equal to the specified deflection angle Ψ' _3Ο 11 of the vehicle.

25b. Fig. 4A shows the position of an overdrive vehicle. The vehicle rotates more rapidly around its height axis than the calculated required yaw rate. In this case, we suggest reducing the lateral force on the left front 601 wheel. This is also done by controlling slip values of 40 to 80% on this wheel. The wheel 601 is therefore denoted by λ.

For each of these cases, the control program may include a subprogram that causes additional pressure reduction relative to the curve on the outer front wheel 601 for the undercontrol (Fig. 25a) or on the inner front wheel 602 for the curve on the outer control (Fig. 25b). These wheels are marked with Pmin. In the case of a left turn, the sides of the respective controls are reversed.

The pressure in each wheel can be controlled by determining a coefficient for each wheel which gives the relationship between the pressure change and the calculated additional Mg deflection torque.

These coefficients are the parameters describing the vehicle and the wheel brakes, as well as the quantities that are changing during the journey • ··· ··· ·························································· · · ··· ·· ··· · ···

- Its 74 functions. They are, in particular, the coefficient of friction μ of the steering angle δ and the road / tire pair (see also section 3.1). In addition to the above-mentioned control, the dependence on the slip of the respective wheel is introduced. The pressure drop at each wheel can be prevented by setting lower limits for the coefficients and replacing the calculated value of the coefficient with the minimum value when it falls below the minimum value.

Figure 26 depicts the corresponding algorithm. First, calculate the additional Mg deflection torque (factor 640 or program). From this torque, the brake force changes or brake pressure changes (641 factors or program sections) for each wheel are determined. The established brake pressures are compared with the Pth thresholds (Factor 642 or Rhombus), determined by the coefficient of friction between the road / tire pair. The thresholds p ^ h indicate whether it is possible to further increase the wheel pressure and simultaneously increase the braking force. If the pressures to be generated by the control are below these limits, the control shall be in accordance with paragraph 3.1. shall be carried out in accordance with the procedure referred to in point. If the calculated brake pressures are above these thresholds, the pressures are calculated using the scheme 644 shown above.

4. Priority switching

From the additional Mg deflection torque, calculate the pressures to be set in the wheel brakes by means of a distribution logic (Figure 3).

point).

• · · · · · · · · · · · · · · · · · · · · · · · · · · · ···

From these pressure values, a slave pressure control circuit generates and outputs control signals to the inlet and outlet valves. In this subordinate pressure control circuit, the actual tire pressure is brought into line with the calculated tire pressure.

If the control signals of other controllers (slip control 7, drive slip control 8, electronic brake force control 9) need to be included (point 1), these control signals must first be converted to pressure values based on the hydraulic model of the wheel brakes in the computer.

The pressure requirements of the deflection torque regulator 10 are then correlated with the pressure requirements of the slip control regulator and other regulators. This takes place in a priority circuit which determines which requirements should prevail and how much pressure should be applied to the pressure of the wheel brakes 5. The pressure controller 5 converts the pressures into switching times for the valves.

Instead of the prescribed pressures, the prescribed pressure changes can be added to the priority switching (see section 7).

In this case, the priority switch 3 outputs the pressure changes Δ p at the outlet according to the rule that the requirement for pressure reduction on one wheel is met with priority and the requirement for pressure to be maintained in one brake has priority over the requirement for pressure increase. , so it is based on the priority switching · · · · · · · · · · · · · · · ·

- 76 when processing certain requirements, the rule is that when there is a requirement for depressurization, the requirements for maintaining or increasing the pressure are ignored. In the same way, there is no increase in pressure if the requirement is to maintain the pressure.

5. Priority switching by directly comparing valve switching times

Alternatively, another method may be used. The distribution logic unit calculates the switching times of the valves directly from additional Mq deflection torque. Other regulators do the same. The valve switching times for the control of the deflection torque can thus be compared, for example, to the required valve switching times of the slip control. Unlike before, priority switching evaluates not different pressure requirements, but different valve switching times.

