JP3320823B2 - Vehicle control device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement in a vehicle control device for controlling the motion of a vehicle in a turning state or the like.

[0002]

2. Description of the Related Art Conventionally, as a control device for a vehicle of this type, for example, when a driver operates a steering wheel, a steering ratio of a rear wheel corresponding to a steering angle of a front wheel is determined according to a vehicle speed. It is known to control the rear wheels in accordance with the steering of the front wheels by steering. However, in this type, it is possible to obtain a steering performance that matches the driver's intention, but immediately after the driver operates the steering. In the initial state, since the front wheels and the rear wheels often have the same phase, the turning performance of the vehicle in the initial state is low.

For this reason, conventionally, for example, Japanese Unexamined Patent Publication No.
In the device disclosed in Japanese Patent Publication No. 268, a control target yaw rate of a vehicle is calculated based on a steering amount of a driver, and a yaw rate of the vehicle is measured, and a yaw rate of the vehicle is measured according to a deviation between the measured value of the yaw rate and the control target value. By performing feedback control of the steering angle of the rear wheels using the feedback control amount, the yaw rate is quickly generated even in the initial state immediately after the steering operation, and the turning performance of the vehicle in this initial state is enhanced. However, in this conventional feedback control, a one-input, one-output control system that calculates a feedback control amount of a rear wheel steering angle only in accordance with a deviation between an actual measured value of yaw rate and a control target value and performs steering control of rear wheels. In some relations, in rear wheel steering corresponding to only the yaw rate deviation, for example, when the side slip angle of the vehicle is large, control hunting is more likely to occur than when it is small, and therefore control hunting is less likely to occur. As a result, it is necessary to always set the control gain to a small value.As a result, the response to the control target yaw rate value is not high, and the actual yaw rate cannot be accurately controlled to the control target value. Even had limitations.

[0004]

Therefore, the present applicant has
Previously, for example, in the specification and drawings such as Japanese Patent Application No. 4-23737, it has been proposed to employ state feedback control of actually measured yaw rate instead of the above-mentioned feedback control. The yaw rate state feedback control measures a vehicle's yaw rate and calculates a plurality of state variables of the vehicle, for example, the sideslip angles of the wheels or the vehicle, and the cornering forces of the front and rear wheels, based on the vehicle's state equation and output equation. A control system that optimally controls, for example, a rear wheel steering angle so as to make the actual yaw rate of the vehicle a control target value using the plurality of state variables by estimating the motion state of the vehicle; Since it is a control system, it is possible to control the steering of the rear wheels and the like with a feedback control amount that matches the motion state of the vehicle, and to improve the turning performance of the vehicle at the time of steering operation. It is possible to improve.

Furthermore, the present applicant has invented the following control in view of the following drawbacks in the above state feedback control. That is, in the state feedback control, the vehicle speed and the cornering power of the wheels are calculated as constants when estimating the sideslip angle and the cornering force of the wheels or the vehicle, but these are actually variables, and the latter are actually variables. The cornering power changes according to the friction coefficient of the road surface and the air pressure of the tire. As a result, as shown in FIG. 12, the motion characteristics of the vehicle change according to the change in the friction coefficient of the road surface. Therefore, under the preset vehicle motion characteristics, control responsiveness and steady-state characteristics can be improved.However, when a large change occurs in the vehicle motion characteristics according to changes in vehicle speed, road surface friction coefficient, and the like, An error occurs in the estimation of the side slip angle of the vehicle and the like, and a deviation occurs in the optimal control. As a result, the responsiveness to the target yaw rate is reduced and the stability of the vehicle cannot be guaranteed.

