JP4251095B2 - Vehicle control device - Google Patents

Description

  The present invention relates to a vehicle control apparatus suitable for use in, for example, auto cruise control.

Auto-cruise control in automobiles is constant-speed traveling control that realizes constant-speed traveling according to the set vehicle speed. However, when there is no preceding vehicle ahead of the host vehicle, the above-mentioned constant-speed traveling control is performed and within a predetermined distance range. Some vehicles also include an inter-vehicle distance control that controls the vehicle speed so as to keep the inter-vehicle distance from the preceding vehicle constant when there is a preceding vehicle. Such auto-cruise control basically obtains a target accelerator opening according to the host vehicle speed detected by the vehicle speed sensor, and increases / decreases the fuel injection amount to the engine according to the target accelerator opening, It is executed while decelerating by appropriately operating the auxiliary brake. By the way, the inter-vehicle distance control uses an inter-vehicle distance sensor such as a laser radar or a millimeter wave radar, or obtains the inter-vehicle distance and relative speed from the preceding vehicle using a camera and based on these information. This is done while appropriately operating the auxiliary brakes equipped with. (For example, refer to Patent Document 1).
Japanese Patent No. 3158924

  Conventionally, however, the vehicle speed is merely controlled according to the deviation or acceleration between the target vehicle speed and the vehicle speed. In other words, the vehicle speed is increased / decreased based on the previously set vehicle motion characteristics without taking into account the changes in the vehicle motion characteristics due to changes in the vehicle state (e.g., engine aging, vehicle weight change due to the load, etc.). Equally, the auto cruise control is just being executed. For this reason, for example, there is a possibility that a deviation amount from the target vehicle speed becomes large, a problem of undershoot or overshoot from the resume of the resume occurs, and a phenomenon such as hunting is caused.

  The present invention has been made in view of such circumstances, and its purpose is to execute appropriate control in consideration of vehicle conditions such as changes in vehicle weight and aging of the engine, for example, auto cruise control. It is in providing the suitable vehicle control apparatus.

According to the present invention, if a vehicle motion characteristic (dynamic characteristic) is learned using a neural network, a current vehicle model that has changed due to a change in vehicle weight due to a vehicle-mounted object or a secular change in an engine is obtained as a learning result. We focus on what we can seek.
Therefore, in order to achieve the above-described object, the vehicle control device according to the present invention performs constant speed traveling control that realizes constant speed traveling according to the set vehicle speed when there is no preceding vehicle ahead of the host vehicle, and is within a predetermined distance range. Auto-cruise control for performing inter-vehicle distance control for controlling the vehicle speed so as to keep the inter-vehicle distance to the preceding vehicle constant when there is a preceding vehicle in the vehicle,
A neural network that learns the motion characteristics of the vehicle according to vehicle behavior information and obtains a vehicle model;
A fuzzy control unit for obtaining a control output value for the speed of the vehicle according to a deviation between a control target value of the auto-cruise control for the behavior of the vehicle and the behavior of the vehicle;
Parameter determining means for determining a parameter of the fuzzy control unit that minimizes the deviation estimated by the fuzzy control unit in accordance with a vehicle model obtained by the neural network.

Preferably, as described in claim 2, the neural network uses, for example, a vehicle model that estimates the current vehicle speed and acceleration from time series data of past vehicle speed, acceleration, and accelerator opening according to the behavior information of the vehicle. Configured to learn. According to a third aspect of the present invention, the fuzzy control unit controls the vehicle speed according to a deviation between the target vehicle speed and the actual vehicle speed of the auto cruise control or according to a deviation between the set inter-vehicle distance and the inter-vehicle distance to the preceding vehicle. The control output value is obtained. Further, as described in claim 4, the parameter determining means increases a deviation between the control target value for the behavior of the vehicle and the behavior of the vehicle and a variation amount of the deviation (for example, a speed deviation and a speed deviation amount). The parameter of the fuzzy control unit is given as a constant value according to the evaluation level evaluated step by step, and the deviation and the variation amount of the deviation in the vehicle model obtained by the neural network are minimized. This is realized as determining parameters of the control unit .

