JPH11147460A - Estimation system for coefficient of friction on road for vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To continually and accurately estimate the coefficient of friction on a road. SOLUTION: An electric control unit (microcomputer) 20 calculates the coefficient of friction μ on a road based on the following quantities of the state of a vehicle by the neutral network calculation; a brake hydraulic pressure BP and a throttle angle SL corresponding to the quantity of the state of vehicle operation which affects the longitudinal movement of the vehicle, a vehicle speed V and a longitudinal acceleration Gx corresponding to the quantity of the state of longitudinal movement of the vehicle, the steering angle θ corresponding to the quantity of the state of vehicle operation which affect the lateral and turning movements of the vehicle, the lateral acceleration Gy corresponding to the quantity of the state of lateral movement of the vehicle, and the yaw rate γcorresponding to the quantity of the state of turning movement of the vehicle or the like.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a road surface friction coefficient estimating apparatus for estimating a road surface friction coefficient of a running road surface based on various state quantities of a vehicle.

[0002]

2. Description of the Related Art Conventionally, this type of apparatus uses various state quantities of a vehicle such as a vehicle speed, a steering angle, a steering torque, a brake pressure, a wheel rotation speed, a wheel drive, a yaw rate, a longitudinal acceleration and a lateral acceleration. There is known an apparatus which calculates a road surface friction coefficient according to a fuzzy inference rule (for example, Japanese Patent Application Laid-Open No. 5-124529). Also,
There is a vehicle in which a road surface friction coefficient is calculated from a vehicle acceleration / deceleration and a slip ratio by a functional expression (for example, Japanese Patent Application Laid-Open No.
No. 16123).

[0003]

However, in the former prior art, since the road surface friction coefficient has extremely high non-linearity, it is difficult to set a membership function in fuzzy inference, and the road surface friction coefficient is accurately estimated. It is difficult. In addition, in fuzzy inference, the road friction coefficient
Although it is possible to estimate as "small", it is difficult to estimate as a continuous amount. Further, in the latter prior art, the non-linearity of the tire that occurs when the vehicle is accelerated or decelerated by depressing the accelerator pedal and the brake pedal or when the vehicle is turned by steering the steering wheel is not considered, and the road surface friction coefficient is not considered. It cannot be estimated accurately.

[0004]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a vehicle road surface friction coefficient estimating apparatus which continuously and accurately estimates a road surface friction coefficient. .

The structural features of the present invention include a plurality of sensors for respectively detecting the state quantities of a plurality of types of vehicles, and a parameter representing the relationship between the state quantities of the plurality of types of vehicles and the road surface friction coefficient. A neural network means for calculating a road surface friction coefficient by a neural network operation based on a plurality of types of vehicle state quantities detected by a plurality of sensors. In this case, as the state quantity of the vehicle, a set of a driving operation state quantity representing a driving operation of the vehicle and a movement state quantity representing a movement state of the vehicle in the front-rear direction that causes a change in the movement of the vehicle in the front-rear direction, A set of a driving operation state quantity representing a driving operation of the vehicle that causes a change in the lateral movement of the vehicle and a movement state quantity representing a lateral movement state of the vehicle may be employed. A set of a driving operation state quantity representing a driving operation and a rotational motion state quantity of the vehicle may be adopted.

As described above, according to the present invention, the road surface friction coefficient is calculated based on the state quantities of a plurality of vehicles by neural network operation. It becomes possible to accurately estimate the road surface friction coefficient that changes in a timely manner. In particular, the road surface friction coefficient is closely related to the longitudinal and lateral movements of the vehicle and the rotational movement of the vehicle, and according to the present invention, the amount of driving operation that changes each of these movements is the same as the movement amount. Is adopted as the state quantity of the vehicle, so that the road surface friction coefficient of the traveling road surface of the vehicle can be accurately estimated.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows a road friction coefficient estimating apparatus according to the embodiment and a vehicle motion utilizing a road friction coefficient estimated by the estimating apparatus. A control device for controlling the state is schematically shown in a block diagram.

The road surface friction coefficient estimating apparatus estimates and calculates a road surface friction coefficient μ based on a plurality of state quantity sensors 10 for detecting various state quantities of the vehicle and various state quantities detected by the state quantity sensors 10. And an electronic control unit 20 to perform the control. The plurality of state quantity sensors 10 include a wheel rotation speed sensor 11, a longitudinal acceleration sensor 12, a lateral acceleration sensor 13, a steering angle sensor 14, and a brake oil pressure sensor 1.
5, throttle opening sensor 16, yaw rate sensor 17
And so on.