[0006] Therefore, for example, H∞ control (for example, “Measurement and Measurement”, which is intended to improve the stability of a control system by satisfying that the loop transfer function of the control system is “1” or less over all frequency ranges). Vol. 29, No. 2 (1990
Focusing on pages 111 to 119 of the month), it is conceivable to improve stability against fluctuations in vehicle characteristics, that is, to improve robust stability. In the H∞ control for the front wheel or rear wheel steering, steady characteristics and quick response of the vehicle are required in a low frequency range of, for example, 1 Hz or less as a frequency of the steering operation, and in a high frequency range of more than 1 Hz, the stability of the vehicle at the time of the characteristic fluctuation is stable. Is required, and these requirements differ in the frequency domain, so it is considered as a mixed sensitivity problem, as disclosed in the journal of the Society, and the control gain of the frequency transfer function of the yaw rate change of the vehicle with respect to the steering of the front wheels or the rear wheels. Two types of characteristics are set: a performance target index that gives the responsiveness and steady-state characteristics of the vehicle, and a performance target index that gives stability when the characteristics of the vehicle have a complementary sensitivity function. The control gain is determined on the frequency axis with emphasis on the former in the low frequency range and the latter in the high frequency range. Therefore, if the H∞ control is used, even if there is a change in the characteristics of the vehicle, if the change is within the set allowable range, all of the speed responsiveness, the steady performance and the stability of the vehicle can always be satisfactorily controlled. .

However, in the yaw rate state feedback control and the H∞ control proposed by the applicant, in any of the controls, the vehicle state equation in which the cornering force of the wheel with respect to the side slip angle of the wheel is established in a linear operation region is used. Because of this, the range over which the wheels that directly control the motion of the vehicle or the sideslip angle of the vehicle can be accurately estimated as the state quantity of the vehicle is narrow, and therefore control such as H∞ control cannot always be applied. There is a limit to guarantee the safety of
There is a regret that H∞ control cannot be used beyond this limit.

Therefore, for example, it is conceivable to actually measure the sideslip angle of a wheel or a vehicle. In this idea, it is necessary to use a non-contact type speedometer and a lateral accelerometer on a road surface. Because of the optical system, it may not be possible to measure the sideslip angle depending on the road surface conditions, and since these instruments are expensive, it is practically difficult to mount them on a real vehicle and put them into practical use in terms of price. It is.

The present invention has been made in view of the above point, and its object is not limited to the above-described yaw rate state feedback control and H∞ control, but widely estimates the side slip angle of a wheel or a vehicle. When controlling the motion of a vehicle using the estimated sideslip angle, the sideslip angle of the wheels or the vehicle is accurately estimated , and the sideslip angle is accurately estimated.
Is to be maintained for a long time .

[0010]

In order to achieve the above object, according to the present invention, the sideslip angle of a wheel or the like is estimated based on a neural network, and the estimation error is timely reduced to a zero value.
I will clear it.

That is, a specific solution of the invention according to claim 1 is to control a vehicle provided with a control means 45 for controlling the movement of the vehicle based on wheels or the sideslip angle of the vehicle as shown in FIG. Targets equipment. Then, the neural network is learned so that at least the steering angle, the vehicle speed, and the vehicle state quantity of the yaw rate are used as input signals, and the measured sideslip angle of the wheels or the vehicle is used as a teacher input to match the wheelslip angle of the wheels or the vehicle with the measured sideslip angle. Slip angle estimating means 4 having
4 is provided, the estimated sideslip angle estimated by the lateral slip angle estimator 44 with a configuration that is reflected in the motion control of the vehicle by the control means 45, the side slip angle estimator 4
4 is configured to perform a discrete system operation for calculating the amount of change in the estimated sideslip angle , and
The integrated value of the estimated sideslip angle in the dispersion calculation is cleared to zero when the vehicle is traveling straight .

[0012]

According to the above-mentioned structure, according to the first aspect of the present invention,
When the steering angle, the vehicle speed, and the yaw rate are input to the sideslip angle estimating means 44 as a vehicle state quantity at the time of turning of the vehicle, for example, a neural network after learning provided therein estimates the sideslip angle of the wheels or the vehicle. Even when the driving state of the vehicle enters the non-linear region of the tire, the actual sideslip angle of the wheel or the vehicle is accurately estimated. As a result, irrespective of the linear region and the non-linear region of the tire, the control means 45 accurately controls the movement of the vehicle based on the side slip angle estimated with high accuracy.

In this case, the neural network of the sideslip angle estimating means 44 is configured to perform a discrete system operation.
Due to the fact that the change is added to the previous estimated slip angle and the current estimated slip angle is calculated, if there is an error in the change, this error is integrated to produce a bias in one direction. However, since the integral value is cleared to zero every time the vehicle goes straight on when the actual skid angle is zero, the skid angle can be accurately estimated over a long period of time .