  According to the vehicle control device having the above-described configuration, the parameter of the fuzzy control unit that obtains the control output value for the vehicle according to the control target value for the vehicle and its deviation is determined according to the vehicle model learned using the neural network. Even if the vehicle weight changes depending on the vehicle-mounted item or the engine performance changes over time, it is possible to obtain a control output value according to the motion characteristics of the vehicle at that time, and to improve its controllability. it can. Moreover, since the control parameters can always be updated optimally according to the movement and regulation of the vehicle, for example, the amount of vehicle speed deviation during auto-cruise control can be reduced, and highly natural control without a sense of incongruity becomes possible. Furthermore, the controllability can be improved only by changing the software (control parameter).

  If the parameters of the fuzzy control unit are determined as constant values according to the evaluation levels in which the deviation of the control object and the amount of change of the deviation (for example, speed deviation and speed deviation amount) are evaluated in multiple stages, membership The control output value can be optimally obtained more easily than the fuzzy control using a function.

Hereinafter, an auto cruise control apparatus will be described as an example of a vehicle control apparatus according to an embodiment of the present invention with reference to the drawings.
The constant speed running control in this automatic cruise control device is to control the vehicle speed to be constant based on the preset vehicle speed when there is no preceding vehicle ahead of the vehicle, and the inter-vehicle distance control is shown in FIG. As shown in FIG. 1, the vehicle speed is controlled so that the inter-vehicle distance L with the preceding vehicle B traveling in front of the vehicle (truck) A is kept constant. This inter-vehicle distance control is executed while detecting the host vehicle speed using a vehicle speed sensor and detecting the inter-vehicle distance L and the relative speed with respect to the preceding vehicle B using, for example, a laser radar or a millimeter wave radar.

  Such inter-vehicle distance control and constant-speed traveling control basically increase or decrease the fuel injection amount to the engine based on accelerator opening information (pseudo accelerator opening) obtained from the above-mentioned inter-vehicle distance L and relative speed. This is realized by controlling the own vehicle speed and decelerating the own vehicle A by operating a plurality of types of auxiliary brakes and main brakes included in the own vehicle A. The auxiliary brake is generated, for example, by opening an engine brake using an engine rotation load, an exhaust brake using exhaust pressure, or a specific valve (third valve) provided in the engine from the compression stroke to the exhaust stroke. It consists of a compression pressure release engine brake using negative work, and a fluid type retarder and an electromagnetic type retarder that adjust the flow of fluid by changing the blade angle of the stator in the fluid coupling.

  For example, as shown schematically in FIG. 2, the main brake basically moves the compressed air (air pressure) obtained in the air tank 1 through the control valve 2 (2FL, 2FR, 2RL, 2RR). The brake chamber 3 (3FL, 3FR, 3RLA, 3RRA, 3RLB, 3RRB) incorporated in each of the left and right wheels (not shown) is respectively applied to apply braking force to each wheel. In particular, in the electronically controlled main brake, the operation of each control valve 2 is controlled by an electric control signal from a brake control computer (EBS-ECU) 5 that detects the amount of depression of the brake pedal mechanism 4 and the like. By doing so, the braking force (chamber pressure) is controlled.

  In this electronically controlled main brake, even if a failure occurs in the electrical control of each control valve 2 due to a failure of the ECU, the brake pedal mechanism 4 has a stepping pressure operation amount. A corresponding chamber pressure is applied to each brake chamber 3 via the control valve 2. That is, an air supply system that supplies compressed air from the air tank 1 to the brake chambers 3 via the brake pedal mechanism 4 is provided in addition to the electronically controlled air supply system described above.

  Now, the auto-cruise control device that executes the above-described own vehicle speed control and deceleration control is schematically configured as shown in FIG. Note that the auto cruise control device having the functional block configuration shown in FIG. 3 may be realized as software on a computer (ECU) responsible for auto cruise control. This auto-cruise control device includes information on the vehicle speed detected by the vehicle speed sensor, information on the set vehicle speed and the set inter-vehicle distance given by an auto-cruise instruction operation, and a preceding vehicle obtained from an inter-vehicle distance sensor such as a millimeter wave radar. Information on the inter-vehicle distance with B and the relative speed are taken as input data, and the control operation is executed.