To generate valve switching times, the distributor logic unit first calculates the pressure changes to be set for each wheel brake.

The connected nonlinear actuator calculates switching times from the pressure changes to control each wheel brake.

For example, this non-linear control element may be a counter.

This counter converts the given pressure changes into stroke numbers. For this, the Tq cycle time is about 3-10 ·· »· · ♦

- 77 ··· • · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·. The maximum number of strokes per cycle time is a fixed quantity that is determined by the goodness of control to be achieved.

The calculated stroke rate determines how long a valve must be controlled within the cycle time.

Since there are usually two valves per wheel brake, one valve to control the flow of pressure fluid to the brake (inlet valve) and another valve to control the flow of pressure fluid from the brake (outlet valve), a total of eight signals should be generated.

These rhythm numbers are assigned to the priority switch, which takes the number of further regulators on additional channels.

Priority switching decides which controller should be prioritized, thus deciding which cycle number to take over for actual valve control.

The response of the vehicle to the braking forces generated by the application of the wheel brakes is the altered deflection angle. This is detected by the deflection angle regulator 10, which also establishes a new additional deflection torque.

Thus, no brake pressure is calculated or set anywhere in the control loop. Thus, the control algorithms do not require information about the wheel brake and, in particular, do not require information about the relationship between the volume of the wheel brakes and the resulting brake pressures.

One way of calculating the schedule times is illustrated in Figure 27.

* »• · • * ·

- 78 From the additional Mg deflection torque, the distributor logic unit 700 calculates the brake pressures that must occur in each wheel brake. Sections 3.1 and 3.2 describe how this happens. As a result of the calculation within the distributor logic unit, four pressures, pj_ - p ^, are available for a four-wheeled vehicle. These quantities must be converted to the switching times of the valves which control the introduction of pressure fluid into the wheel brakes (pressure increase) and the discharge of pressure fluid from the wheel brakes (pressure reduction). The valve switching times, as already mentioned, are not calculated from the absolute values of the given pressures, but from the change in the given pressures. Therefore, all p _n values (n = 1 ... 4) a

We enter 701 shift registers. For the first register location 702, enter the current value. The second register location 703 passes the previous value from the register location 702 so that the pressure requirement from the previous calculation cycle is placed here. This value is denoted by p _n *.

In the next step 705, the current pressure requirement p _{n is} read from the first register location 702. If this value is 0 or less than a minimum value, the program branches to loop 706, which is responsible for ensuring that sufficient pressure is exerted from the wheel brake so that the set pressure is 0. For this, the inlet valve closes and the outlet valve opens for at least one Tq cycle time.

If the required pressure value is above the minimum value, then the difference between the register value 702 and the register location 703 is formed. This is what happened in the 707 difference separator. · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · P · Estonian

- 79 nicks. The calculated differential pressure Δρ may be greater than or equal to 0. If greater than 0, increase the pressure in the given wheel brake. If less than 0, the pressure in the given wheel brake shall be reduced. In the case of pressure increase, the program goes through the right decision path 710. Taking into account the differential pressure to be set, the pressure requirement, or if there are adequate signals, the opening time Trans _e j_ _{n for} the inlet valve is calculated from the actual pressure in the wheel. The opening time A _{aus of} the outlet valve is set to zero. Conversely (decision path 711), if a reduction in pressure is required, the opening time At _e j_ _{n of} the inlet valve is set to zero, while the opening time of the outlet valve At _{aus is set} from the differential pressure required and the actual pressure in the brake.

There is usually a linear relationship between the opening time Trans and the desired pressure change Δρ.

Once this has been clarified, we do not count the opening times themselves, but the pace numbers. This diagram is explained in more detail in Figure 28. The calculations described above are always performed at constant intervals (Tg cycle time), and as a result of a calculation, control signals are recorded for the wheel brake valves in the following cycle. The duration of a Tg cycle is approx. 3 ms

·· ·· * · ·· • ···································································· · ·

- 80 Depending on how fine the control should be, we divide each Tg cycle time into N periods.