[0014]

As described above, according to the vehicle control apparatus of the first aspect of the present invention, the sideslip angle of the wheels or the vehicle is accurately estimated using the neural network, and the estimated sideslip angle of the vehicle is estimated. If the Ru is reflected in the motion control, a discrete system calculates the estimated sideslip angle at the New <br/> Rarunetto, its
Since the integrated value of definitive to the discrete calculation estimated lateral slip angle is cleared to zero value for each vehicle straight, more accurate estimation of sideslip angle
It can perform well and has high accuracy of its side slip angle during running
Vehicle motion while maintaining a long-term estimate.
It can be controlled exactly as desired.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS.

FIG. 2 is a schematic plan view of a vehicle steering system according to the present invention, wherein 1 is a steering wheel, 2 is left and right front wheels, 3
Is a left and right rear wheel, 10 is a front wheel steering device that steers the left and right front wheels 2 and 2 by operating the steering 1, 20 is a left and right rear wheel 3 according to the steering of the front wheels 2 and 2 by the front wheel steering device 10, 3 is a rear wheel steering device that steers the steering wheel 3.

The front wheel steering device 10 has a relay rod 11 arranged in the width direction of the vehicle body, and both ends of the rod 11 are tie rods 12, 12 and a knuckle arm 1, respectively.
It is connected to the left and right front wheels 2, 2 via 3, 13.
The relay rod 11 is provided with a rack and pinion mechanism 14 for moving the relay rod 11 right and left in conjunction with the operation of the steering wheel 1. In this configuration, the front wheels 2, 2 are steered.

On the other hand, the rear wheel steering device 20 has a relay rod 21 disposed in the vehicle width direction in the same manner as described above, and both ends of the rod 21 are respectively connected via tie rods 22, 22 and knuckle arms 23, 23. And are connected to the left and right rear wheels 3,3. The relay rod 21 is provided with a centering spring 22 for urging the rod 21 to a neutral position, and a rack-and-pinion mechanism 23. The mechanism 23 includes a clutch 24, a speed reduction mechanism 25. ,
When the clutch 24 is engaged, the relay rod 21 is moved in the vehicle width direction via the rack-and-pinion mechanism 23 by rotating the motor 26 when the clutch 24 is engaged, so that the rear wheels 3 are moved to the motor 26. The steering is steered by an angle corresponding to the rotation amount.

The motor 26 is connected to the control unit 2
9 is driven and controlled. The control unit 29
As shown in FIG. 3, the actual yaw rate of the vehicle is set to the control target yaw rate based on at least the actual yaw rate of the vehicle and the state equation and the output equation of the vehicle in which the estimated side slip angle of the wheels or the vehicle is the state quantity of the vehicle. FIG. 5 shows the state feedback control means 46 for controlling the steering angle of the rear wheel 3 by the motor 26 based on the control flow of FIG. 4 so as to perform the feedback control, and the control gain of the state equation and the output equation of the state feedback control means 46 in FIG. A control gain calculating means 47 for calculating based on the control gain calculating flow shown is provided, and these two means 46, 47 constitute a control means 45.

Further, in the controller 29, a skid angle estimating means 4 for estimating a skid angle at the center of gravity of the vehicle is provided.
4 are provided. The side slip angle estimating means 44 has a neural network 50 shown in FIG. 6 as described later.

In FIG. 3, reference numeral 35 denotes a lateral acceleration sensor for detecting a lateral acceleration acting on the vehicle; 36, a yaw rate sensor for detecting an actual yaw rate acting on the vehicle;
Reference numeral 7 denotes a rear wheel steering angle sensor for detecting the steering angle of the rear wheel 3, and reference numeral 38 denotes a front wheel steering angle sensor for detecting the steering angle of the front wheel 2. The detection signals of these sensors 35 to 38 are transmitted to the control unit 29. Is input to the state feedback control means 46. Further, in the figure, reference numeral 39 denotes a vehicle speed sensor for detecting a vehicle speed, and the detection signal is input to the control gain calculating means 46. The yaw rate 36, the front wheel steering angle sensor 38, and the vehicle speed sensor 39 are input to the sideslip angle estimating means 44.