  When the auto-cruise control is not executed, the auto-cruise control switch 11 is switched to the OFF (OFF) side, so that it is actually obtained from an accelerator opening sensor attached to an accelerator pedal (not shown). Information on the accelerator opening operated by the driver is given to the engine ECU 12. The fuel injection amount is determined based on the accelerator opening information, and thereby the vehicle speed is controlled. On the other hand, when the auto-cruise control is executed, the auto-cruise control switch 11 is switched to the (ON) side, and the accelerator opening degree increase / decrease control unit 13 obtains based on the own vehicle speed or the like as described below. The pseudo accelerator opening / closing information is given to the engine ECU 12 via the adder 14. . The adder 14 adds the pseudo accelerator opening degree increase / decrease information output from the accelerator opening degree increase / decrease control unit 13 to the previous accelerator opening degree information given to the engine ECU 12 so that the engine described above is added. It plays a role of obtaining current accelerator opening information (pseudo accelerator opening information) to be given to the ECU 12.

  Here, the accelerator opening increase / decrease control unit 13 will be described in more detail. When there is no preceding vehicle B, the accelerator opening increase / decrease control unit 13 determines the difference between the own vehicle speed and the set vehicle speed obtained using the subtractor 15. (Vehicle speed deviation) and a value obtained by differentiating the vehicle speed deviation through the differentiator 16 (rate of change of the vehicle speed deviation) are input, and a pseudo accelerator is set to make the vehicle speed deviation 0 (zero) according to the information. The amount of opening / closing increase / decrease (accelerator opening / closing information) is obtained. Specifically, when the host vehicle speed exceeds the set vehicle speed, information is output to decrease the aforementioned pseudo accelerator opening, and conversely, when the host vehicle speed is less than the set vehicle speed, the pseudo accelerator opening is increased. Information is output. Then, the accelerator opening degree increase / decrease information is given to the engine ECU 12 via the adder 14 described above. As described above, the fuel injection amount is determined, and constant speed running control is realized.

  On the other hand, when the preceding vehicle B exists, the accelerator opening degree increase / decrease control unit 13 uses the target vehicle speed obtained by the vehicle speed increase / decrease control unit 17 based on the relative vehicle speed and the inter-vehicle distance instead of the set vehicle speed. That control is being executed. Specifically, the difference (vehicle speed deviation) between the target vehicle speed obtained by the vehicle speed increase / decrease control unit 17 and the host vehicle speed is calculated by the differentiator 15, and the vehicle speed deviation is reduced to 0 according to the vehicle speed deviation and its differential value. The above-described pseudo increase / decrease amount of the accelerator opening (accelerator opening increase / decrease information) is obtained so as to be (zero). The accelerator opening / closing information is input to the engine ECU 12 to perform the above-described constant speed traveling control and inter-vehicle distance control.

  The information on the target vehicle speed obtained by the vehicle speed increase / decrease control unit 17 is given to the target deceleration control unit 18 at the same time. The target deceleration control unit 18 needs to apply braking to the host vehicle A according to the inter-vehicle distance deviation obtained by the calculator 19 according to the set inter-vehicle distance and the inter-vehicle distance and the target vehicle speed obtained by the vehicle speed increase / decrease control unit 17. It plays the role of judging whether or not. Specifically, as described above, the fuel injection amount is determined based on the pseudo accelerator opening, and it is determined whether or not the inter-vehicle distance can be kept constant only by controlling the vehicle speed (or engine torque). Yes. If the inter-vehicle distance cannot be kept constant only by the vehicle speed control, the target vehicle speed and the inter-vehicle distance deviation are obtained based on the above-described target vehicle speed and inter-vehicle distance deviation information, which are specifically obtained from these information according to the inter-vehicle distance and relative speed. The deceleration f0 is calculated, and the information (target deceleration f0) is given to the target braking force calculator 20 to execute the brake control.

  The target braking force calculation unit 20 calculates a braking force (target braking force) necessary for obtaining the above-described target deceleration f0 from the motion characteristics of the vehicle A and the like. This target braking force is given to the auxiliary brake braking force calculation unit 21 to calculate the braking force obtained by the operation of the auxiliary brake 22 provided in the host vehicle A, and the auxiliary braking force is calculated under the control of the auxiliary brake braking force calculation unit 21. The brake 22 is driven. Further, the braking force by the auxiliary brake 22 obtained by the auxiliary brake braking force calculation unit 21 is given to the subtractor 24, and the difference from the target braking force obtained by the target braking force calculation unit 20 is particularly large. A shortage of the target braking force that cannot be satisfied only by the braking force obtained by the operation of the auxiliary brake 22 is required.