In the diagram of Fig. 28, the division is six steps. Switching times are not given to valves as time quantities, but as the number of cycles within a cycle in which the valve must be open. For example, if n = 3, the opening time is 1.5 ms, as shown in FIG.

If the required opening time is longer than the cycle time, then n is set to the respective maximum N (six in the example shown).

This calculation is done for each wheel brake, which is four times for a four-wheeled vehicle. The calculations can be performed in parallel or sequentially. As a result, eight values are obtained, four for inlet valves and four for outlet valves. These values are assigned to the modified priority switch 720. This 720 modified priority switching includes slip control (blocking control) and switching time requirements for other controllers, also expressed in time.

This discharge is made so that there is a change in pressure in the wheel brakes, thereby changing the braking forces and the torque applied to the vehicle. Thus, there is a change in the quantities describing the driving dynamics of the vehicle. These sensors are detected directly or indirectly by the sensors and re-calculated.

· · »· · ·»

81 This results in a modified torque requirement which is converted to new control signals for the valves as described above.

The calculation of the differential pressure to be set is based on the pressure requirements obtained in the previous calculation cycle. However, they do not need to be actually adjusted so that the actual pressures in the wheel brakes differ from the calculated pressure requirements. Therefore, in certain situations, the actual pressure in the wheel brakes must be balanced against the pressure requirements. This is most easily accomplished when the pressure requirement is zero, i.e., the distributor logic unit 700 requires a value equal to zero pressure in one of the wheel brakes. In this case, we do not discriminate from the previous value and derive no control signal from it, but in step 705 branch to loop 706 to calculate the switching times, which must ensure that there is actually zero pressure. This is done by adjusting the A _aus switching time of the outlet valve to at least the cycle time Tq.

It may also be necessary to provide appropriate information for the 720 modified priority switching so that preset values from other controllers do not superimpose on this time requirement, which must produce zero pressure on one of the brakes. In addition, this information may record that the pressure drop must occur over several cycle times, so that it is ensured that a full pressure drop is achieved.

·· · * »• ···· • · * · ···, · · · ·.

6. Detect tire pressure

The stability control pressure control result described in point 4 provides brake pressure values for the wheel brakes. These specified values must be implemented. One method is to measure the pressure in the wheel brakes and compare it to the values given. A normal pressure regulator regulates the tire pressure to the specified value. This procedure requires one pressure sensor per wheel brake, ie four pressure sensors for a four-wheeled vehicle.

Usually they are trying to use as few sensors as possible because of the costs involved. In addition, each sensor is an additional potential source of interference. It may be necessary to shut down the entire control system due to a sensor failure.

Therefore, we recommend the use of a rating system that derives a pressure value based on data from existing sensors that corresponds to the pressure in the wheel brakes. We recommend the following concept for this.

The pressure in each wheel brake, as mentioned, is controlled by two valves. The inlet valve controls the discharge of the pressure medium, while the outlet valve controls the discharge of the pressure medium.

Thus, the signals that a pressure regulator can send are control times that indicate how long a valve must be open or closed. The cycle time is divided into a fixed number of periods (cycles). Controlling times like this83 • tofe · ··· · can be specified as a number, which defines how many periods of time a valve must be open or closed.

The basic consideration is that these signals are not only applied to the wheel brakes, but are also calculated as a running artificial model. The real vehicle reacts to the applied brake pressures by setting a defined center of gravity v and setting the individual wheels (ΰ £ rpm).

The speed of the vehicle is not measured directly, but is also calculated from the wheel revolutions of each wheel in separate calculation steps. This is therefore referred to as the reference speed v ^ _e f.

Corresponding values can be replicated within a vehicle model.