Next, the drive control of the motor 26 by the control unit 29 will be described with reference to the control flow of FIG. In the figure, in step S1, for example, 20 ms
Every time the control timing of each set cycle of ec is reached, the vehicle speed Vsp, the front wheel steering angle Fstg, and the rear wheel steering angle Rst are determined in step S2 based on the detection signals of the sensors 36 to 39.
g, the yaw rate yr generated at the time of turning of the vehicle, and the motion state amount of each vehicle of the lateral acceleration Yg acting on the vehicle are measured.

In step S3, a control target yaw rate yrt of the vehicle is set based on the following equation, and a yaw rate deviation en between the target yaw rate yrt and the actual yaw rate yr is calculated.

[0024]

(Equation 1)

Where A is a stability factor,
L is the wheelbase of the vehicle.

Thereafter, in step S4, the steering angle r of the rear wheel 3 is set to H
す る Determined based on the following vehicle state equation (1) and output equation (2) in control.

[0027]

(Equation 2)

Here, Xh is a state quantity corresponding to a plurality of state quantities of the vehicle (hereinafter, “state quantity” in the description of this specification).
The state quantity of the vehicle is, for example, the side slip angle of the vehicle, the steering angle of the rear wheel and its changing speed, the cornering force of the front and rear wheels, and the yaw rate acting on the vehicle. And the like are estimated by the above two equations.

Further, the control gains Acl, Bcl, Ccl and Dcl of the above-mentioned vehicle state equation (1) and output equation (2).
Is calculated and determined in advance based on the control gain determination flow shown in FIG.

Next, a control gain determination flow in the H∞ control shown in FIG. 5 will be described. First, in step Sa, a correction coefficient (weight) Wn for changing a performance target index W3 that gives stability when a characteristic of a vehicle fluctuates as a control gain characteristic of a frequency transfer function of a yaw rate change of the vehicle with respect to the steering of the rear wheel 3 is controlled. Is changed in accordance with the estimated sideslip angle β estimated by the sideslip angle estimation means 44. This change is performed by determining the correction coefficient Wn to a large value in accordance with an increase in the estimated side slip angle β, as shown in step Sa.

Then, in step Sb, a performance target index W3 for providing stability when the characteristics of the vehicle fluctuates is determined based on the following quadratic expression of Laplace variable s.

[0032]

(Equation 3)

Here, a and c are constants. Based on the above equation, as shown in FIG. 9, the performance target index W3 that provides stability when the characteristics of the vehicle fluctuates is changed to the low frequency side when the correction coefficient Wn is large, and the correction coefficient Wn is small. When the value is a value, it is changed to the high frequency side.

Subsequently, in step Sc, a performance target index W1 (provided that the vehicle has quick response and steady-state characteristics as control gain characteristics of the frequency transfer function of the yaw rate change of the vehicle with respect to the steering of the rear wheel 3) is shown in FIG. , Reciprocal representation) are designated as fixed values based on the following cubic expression of the Laplace variable s.

[0035]

(Equation 4)

Here, d, e, f, g, and h are constants.

In step Sd, the two performance target indices W3 and W1 are converted into state equations, respectively, and these two state equations and the vehicle state equation (3) and the output equation (4) shown below are calculated. By combining the four into one state equation, decomposing the combined state equation into two Riccati equations, and solving the equations as eigenvalue problems using a Hamilton matrix, the control gains Acl to Find Dcl.

[0038]

(Equation 5)

Here, in the above-mentioned vehicle state equation (3) and output equation (3), d ヨ ー / dt is the yaw rate of the vehicle, β is the sideslip angle at the vehicle center of gravity, Cf and Cr
Is the cornering force of the front and rear wheels, δr is the steering angle of the rear wheel, lf and lr are the distances between the front and rear axles and the center of gravity of the vehicle, I is the moment of inertia of the vehicle with respect to yaw rate, m is the body mass, and Vsp is the vehicle speed. , Gy are the lateral elastic coefficients of the wheels, and Kyf and Kyr are the cornering powers of the front and rear wheels, respectively. Here, the vehicle speed Vsp is the vehicle speed sensor 3
9, the actual vehicle speed detected is used, and the others are fixed values set in advance.

Thereafter, at step Se, the H∞ obtained above is obtained.
The control gain of the control is changed from the continuous control system to the digital control system by the bilinear conversion method, and the processing is terminated. The control gains Acl to Dcl may be determined by a method of sequentially reading values stored in a map in advance in accordance with the motion state of the vehicle, in addition to the above-described sequential calculation.