  The target chamber pressure calculation unit 25 calculates the chamber pressure for generating the insufficient braking force by the main brake 26 according to the information on the insufficient braking force obtained by the subtractor 24, and obtains the information on the chamber pressure. The main brake 26 is applied to actuate the main brake 26. That is, the target chamber pressure calculation unit 25 is driven when the target deceleration cannot be achieved only by the operation of the auxiliary brake 22, and in order to obtain an insufficient deceleration by the operation of the main brake 26, The chamber pressure applied to the main brake 26 is obtained. The control of the deceleration necessary for performing the above-mentioned inter-vehicle distance control is executed by the operation control of the auxiliary brake 22 and the operation control of the main brake 26.

  Basically, in the auto cruise control apparatus configured as described above, the vehicle control apparatus according to the present invention includes, for example, the accelerator opening increase / decrease control unit 13 for calculating the accelerator opening described above as a fuzzy control unit. This is realized by adaptively setting the parameters of the fuzzy control unit (accelerator opening degree increase / decrease control unit 13) according to the vehicle model learned using the neural network 31 as shown in FIG. That is, by using the fuzzy control unit 30 to obtain the pseudo accelerator opening u (t) corresponding to the vehicle speed deviation ei (t), the accelerator opening increase / decrease control unit 13 is realized and obtained from the neural network 31. The control parameter of the fuzzy control unit 30 is set so as to reduce the deviation between the control target value and the control output value according to the vehicle model. The vehicle speed increase / decrease control unit 17 and the target deceleration control unit 17 can be similarly realized as the fuzzy control unit 30.

The neural network 31 used in this vehicle control device has a three-layer structure of an input layer, an intermediate layer, and an output layer as shown in FIG. 5, for example. And two output layers. The number of neurons is determined by trial and error by simulation. The input / output function of each neuron is, for example, Y = {1−exp (−X)} / {1 + exp (−X)}
Is given as a sigmoid function.

In order to learn the motion characteristics of the vehicle using this neural network 31, for example, the input is a vehicle speed deviation and a vehicle speed deviation change amount 3 seconds ago, and a pseudo accelerator opening control amount every second from 10 seconds ago. The current vehicle speed deviation and the vehicle speed deviation change amount are obtained as output. However, the vehicle speed deviation is standardized at 4 km / h, the vehicle speed deviation change amount is 0.2 m / s ² , and the accelerator opening is 100%.

When a general back-propagation algorithm is used as a learning algorithm for such a neural network 31, the evaluation function J to be minimized by the learning is J = (1/2) Σ (e _i −e _Ni ) ²
As given. In the above equation, Σ represents a set having (i = 1, 2) as a parameter. When the update amount of the coupling load for decreasing the evaluation function J is obtained by using, for example, the steepest descent method, the update amount Δw (t) of the coupling load w is Δw (t) = α [∂J / ∂w ] + ΒΔw (t−Δt)
As given.

In the above equation, ∂ represents a partial differentiation, and the connection weight w (t) is a weighting factor representing the influence of the i-th neuron in the k-th layer on the j-th neuron in the (k + 1) -th layer, The combined load w (t−ΔT) indicates a weighting coefficient before ΔT seconds. Further, α is a learning coefficient that determines the learning speed, and β indicates a stabilization constant that suppresses vibration at the time of convergence using the previous weight correction amount. The combined load w obtained by reducing the evaluation function J from the correction amount of the weighting coefficient obtained according to this equation is w (t + ΔT) = w (t) + Δw (t)
As required.

On the other hand, the input / output characteristics of the above-described fuzzy control unit 30 show evaluation characteristics including ambiguity such as “the vehicle speed deviation is large”, “the vehicle speed deviation change is small”, etc. as a triangle, also if it indicates quantitatively the consequent part membership function a constant such as "the accelerator opening up 1%" _{Δu (t) = Σ i Σ} j f ij μ i (e1) ν _j (e2)
u (t) = U (t−ΔT) + Δu (t)
As shown. However, u (t) represents the pseudo accelerator opening instruction value, and Δu (t) represents the pseudo accelerator opening increase / decrease amount. Further, Σ _i indicates a set with i as a parameter, and Σ _j indicates a set with j as a parameter. Further, μ _i represents the degree of conformity to the i-th rule relating to the vehicle speed deviation, and ν _j represents the degree of conformity of the vehicle speed deviation change amount to the j-th rule. Further f _ij is given as a constant value of the consequent membership functions in applying i-th row j-th column of the rules table format as shown in FIG.