By comparing the actual values of C £> i and VR _e f with the values calculated by vp> _e f or estimated by the vehicle model, a correction quantity can be determined for the pressure in each wheel brake. By correcting the amount, the pressure calculated by the hydraulic model can be modified so that the tire pressure can be better estimated.

The conceptual structure just described will be explained in more detail in connection with FIG.

Here, instead of the pressure controller 5 of Figure 1, the pressure controller 800 is used. The pressure controller 800 calculates control times for the wheel brake valves from the first value 801 for the pressure to be adjusted and a second value 802 for the estimated or measured pressure in the wheel brake. The control «fc«

· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · The vehicle is marked with 810. This is intended to illustrate that the vehicle is responding to forces exerted by the pressures set on the wheel brakes. The wheel speed of each wheel will then change.

The vehicle 810 also needs to have wheel sensors that detect the needle wheel speeds so that the o> i values are directly available.

The vehicle 810 also has a rating unit, usually a part of an anti-lock controller. This forms a reference speed, v _re f, of the wheel speeds of each wheel under defined rim conditions, which must correspond to the actual speed of the vehicle.

A slip can be calculated for each wheel from each wheel speed and from the vehicle reference speed.

The values a> i and VR _e f are available as starting values. The slip is available as a value of 812.

The total model used, the 820 total model, contains three sub-models. These are the 821 hydraulic model, the 822 vehicle model, the 823 tire model.

Hydraulic model 821 describes, by means of two approximate formulas, the relationship between brake pressure p and volume V enclosed in the wheel brake, as well as the change in volume Δν when the inlet valve or outlet valve is open for a period of time.

F 6.1 p = a * V + b * V ²

F 6.2

Δν = + - c * t _e i _n / _aus * Δρ

Parameters a, b and c are quantities describing the braking system and are stored as values in appropriate stores, p describes the current brake pressure. V is the current volume enclosed in the wheel brake.

Δρ is measured either on the inlet valve or on the outlet valve.

When measuring at the inlet valve, a difference between a pressure source and p is observed, while at the outlet valve, a difference is found between p and a pressure in the container. The pressure in the container is generally around 1 bar and is thus negligible.

Assuming that the pressure in the wheel brakes and the closed volume is set to 0 at the start of an adjustment, the volume change will be followed by the valve opening times ·· · ·

- 86 and thus the pressure change in each wheel brake can be adjusted.

In any case, it is clear that the given formulas can only reproduce the actual conditions very roughly, so proper correction is needed. The vehicle in Vehicle Model 822 is generally described as having a rigid body which is at a single plane at the four abutment points (wheel surface).

The body can move parallel to the plane, ie in the x and y directions, and rotate about its center of gravity. The axis of rotation is perpendicular to the plane of motion.

The forces acting on the body are the braking forces acting on the contact surfaces of the wheels and the air drag forces.

The wheel loads Fz, v and F _z , h shall be based on these considerations:

F 6.3a ♦ M (-F _xv - F _xh ) - h ^ v _ref z, v

F 6.3b · h - (- F _x , _v - F _x , „) mgl„ - h-ni.v _ref

A'A

A A

Such a model is generally sufficient to perform the desired pressure correction. The model can, of course, be refined if necessary. For further calculations, the model essentially provides F _x loads of the bearing surfaces as a function of the deceleration of the center of gravity. The wheel is considered to be a rotatable disc having a certain moment of inertia.

F 6.4

K, * F raa x Br θ

The deceleration torques acting on the wheel are determined linearly from the wheel brake pressure.

F 6.5 ^M Br ^{= c} Br * P

In the tire model, it is assumed that the force utilization f, i.e. the ratio of the brake force to the wheel load, changes linearly with the wheel slip.

F 6.6

F _x λ * F _z

The given equations allow calculating the wheel speed of each wheel and the reference speed of the vehicle model.

These values can be compared with actual 811 baseline values. This is at 830 comparison points. From the difference between the measured and estimated wheel speeds of each wheel, an additional volume can be calculated by taking into account a correction factor k.