Next, shows the specific structure of the side slip angle estimator 44 in Fig. The side slip angle estimating means 44 shown in the figure has a neural network 50, which is composed of an input layer 51, an intermediate layer 52, and an output layer 53. There are six input layers 51, seven intermediate layers 52, and one output layer 53. In the input layer 51, the estimated sideslip angle β (k), the actual yaw rate yr (k), the front wheel steering angle Fstg (k), the rear wheel steering angle Rstg (k), the reciprocal 1 / Vsp (k) of the vehicle speed,
The reciprocal 1 / Vsp ² (k) of the square of the vehicle speed is input, and the output layer 53 outputs a change with respect to the estimated side slip angle (k).

That is, the change amount of the estimated side slip angle β is obtained by setting the control timing to Δt, as shown in the following equation, the immediately preceding estimated side slip angle β (k) and the immediately preceding front wheel steering angle Fstg (k).
And the estimated side slip angle β (k)
The neural network 50 itself is configured to perform a discrete system operation for calculating a change in the estimated sideslip angle β by inputting the above and the like as an input, and outputting the current estimated sideslip angle β (k + 1).

[0043]

(Equation 6)

Therefore, the output layer 5 of the neural network 50
3 is added with the previous estimated sideslip angle β (k).

The output functions f1, f2, and f3 of the three layers of the neural network 50 shown in FIG.
It is as follows.

[0046]

(Equation 7)

Further, each weight of the intermediate layer 52 and the output layer 53 is determined in advance based on a neural network 60 shown in FIG.
Using a predetermined evaluation function so that the estimated sideslip angle β matches the actual sideslip angle, the characteristic is designed and configured to match the estimated sideslip angle β with the actual sideslip angle by learning.

The neural network 60 shown in FIG.
1, an intermediate layer 62 and an output layer 63. The input layer 61 has an actual yaw rate yr, a front wheel steering angle Fstg, a rear wheel steering angle Rstg, and a vehicle speed measured when the vehicle travels on a typical curved road many times. reciprocal of vsp 1 / Vsp, with reciprocal 1 / Vsp ² of the square of the vehicle speed, and the vehicle state quantity of the friction coefficient μ of the road surface is large number entered in a given time series, the estimated lateral slip angle β and the output layer 63 An estimated yaw rate yr is output, and the estimated sideslip angle β and the estimated yaw rate yr are obtained by using the actual sideslip angle and the actual yaw rate of the vehicle center of gravity of the discrete time series measured together with the vehicle state quantity at the time of traveling on the curved road as teacher inputs. It has been identified to match the measured value. Here, the estimated yaw rate yr is estimated by:
This is for confirming that the measured actual yaw rate matches the estimated value yr and confirming that the system identification has been correctly performed.

The vehicle speed Vsp input to the neural network 50 shown in FIG.
The form of sp and its reciprocal 1 / Vsp ² is represented by the following equation for a linear motion model of a normal two-wheel steering vehicle, and the terms relating to the vehicle speed in the arithmetic equation are 1 / Vsp and 1
/ Vsp ² in order to match this arithmetic expression.

[0050]

(Equation 8)

Here, δf is the steering angle of the front wheels.

Next, the operation of the side slip angle estimating means 44 will be described with reference to the flow chart of FIG. In the same figure, every time the control timing comes in step S1, the actual yaw rate yr (k) and the front wheel steering angle Fstg in step S2.
(K), rear wheel steering angle Rstg (k), vehicle speed Vsp (k)
After measuring the motion state amount of the vehicle at step S3, the absolute value | Fstg | of the front wheel steering angle is compared with a small set value Fstgl at step S3, and the absolute value | yr | of the actual yaw rate is set at a small set value yrl at step S4. Compare. And |
When Fstg | <Fstgl and | yr | <yrl, it is determined that the vehicle is traveling straight, and only in this case, the estimated sideslip angle β (k) is set to β (k) = 0 in step S5, and the sideslip angle is set. Clear the integrator to zero.

Thereafter, in step S6, the vehicle state quantity such as the yaw rate yr measured above and the estimated side slip angle β
(K) is input to the neural network 50 to estimate the estimated sideslip angle β (k + 1). In step S7, the estimated sideslip angle β (k + 1) is controlled by the H∞ control in step Sa of the control gain determination flow of FIG. Then, the process returns to step S1 as the change of the correction coefficient Wn of the performance target index W3 that gives stability when the characteristic of the vehicle changes.