The wheel rotation speed sensor 11 detects a wheel rotation speed Vw. It should be noted that the wheel rotational speed Vw is determined by executing a program to be described later.
And used as one of the state quantities of the vehicle. The longitudinal acceleration sensor 12 and the lateral acceleration sensor 13 detect accelerations Gx and Gy in the longitudinal and lateral directions of the vehicle, respectively. The steering angle sensor 14 detects the steering angle θ of the front wheels by measuring the steering angle of the steering wheel. The brake oil pressure sensor 15 detects the oil pressure BP of the brake oil supplied to the wheel cylinder by the depression operation of the brake pedal. The throttle opening sensor 16 detects the opening S of the throttle valve, which changes when the accelerator pedal is depressed.
L is detected. The yaw rate sensor 17 detects a yaw rate γ acting on the vehicle.

The brake oil pressure BP and the throttle opening SL of the vehicle state variables correspond to driving operation state variables representing the driving operation of the vehicle that causes a change in the longitudinal movement of the vehicle. The vehicle speed V and the longitudinal acceleration Gx correspond to a motion state quantity representing a motion state of the vehicle in the front-rear direction. The steering angle θ corresponds to a driving operation state quantity representing a driving operation of the vehicle that causes a change in the lateral motion of the vehicle and the rotational motion of the vehicle. The lateral acceleration Gy is equivalent to a motion state quantity representing a lateral motion state of the vehicle. The yaw rate γ corresponds to the rotational motion state quantity of the vehicle.

The electronic control unit 20 includes a CPU, an RO,
The microcomputer is constituted by a microcomputer including M, RAM, I / O, timer, etc., and repeatedly executes a program corresponding to the flowchart of FIG. 2 stored in the ROM to estimate a road surface friction coefficient μ by a neural network operation. And output.

Here, a neural network operation realized by executing the program will be described. FIG. 3 is a block diagram functionally showing a neural network NN realized by executing the program. The neural network NN includes a vehicle speed V, a longitudinal acceleration Gx, and a lateral acceleration G as state quantities of the vehicle.
y, steering angle θ, brake oil pressure BP, throttle opening S
L, yaw rate γ, etc., are input data IDi (i
= 1 to m) input units 21-1 to 21-m (m>)
7) and intermediate units 22-1 and 22-2
An intermediate layer 22 made of 2-n (n> 7) and an output layer 23 composed of a single output unit 23-1 are provided. Input data IDi (i = 1 to m), output values Xi (i = 1 to m) of each input unit 21-1 to 21-m, output values Yj (j = j) of each intermediate unit 22-1 to 22-n 1
To n) and the output value Z of the output unit 23-1 are defined as in the following equations (1) to (3).

[0013]

Xi = f (ω0i · IDi−θ0i)

[0014]

(Equation 2)

[0015]

(Equation 3)

Each of the coupling coefficients ω0i, ω1ij, ω2j and each of the threshold values θ0i, θ1j, θ2 are constants representing the results of prior learning, and the function f (X) is a function such as a step function or a sigmoid function. And these coupling coefficients ω0i, ω1ij, ω2
j, threshold values θ0i, θ1j, θ2, and function f (X) are stored in advance in the ROM of the microcomputer, and are collectively referred to as parameters in this specification. In addition,
In the present embodiment, the intermediate layer 22 is provided in only one stage, but may be provided in a plurality of stages, such as two, three, and the like.

Next, the setting of these parameters will be described. First, the road surface friction coefficient μ0 (for example,
0.1, 0.2, 0.3 ...) on various road surfaces in which the same type of vehicle as the vehicle on which the present friction coefficient estimating device is mounted is driven in various driving states to obtain various road surface friction coefficients μ0. , The vehicle speed V, the longitudinal acceleration Gx, the lateral acceleration Gy, the steering angle θ, the brake oil pressure BP, the throttle opening SL, and the yaw rate γ. FIG. 4 shows an example of the measurement result.