This additional volume Δν is added to the calculated nominal volume and the new nominal volume is obtained, from which a wheel pressure can be deduced from formula F 6.1 which is relatively exactly the actual tire pressure.

Of course, the accuracy of the estimate depends on the correction factor k, which may need to be determined experimentally.

This factor will vary from vehicle to vehicle and will depend, among other things, on how well the vehicle model reflects actual conditions.

The additional volume may also include a tolerance volume. This means that the volume through the valves is not proportional to the switching times. When the valve is opened and closed, the valve cross-section is only slowly expanded or narrowed, so that there is only a reduced volume in those periods where the entire cross-section of the valve has not yet been created or eliminated.

7. Replacement of a deflection angle gauge

In the above described control, the deflection angle velocity is a particularly important quantity because it serves as a control characteristic whose deviation ΔΨ 'must be minimized. Preferred ·· · ·

However, other regulatory features may be used, which are described below. For the sake of simplicity, the following notations are used here:

ifc

Ψ 'j4 _ess = gi as the actual value of the deflection angle,

Ψ '' Mess ⁼ 9 * 1 ' ^{as the} actual value of ^the deflection acceleration, άΨ''Mess / ht = g''j as the measured value of the deflection change (second order acceleration).

The same applies to the set values of Fig. 9, which are indexed by s.

The measured deflection angle velocity is determined in FIG. 12 in a conventional manner by a deflection angle detector 321 which outputs the output signal gj. However, such known deflection angle sensors, which directly transmit the deflection angle velocity, are very complicated to construct and therefore very expensive. The same applies to the comparators that are connected to them and to the controllers for the control circuit. That's why we strive to help with this and suggest simpler sensors and a simpler controller.

Figure 13 schematically illustrates a novel deflection angle sensor 321, which is a first cross-accelerometer 322 and a second 323 cross-accelerometer.

consists of a cross-accelerometer. The two transverse accelerometers 322 and 323 are located on the longitudinal axis of the vehicle, above the front and rear axles respectively. In principle, cross-accelerometers, wherever they may be located outside the center of gravity of the SP, require only proper conversion. Figure 15 shows a rectangular vehicle line 324 with tires 325 and sensors. According to this arrangement, the front cross-accelerometer 322 measures the cross-acceleration aq _{V at} the height of the front axle 326 and the rear cross-accelerometer 323 measures the cross-acceleration aqh at the height of the rear axle 327.

The two cross-accelerometers can deliver a quantity that depends on the deflection angle. From mathematical derivations it can be shown that from the results measured by the cross- _{accelerometers} the deflection angle and the cross-acceleration aq _{Uer of} the center of gravity SP can be determined as follows:

F 7.1 a a

Ψ = ^{gh gv} ~ 1. + 1 bv

F 7.2 quer. * 2 ♦ a ♦ 1.

qh v qv hk * ^J v

Here, as shown in Figure 13, l _v , 1 is the distance of the accelerometer 322, 323 from the center of gravity SP, v is the speed of the vehicle and β is the drift angle. Thus, from the cross accelerations and the distance of the cross accelerometer 322, 323, the deflection angle acceleration g'i can be calculated. Therefore, we recommend replacing the deflection angle acceleration g'i instead of the deflection angle rate suggested in the previous paragraphs. It is also possible for each input value to be linearly weighted for the comparison unit, as is known in the art. Then, the deflection angular velocity g and the drift angle β are calculated from the deflection angle pressure g '[angular acceleration?] And the deflection angular velocity β' by means of a band-wide integration or a first-order scaled low-pass filter to obtain the deflection angles 302 with the dimension of outgoing quantities of the vehicle reference model (section 2.3.1).

For bandwidth integration, it applies to

F 7.3

while applying a low pass filter gives the following relationship:

F 7.4

G (zp r _x * (1-λ)

- λ * ζ ' ¹ «·«

The angle of drift is a

F 7.5 is obtained after evaluating _q = ν * (Ψ '+ β').