Therefore, the actual yaw rate y of the vehicle is determined by steps S3, S5 and S6 of the control flow shown in FIG.
The actual yaw rate yr of the vehicle is calculated based on the vehicle's state equation (1) and the output equation (2), where r, the estimated sideslip angle β of the vehicle, and the cornering forces Cf, Cr, etc. of the front and rear wheels, etc. The rear wheel steering device 20 is configured to perform state feedback control to the control target yaw rate yrt.
Is configured to control the state feedback control means 46 of FIG.

Further, according to the control gain determination flow of FIG. 5, a performance target index W3 for providing stability when the characteristic of the vehicle fluctuates as a control gain characteristic of the frequency transfer function of the yaw rate change of the vehicle with respect to the steering of the rear wheel 3 and the vehicle. Of the vehicle based on two types of performance target indices W1 that give the responsiveness and steady-state characteristics of the vehicle, and the state equation (3) and the output equation (4) of the vehicle that receive the steering angle δr of the rear wheel 3 as an input. The characteristic (control gain characteristic) is set at a set frequency (0.3 rad / s in FIG. 9) at which the loop transfer function becomes 0 db as shown by the solid line in FIG.
ec), it is located immediately below the performance target index W3 that provides the above stability, and is set at the set frequency (0.3r
(ad / sec) or less, the control gain Acl of the state feedback control means 46 is located just above the performance target index W1 that gives the steady-state characteristics and the like.
ＤDcl is configured in the control gain calculating means 47 of FIG. Then, in step Sa of the control gain determination flow of FIG. 5 and step S5 of FIG. 4, the estimated side slip angle β (k + 1) estimated by the side slip angle estimation unit 44 is changed by the control unit 45 to the rear wheel steering at the time of turning the vehicle. The H∞ control is used to reflect the change in the performance target index W3.

Therefore, in the above embodiment, the side slip angle estimating means 44 includes the neural network 5 designed by the neural network 60 in which the vehicle system is identified.
0, and the sideslip angle at the vehicle center of gravity is estimated based on the neural network 50, so that the estimated sideslip angle β accurately matches the actual sideslip angle at the vehicle center of gravity. As a result, the performance target index W3 that gives stability when the characteristics of the vehicle fluctuates in the H∞ control of the rear wheel steering is changed based on the estimated sideslip angle β that is accurately estimated. As shown by the solid line in FIG. 9, the set frequency (0.3 r
ad / sec), it is located directly below the performance target index W3 for providing stability, and is 0.3 rad / s.
In a low frequency range equal to or lower than ec, it is located immediately above the performance target index W1 that gives a steady characteristic or the like. As a result, the control gain becomes small in a high frequency range exceeding 0.3 rad / sec with respect to the motion characteristics shown in FIG. In the low frequency range, the control gain becomes a large value. As a result, in the high frequency range, the stability against the characteristic fluctuation of the vehicle is improved and the robust stability is improved due to the small control gain.

In particular, the neural network 60 of FIG. 7 for designing the neural network 50 of the side slip angle estimating means 44 is as follows.
The friction coefficient μ of the road surface is input as an input signal, and the vehicle system is identified based on the friction coefficient μ.
When estimating the sideslip angle based on the following formula, the estimated sideslip angle β is output with a good assumption of the friction coefficient without measuring the friction coefficient of the road surface, so that the estimated sideslip angle β can be accurately estimated. .

In addition, the input of the vehicle speed Vsp to the neural network 50 of the side slip angle estimator 44 is based on the vehicle speed reciprocal 1 / Vsp and the vehicle speed two reciprocal 1 / Since the calculation is performed in the form of Vsp ² , the calculation speed can be increased.

Further, since the neural net 50 of the side slip angle estimator 44 estimates the estimated side slip angle β by a discrete system operation, the estimated side slip angle β can be calculated more accurately. In this case, even if the integrated value of the estimated side slip angle β includes an error, the integrated value is cleared to zero every time it is determined that the vehicle is going straight ahead according to the front wheel steering angle Fstg and the yaw rate yr. Therefore, the sideslip angle at the center of gravity of the vehicle can always be accurately estimated when the vehicle is running.