In FIG. 6 , the evaluation levels NB, NM, NS, ZO, PS, PM, and PB of the vehicle speed deviation μ _i and the vehicle speed deviation variation ν _j are large in the negative direction, medium in the negative direction, small in the negative direction, and zero. The evaluation levels for membership functions evaluated in multiple stages are shown as small in the positive direction, medium in the positive direction, and large in the positive direction. By searching for such a table, for example, "large vehicle speed deviation is negative (NB), when the vehicle speed deviation change amount is zero (ZO) is, f ₁₄ only a pseudo accelerator opening increasing" says output A value will be required. Then , by calculating the sum of the fitness and constant product of the consequent part membership for all the rules, the pseudo accelerator opening increase / decrease amount Δu (t) is obtained as shown in the above formula. .

  Therefore, by reducing the estimated value of the vehicle speed deviation amount and the vehicle speed deviation change amount, which is the output of the neural network 31 that has acquired the dynamic characteristics (motion characteristics) of the vehicle as described above, with respect to such a fuzzy control section 30. If the consequent part membership function in the fuzzy control unit 30 is changed so as to suppress the deviation from the target value and the acceleration / deceleration, a control output value corresponding to the motion characteristic of the vehicle can be obtained.

Specifically, E = (1/2) [w (t) −y (t)] ²
In order to calculate the error between the target vehicle speed w (t) and the actual vehicle speed y (t) and obtain the update amount of the above-mentioned fuzzy consequent part constant by the gradient method, the following learning signal 信号 E / ∂w _ij = ∂ E / ∂y (t) ΣA ・ B ・ C
A = ∂y (t) / ∂u (t−nΔT)
B = ∂u (t−nΔT) / ∂δu (t−nΔT)
C = ∂δu (t−nΔT) / ∂w _ij
Is calculated. However, u (t) is a pseudo accelerator opening, δu (t) is a pseudo accelerator opening increase / decrease amount, and w _ij is a fuzzy consequent part constant (control parameter).

Each partial differential is given by ∂E / ∂y (t) = y (t) -w (t)
∂u (t−nΔT) / ∂δu (t−nΔT) = 1
∂δu (t−nΔT) / ∂w _ij = μ _ij
Can be expressed as Note that μ _ij is the degree of conformity to the (i, j) -th fuzzy inference, and for ∂y (t) / ∂u (t−nΔT), an emulator configured by the neural network 31 is an actual vehicle. Can be calculated.

Based on the learning signal obtained as described above, the change amount of the fuzzy consequent part constant is expressed as follows: Δw _ij = −α [∂E / ∂w _ij ] + ηΔw _ij
It is possible to obtain the optimum fuzzy parameter by repeatedly executing these processes. In the above equation, α is a forgetting rate and η is a constant representing a learning rate.

  As a result, even if the actual vehicle speed is controlled by increasing / decreasing the pseudo accelerator opening with respect to the target vehicle speed, for example, as shown in FIG. According to the estimation result of the motion characteristic of the vehicle in the neural network 31, the control parameter is updated according to the change of the vehicle weight or the change of the engine characteristic due to the vehicle-mounted object. Therefore, as shown in FIG. Stable control with reduced error with respect to vehicle speed is possible. In other words, if the fuzzy rule is maintained in the initial state, the vehicle model is learned and the control parameters are updated as described above even if the control characteristics are as shown in FIG. Thus, good control characteristics as shown in FIG. 7B can be obtained.

  Incidentally, in FIG. 7A, the actual vehicle speed is stabilized at about 67 km / h with respect to the target vehicle speed of 70 km / h, a vehicle speed deviation of about 3 km / h remains, and the pseudo accelerator opening is also between 20 and 40%. This shows that hunting is controlled. On the other hand, when the present invention is applied, as shown in FIG. 7B, the deviation from the target vehicle speed is stable within 1.0 km / h, and the vehicle speed stability is excellent. Indicated.

  Also, the output value for determining the fuzzy consequent part constant by learning as described above is adaptively adapted from the initial state shown in FIG. 8 (a) according to the motion characteristics of the vehicle as shown in FIG. 8 (b). It will be corrected. In this case, as shown in FIG. 8B, the pseudo accelerator opening degree control amount is not simply set to increase or decrease according to the vehicle speed deviation or the vehicle speed deviation change amount as in the fuzzy rule assigned by the inference by the operator. At first glance, the pseudo accelerator opening control amount is such that a random number is assigned to each rule. However, in order to calculate an appropriate control amount comprehensively to control the vehicle speed to the target vehicle speed, it is considered that the pseudo accelerator opening control amount is set by skillfully using the fuzzy rule boundary ambiguity Can do.