Thus, it can be seen that two cross-accelerometers can be used to replace a known deflection angle velocity. However, the measures just described must be implemented to convert the deflection angle acceleration to the deflection angle velocity. After the training of Ag and Ag ', the 16 programs (regulatory act) can be connected. In Fig. 14, the thus calculated Mg deflection torque in the program 16 is additionally converted to the torque change M by time derivation.

However, under certain circumstances, it is preferable to switch to the nonlinear control of Figure 17, in which the deflection angle acceleration g ', both as an actual value and a target value, is given to the comparator unit 303 as a result of vehicle reference model 302. For this purpose, appropriate derivatives must be formed in the vehicle reference model.

As a result, at the output of the comparator unit 303, instead of the deflection angle difference, Ag, the deflection angle acceleration, Ag ', is changed, and as input, this is transmitted to program 16. The 16 programs (the deflection torque rule) ········································· ·

- 93), the drift angular velocity β 'can be additionally entered, as shown in Fig. 15, for a more precise definition of the torque change.

As already mentioned in FIG. 14, the additional deflection torque Mg can be omitted as the output signal of program 16 and instead the output signal M 'can be used as output signal. In a modified distribution logic unit, the torque change M ', i.e. the derivative of the additional Mg deflection torque, is converted to individual pressure changes. This means that the pressure changes are distributed to each wheel so that the desired additional Mg deflection torque is generated together. Details of this are described below with reference to Figure 16.

It should be taken into account that, at the same time, there is a defined pressure distribution in the wheel brakes through the driver's application of the brakes. In this case, it is more advantageous to determine the Mg deflection torque by integrating the torque change M '. From this, the differential pressure which can be generated by taking into account the pressure in each wheel brake can be directly determined. The advantageous enhancement described above can be combined with the distribution logic of paragraph 3 by using derivatives of the control features used in points 1 to 3. Thereby, two control principles are available, one providing an auxiliary deflection torque Mg and the other providing a change in the auxiliary deflection torque M 'as a predetermined (to be achieved) value. Switching between principles is possible. In particular, a change to another regulatory principle should be made if the former considers that the ancillary ··· ···

- The calculation of 94 control characteristics (drift angle, etc.) cannot be performed with sufficient precision (see, for example, section 2.2.2). It should also be noted that, as shown in Figure 15, Ag '[?] Can be added to the program 16 in addition to Ag' as a corrective amount.

In the program 16, in Fig. 15, the adapter has two threshold switches S2, S3 outside the amplifier k2, k3 which improve the control behavior within the program 16 and adapt the input quantities optimally to the ideal control behavior depending on the speed. Similarly, the ki, k2, k3 amplifier has a similar function. Each value is then summed up and output as the output signal of the deflection torque controller 10. General explanations for the 16 programs (regulatory acts), which are valid here, are provided in Section 2.4. .

Referring to Figure 1, it is shown how the pressures to be realized in a priority switch 3 at the output of the controller 7, 8, 9 are related to the pressures to be realized by a distribution logic unit 2. Applying the pressures to be realized presupposes a prior transformation in the equipment delivering these realizable values. The measures described below can simplify the exchange of information between program modules in the scope.

Figure 16 is a very simplified representation of the thread stability control loop of Figures 9 and 14. The markings in the previous figures have been retained.

* · ·

The deflection torque regulator 10 of Fig. 1 is modified here so that the output has a change in the additional deflection torque Mq, which is applied to the distribution logic unit 2 along with the pressure distribution desired by the driver on the brakes. . Referring to Figure 12 for the calculation of M '.