In this embodiment, the sideslip angle at the center of gravity of the vehicle is estimated. However, the sideslip angle of each wheel may be estimated instead.

FIG. 10 shows another embodiment. In the above embodiment, the front wheels 2, 2 are electrically controlled separately from the steering instead of performing the steering control of the rear wheels 3, 3 using the rear wheel steering device 20. This is applied to a system that performs steering control.

That is, the steering system shown in FIG. 10 does not include the rear wheel steering system 20 shown in FIG.
And a rack and pinion mechanism 50 disposed on the relay rod 11 in parallel with the motor 5 for driving the mechanism 50.
1 and the motor 51 is connected to the control unit 29.
The driving is controlled by the Other configurations are the same as those of the above-described embodiment. However, in the present embodiment, when the rear wheels are steered and the rear wheels are steered in the opposite phase to the front wheels, the front wheels are steered. When the steering control is performed to increase the angle and the rear wheels are controlled to have the same phase in the above embodiment, the steering control may be performed to decrease the steering angle of the front wheels in the present embodiment.

In the above description, in the state feedback control of the rear wheel steering angle, the estimated observation amount of the vehicle is 6%.
The state of the vehicle was accurately observed using the seeds, namely, the sideslip angle of the wheel, the steering angle of the rear wheel and its changing speed, the cornering force of the front and rear wheels, and the yaw rate acting on the vehicle. To do so, it is sufficient to observe at least two types: the actual yaw rate of the vehicle and the estimated sideslip angle of the wheels or the vehicle.

Further, in the above description, the control means 45 is configured to perform the H∞ control for the rear wheel steering. However, the present invention is not limited to this. The present invention may be applied to a method of estimating and controlling the motion of a vehicle, for example, a suspension device, using the estimated sideslip angle.

[Brief description of the drawings]

FIG. 1 is a block diagram of the first embodiment of the present invention;

FIG. 2 is a diagram illustrating an overall configuration of a steering device that also steers a rear wheel of a vehicle.

FIG. 3 is a diagram showing a block configuration of steering control of a rear wheel.

FIG. 4 is a diagram illustrating a control flowchart of steering control of a rear wheel.

FIG. 5 is a flowchart for determining a control gain of H∞ control.

FIG. 6 is a configuration diagram of a neural network provided in a side slip angle estimator.

FIG. 7 is a configuration diagram of a neural network for vehicle system identification.

FIG. 8 is a flowchart showing the operation of the side slip angle estimator.

FIG. 9 is a diagram showing performance target indexes W3 and W1 used for H∞ control.

FIG. 10 is a diagram showing an overall configuration of a steering device that steers a front wheel.

FIG. 11 is a diagram illustrating gain characteristics of a frequency transfer function of a vehicle motion characteristic with respect to rear wheel steering when H∞ control is not performed.

FIG. 12 is an explanatory diagram illustrating how a gain characteristic of a frequency transfer function of a vehicle motion characteristic with respect to rear wheel steering changes when a friction coefficient of a road surface changes.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Steering 3 Rear wheel (wheel) 20 Rear wheel steering device 26 Motor 29 Control unit 44 Side slip angle estimation means 45 Control means 46 State feedback control means 47 Control gain calculation means 50 Neural net 60 Neural net for vehicle system identification

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁷ identification code FI B62D 137: 00 B62D 137: 00 (58) Investigated field (Int.Cl. ⁷ , DB name) B62D 6/00-6/06 B62D 7/14

Claims (1)

(57) [Claims] 1. A control device for a vehicle comprising control means for controlling the motion of a vehicle based on wheels or sideslip angle of the vehicle, wherein at least a steering state angle, a vehicle speed, and a vehicle state quantity such as a yaw rate are input signals. A skid angle estimating means having a neural net learned so that the measured skid angle of the wheel or the vehicle matches the measured skid angle using the measured skid angle of the wheel or the vehicle as a teacher input, and provided by the skid angle estimating means. estimated side-slip angle while being reflected in the motion control of the vehicle by the control means, the neural network provided in the side slip angle estimating means estimated
Performs a discrete system operation to calculate the amount of change in the constant sideslip angle.
When the integrated value of the estimated sideslip angle in the discrete system operation is
A vehicle control device characterized by being cleared to zero value .