Next, various combinations of the measured vehicle state quantities are sequentially input to a neural network NN constructed in a software or hardware manner in the same manner as in this embodiment, and the road surface friction coefficient μ0 at each combination is instructed. Neural network NN given as data
Let them learn. However, in this neural network NN, each coupling coefficient ω0i,
ω1ij, ω2j and threshold values θ0i, θ1j, θ2 are stored in a writable memory device such as a RAM. Specifically, a set of input data IDi (i = 1
To m), the state quantities V, Gx, Gy, θ, B of the vehicle
Each time P, SL, γ... Are input to the input layer 21, the output layer 2
3 to obtain a road surface friction coefficient μ as an output value Z, a road surface friction coefficient μ0 (= 0.1, 0.2, 0.3...) Corresponding to the set of input data IDi as teacher data, and a road surface as the output value Z. The deviation from the friction coefficient μ is fed back to the neural network NN, and each coupling coefficient ω0 is determined by using the back propagation method so that the deviation is minimized.
i, ω1ij, ω2j and threshold values θ0i, θ1j, θ2 are sequentially updated. Then, all the parameters thus obtained in advance are set and stored in the ROM of the microcomputer.

The electronic control unit 20 is connected to an anti-lock brake control device 31, a vehicle stability control device 32, a brake assist control device 33, and the like as control devices for controlling the motion state of the vehicle. The antilock brake control device 31 avoids locking of the wheels during braking. The vehicle stability control device 32 avoids abnormal behavior of the vehicle (spin, drift-out, etc.). The brake assist control device 33 increases the brake oil pressure for the wheel cylinder in order to obtain a sufficient braking force when the brake pedal is suddenly depressed. Each of these control devices 31 to 31
33 uses the road surface friction coefficient μ estimated by the electronic control unit 20 for the control.

Next, the operation of the embodiment configured as described above will be described with reference to flowcharts. When the ignition switch (not shown) is turned on, the electronic control unit 20 starts executing the program in step 50 of FIG. 2 and repeatedly executes the program including steps 52 to 68. In step 52, the wheel rotation speed sensor 11, the longitudinal acceleration sensor 12, the lateral acceleration sensor 13, the steering angle sensor 14, the brake oil pressure sensor 15, the throttle opening sensor 16, the yaw rate sensor 17, and other vehicle state quantity sensors 10 are used to determine the wheels. Rotation speed Vw, longitudinal acceleration Gx, lateral acceleration Gy, steering angle θ, brake oil pressure B
P, throttle opening SL, yaw rate γ, etc. are input.

Next, at step 54, the input wheel rotational speed Vw is converted into an actual vehicle speed Vx, and at step 56, an estimated vehicle speed Vx 'is calculated according to the following equation (4).

[0022]

Vx ′ = Vxo + ∫Gxdt Equation (4) derives the vehicle speed V by adding the time integral value of the longitudinal acceleration Gx to the reference vehicle speed Vxo until the time elapsed from the reference vehicle speed Vxo. Means that. In the actual processing of step 56, a value obtained by multiplying the average value of the longitudinal accelerations Gx at the current processing timing and the previous processing timing by the time from the previous processing timing to the current processing timing is accumulated in the reference vehicle speed Vxo. . Note that the reference vehicle speed Vxo is initially set to “0” and is updated each time the process of step 62 described later is performed. The time from the previous processing timing to the current processing timing is measured by a timer built in the electronic control unit 20.

Next, at step 58, it is determined whether or not the wheels are slipping. In this determination, the actual vehicle speed Vx and the estimated vehicle speed V calculated in steps 54 and 56 are used.
It is determined whether or not the absolute value | Vx−Vx ′ | of the difference from x ′ is equal to or greater than a predetermined error ΔV. In this case, the absolute value | Vx−
If Vx ′ | is smaller than the predetermined error ΔV, “NO”, that is, it is determined that the wheel is not in the slip state, and step 6 is performed.
At 0, the vehicle speed V is set to the actual vehicle speed Vx. Step 6
After the processing of 0, the reference vehicle speed Vxo is updated to the actual vehicle speed Vx in step 62, and the program proceeds to step 66. On the other hand, if the absolute value | Vx−Vx ′ | is equal to or larger than the predetermined error ΔV, “YES”, that is, it is determined that the wheels are in the slip state, and the vehicle speed V is calculated at step 64 as the estimated vehicle speed V.
x 'and the program proceeds to step 66.
In detecting the slip state, the slip state may be determined in consideration of the yaw rate γ, the steering angle θ, and the like.