The distribution logic unit 2 has a logic block 340 and a pressure gradient switch 341. The essential function of the logic block 340 is to ensure that, despite the action taken to control the stability of the vehicle, it does not brake more strongly in the vehicle as a whole by providing the driver's pressure signal at the input of the distributor logic unit. This is to prevent the driving stability regulation from triggering additional instability. Thus, if there is brake pressure on one wheel based on the driver's braking preference, and on the other hand, through the stability control device, one or two wheels must be inflated and the opposite wheels depressurized to achieve additional deflection torque, there may be conflicting requirements requirement for simultaneous pressure increase and pressure drop. For other wheels, it may be a requirement that the pressure increase not only as a result of the driver's braking desire but also as a result of stability control. The logic block then ensures that the pressure in the appropriate wheels is reduced first and then the brake pressure can be increased to a specific limit as desired by the driver. This ensures that the average braking force relative to all wheels ··· ·, taking into account the additional torque generated by the anti-lock braking system, is not greater than the driver wishes.

As already explained in section 3.2, a targeted increase in λ longitudinal slip on one wheel can be used to reduce lateral forces while retaining braking force in the longitudinal direction. In this way, a deflection torque can be exerted without reducing vehicle deceleration.

In the pressure gradient switch 341 of the distribution logic unit 2, the pressure changes ΔΡ _ΧΧ occurring on each xx wheel are _{calculated from a} predetermined constant d _xx and the torque change M '. The calculation also includes the difference between the PFahrer brake pressure required by the driver and the actual Pxxist brake pressure measured,

F 7.6

and xx e [vr, vl, hr, hl] and gi = proportionality factor.

- 97 The actual brake pressure P ist _xx or pressure gauge is taken up in the wheel in question, or calculated by the calculation model, which follows the pressure changes imposed on the wheel, and thus enter the occurrent pressure wheel (claim 6).

The calculated pressure requirements are given and evaluated for 3 priority switches (see section 4 above).

The above description assumes that pressure gradients were processed directly in the priority switching. However, this is not necessary. It is also possible that in the 3 priority switches we process the valve switching times (point 5). In this case, however, a valve switching timing 343 must be switched between the distribution logic unit 2 and the priority switching unit 3. Then the valve switching times are taken from the other 7, 8, 9 controllers. The priority switching then processes the valve switching times A that has been entered according to the appropriate scheme already described in section 4 for brake pressures. Output quantities for priority switching are valve switching times.

The required Transverse _xx pressure changes for each xx wheel are converted to APp valve switching times [?] Using the following equation:

F 7.7 ^s xx = Kr P _xx lst · ^Δ Ρχχ ·

Kr (p.) = XX ^r xxist according to Rule ¹

F 7.9 {pxxistJ

Here, Kr _xx [?] Is a gain factor which depends on the actual pressure of each wheel and when increasing the pressure,

F 7.8 ___1_

Dv * 21 * Ja ² + 4 * b * p. "♦ ./160-p a ^ 0 V ^ xxist y ^ xxist is calculated, while the pressure drop is _ 1_ ^DV ab * ^T 0 * / ^{32 + A} * ^b * Pxxist * yJPxxist, xx is again an index which is for each wheel indicates its location.

Claims (3)

PATENT CLAIMS 1. A system for controlling the deflection torque of a four-wheeled vehicle with a device for detecting the current coefficient of friction in cornering, which detects the coefficient of friction only when the vehicle is in cornering, wherein the system provides at least constant control of the coefficient of friction is lysed and calculated according to the following formula: . fa. f. lAr "int flame where μ is the coefficient of internal friction, aq _{Uer is} the measured transverse acceleration, aiong is ^the measured or calculated longitudinal acceleration of ^the vehicle, g is the gravitational constant, and 1 <f. System according to claim 1, characterized in that the deflection torque control device functions as a parameter with a coefficient of friction μ> = 1 until needed. it appears a regulatory intervention with this parameter that friction coefficient detection occurs only then and 100 that this result replaces 1 as a parameter to examine the need for regulatory intervention. A system according to claim 2, characterized in that, if the friction coefficient (μ ^Λ ) detection results in a finding that no regulatory intervention is required, this result is updated several times, and that no further friction coefficient detection occurs. if, after applying a certain number of tests, regulatory intervention still does not appear to be necessary.