  Therefore, according to the present invention, it is possible to learn the vehicle model by estimating the motion characteristics of the vehicle, and to update the control parameters as needed according to the learned vehicle model, so that the tuning for the control system is automatically executed. It becomes possible. And it becomes possible to reduce the vehicle speed deviation | shift amount with respect to a target vehicle speed, and to perform the appropriate control which does not depend on the change of a vehicle environment. In addition, effects such as the ability to improve the control performance only by updating the control parameters can be achieved.

The present invention is not limited to the embodiment described above. Here, the vehicle speed control by increasing / decreasing the pseudo accelerator opening is described as an example, but the present invention can be similarly applied to other control systems such as a braking control system. The configuration of the neural network 31 and the number of neurons in each layer may be set according to the control target.
Further, in the vehicle speed control in the embodiment, the fuel injection amount is determined based on the pseudo accelerator opening information, and thus the vehicle speed control in the engine ECU 12 is performed. However, the vehicle speed and the inter-vehicle distance are not obtained without obtaining the pseudo accelerator opening information. Alternatively, the speed may be controlled by directly obtaining the fuel injection amount in accordance with the relative vehicle speed or the like. In addition, the present invention can be implemented with various modifications without departing from the gist thereof.

The figure which shows the concept of the inter-vehicle distance control in auto-cruise control. The schematic block diagram of an electronic main brake. The principal part schematic block diagram of the auto-cruise control apparatus with which the vehicle control apparatus which concerns on one Embodiment of this invention is integrated. 1 is a diagram illustrating a schematic configuration of a vehicle control device according to an embodiment of the present invention and a concept of setting control parameters thereof. The figure which shows the structural example of the neural network applied to this invention. The figure which shows the example of the fuzzy rule tabulated used by this invention. The figure which contrasts and shows the vehicle speed control characteristic in the case where a fuzzy rule is kept constant, and the case where a fuzzy rule is updated according to the vehicle model learned with the neural network. The figure which contrasts the fuzzy rule set by the inference by an operator, and the fuzzy rule updated according to the learned vehicle model.

Explanation of symbols

12 Engine ECU
13 Accelerator opening / closing control unit 17 Vehicle speed increase / decrease control unit 18 Target deceleration control unit 20 Target braking force calculation unit 22 Auxiliary brake 26 Main brake 30 Fuzzy control unit 31 Neural network

Claims (4)

When there is no preceding vehicle ahead of the host vehicle, constant speed running control is performed to achieve constant speed driving according to the set vehicle speed. When there is a preceding vehicle within a predetermined distance range, the distance between the preceding vehicle and the preceding vehicle is kept constant. A vehicle control device that performs auto-cruise control that performs inter-vehicle distance control that controls the vehicle speed so as to maintain,
A neural network that learns the motion characteristics of the vehicle according to the vehicle behavior information and obtains a vehicle model;
A fuzzy control unit for obtaining a control output value for the speed of the vehicle according to a deviation between a control target value of the auto-cruise control for the behavior of the vehicle and the behavior of the vehicle;
A vehicle control apparatus comprising: parameter determination means for determining a parameter of the fuzzy control unit that minimizes the deviation estimated by the fuzzy control unit in accordance with a vehicle model obtained by the neural network. The said neural network learns the vehicle model which estimates the present vehicle speed and acceleration from the time series data of the past vehicle speed, acceleration, and accelerator opening according to the behavior information of the said vehicle . Vehicle control device. The fuzzy control unit obtains a control output value for controlling the vehicle speed according to the deviation between the target vehicle speed and the actual vehicle speed of the auto cruise control or according to the deviation between the set inter-vehicle distance and the inter-vehicle distance to the preceding vehicle. The vehicle control device according to claim 1. The parameter determining means is a value obtained by making the parameters of the fuzzy control unit constant according to an evaluation level in which a deviation between a control target value for the behavior of the vehicle and a behavior of the vehicle and a change amount of the deviation are evaluated in multiple stages. To give
The vehicle control device according to claim 1 , wherein the parameter of the fuzzy control unit is determined so as to minimize the deviation and a variation amount of the deviation in the vehicle model obtained by the neural network .

Images