In step 66, a "neural network operation routine" is executed to calculate a road friction coefficient μ. That is, the calculated vehicle speed V and the input longitudinal acceleration Gx, lateral acceleration Gy, steering angle θ, brake oil pressure BP, throttle opening SL, yaw rate γ, and the like are input data IDi (i = 1 to m) in the ROM. By using the coupling coefficients ω0i, ω1ij, ω2j and the thresholds θ0i, θ1j, θ2 stored in the input layer 21 and the intermediate layer constituting the neural network operation, 22 and output values Xi (i =
1 to m), Yj (j = 1 to n), and Z (= μ). In addition, as the vehicle speed V, steps 58, 60, 64
The actual vehicle speed Vx is used when the wheel is not slipping, and the estimated vehicle speed Vx 'is used when the wheel is slipping.

Then, in step 68, the calculated output value Z, that is, the road surface friction coefficient μ is applied to the anti-lock brake control device 31, the vehicle stability control device 32,
Output to the brake assist control device 33, respectively. Each of these control devices 31 to 33 controls the avoidance of wheel lock, the avoidance of abnormal behavior of the vehicle (spin, drift-out, etc.), the increase in brake oil pressure, and the like using the road surface friction coefficient μ.

As described above, according to the above-described embodiment, the brake hydraulic pressure BP and the throttle opening as the driving operation state quantities representing the driving operation of the vehicle causing a change in the longitudinal movement of the vehicle by the neural network calculation. SL, a vehicle speed V and a longitudinal acceleration Gx as motion state quantities representing the longitudinal motion state of the vehicle, and a driving operation state quantity representing a driving operation of the vehicle which causes a change in the lateral motion of the vehicle and the rotational motion of the vehicle. The road surface friction coefficient μ is calculated based on the steering angle θ, the lateral acceleration Gy as a motion state quantity representing the lateral motion state of the vehicle, and the yaw rate γ as the rotational motion state quantity of the vehicle. Thus, the continuously changing road surface friction coefficient μ can be accurately estimated in consideration of various running states of the vehicle. In addition, the control using such an accurate estimated road surface friction coefficient μ can accurately control various vehicles.

[Brief description of the drawings]

FIG. 1 is a block diagram of a road surface friction coefficient estimating apparatus according to an embodiment of the present invention and a control device connected to the estimating apparatus.

FIG. 2 is a flowchart of a program executed by an electronic control unit (microcomputer) in FIG. 1;

FIG. 3 is a functional block diagram of a neural network operation.

FIG. 4 is a diagram showing measured values of various state variables and a road surface friction coefficient by an experiment.

[Explanation of symbols]

10 state sensor, 11 wheel speed sensor, 12
Longitudinal acceleration sensor, 13 lateral acceleration sensor, 14 steering angle sensor, 15 brake oil pressure sensor, 16 throttle opening sensor, 17 yaw rate sensor, 20 electronic control unit (microcomputer), 21 input layer,
22: middle layer, 23: output layer.

Claims (4)

[Claims] A plurality of sensors for respectively detecting state quantities of a plurality of types of vehicles, and a parameter representing a relationship between a state quantity of the plurality of types of vehicles and a road surface friction coefficient are stored. A vehicle road surface friction coefficient estimating apparatus comprising: a neural network means for calculating a road surface friction coefficient by a neural network operation based on the detected state quantities of a plurality of types of vehicles. 2. The state quantity of a plurality of types of vehicles according to claim 1, wherein the state quantity of a driving operation representing at least a driving operation of the vehicle causing a change in the movement of the vehicle in the front-rear direction, and the movement amount of the vehicle in the front-rear direction. A road surface friction coefficient estimating device for a vehicle, comprising: a motion state quantity representing a state. 3. The state quantity of a plurality of types of vehicles according to claim 1, wherein the state quantity of the driving operation represents at least a driving operation of the vehicle that causes a change in the lateral movement of the vehicle, and the lateral movement of the vehicle. A road surface friction coefficient estimating device for a vehicle, comprising: a motion state quantity representing a state. 4. The state quantity of a plurality of types of vehicles according to claim 1, wherein at least a driving operation state quantity representing a driving operation of the vehicle that causes a change in the rotational movement of the vehicle and a rotational movement state quantity of the vehicle. A road surface friction coefficient estimating apparatus for a vehicle, comprising: