CN108422997A - A kind of automobile active safety cooperative control system and method based on wolf pack algorithm - Google Patents

Abstract

The invention discloses a kind of automobile active safety cooperative control systems and method based on wolf pack algorithm, including sensing system, top level control device, lower layer's controller, chassis execution system.Pilot control signal is monitored by sensing system first, driving states signal, top level control device calculates top level control object (speed according to driver's desired motion and the difference of actual motion by wolf pack algorithm, side slip angle, side acceleration, yaw velocity) optimal solution, and it is output to lower layer's controller, lower layer's controller uses neural network algorithm, with the optimal solution of upper layer control object target in order to control, calculate lower layer's control object (front wheel angle, each wheel brake pressure) optimal solution, it is sent to chassis and executes system, so as to adjust the transport condition of automobile, realize the automobile active safety Collaborative Control under emergency work condition.The present invention optimizes people-vehicle as closed-loop system, can realize the optimum control of the vehicle overall situation.

Description

A vehicle active safety cooperative control system and method based on wolf pack algorithm

technical field

The invention relates to the technical field of automobile active safety, in particular to an automobile active safety cooperative control system and method based on wolf pack algorithm.

Background technique

With the progress of the times, people have higher and higher expectations for the safety performance of automobiles, and active safety control systems have gradually become the standard configuration of various models. The active safety system is composed of multiple subsystems, and each system needs to cooperate closely to realize the control intention. Therefore, the cooperative control of the active safety system has become a research hotspot in the field. However, most of the current research only considers the instability of the car itself, while ignoring the influence of the driver's manipulation on the driving state of the car, which reduces the control accuracy; at the same time, because the widely used optimization algorithm convergence Poor, most of the optimization schemes only consider one control objective, the optimization results only achieve local optimization, and fail to achieve the overall optimal control of the whole vehicle.

Contents of the invention

In view of the deficiencies of the above-mentioned prior art, the present invention provides a vehicle active safety cooperative control system and method based on wolf pack algorithm, which can realize the overall optimal control of the whole vehicle.

To achieve the above object, the present invention adopts the following technical solutions:

A vehicle active safety cooperative control system based on the wolf pack algorithm, including a sequentially connected sensor system, an upper controller, a lower controller, and a chassis execution system;

The sensor system includes a driver sensing module for collecting driver action signals, and a vehicle sensing module for recording the current driving condition of the car; the driver sensing module and the vehicle sensing module transmit sensor signals to the upper controller respectively;

The upper controller includes a signal recognition module, a decision module, and an optimization module; the signal recognition module receives sensor signals, and sends them to the decision module after preprocessing; the decision module predicts the driver's expected motion parameters according to the sensor signals, and sends them to An optimization module, the optimization module takes the minimum difference between the expected motion parameters and the actual motion parameters as the control target, takes the actual motion parameters as the upper-level control object, uses the wolf pack algorithm to perform optimization calculations, and uses the optimal solution of the upper-level control object sent to the lower controller;

The lower-level controller takes the optimal solution of the upper-level control object as the control target, performs optimization calculations on the front wheel rotation angle and the brake hydraulic pressure of the wheels as the lower-level control objects, calculates the optimal solution of the lower-level control object, and outputs control signals to the chassis execution system;

The chassis execution system is used to receive the control signal output by the lower controller, control the rotation of the steering execution motor and adjust the wheel brake hydraulic pressure.

Further, the motion parameters include vehicle speed, side slip angle of center of mass, lateral acceleration, and yaw rate.

Further, the driver sensing module includes a steering wheel angle sensor and a brake pedal displacement sensor for collecting driver action signals; the vehicle sensing module includes a wheel sensor, a sideslip sensor, and a gyroscope sensor for recording Current driving conditions.

A vehicle active safety cooperative control method based on wolf pack algorithm, comprising the following steps:

Step 1: The sensor system collects the driver's action signal and the current driving status of the car, and transmits it to the upper controller;

Step 2: After the upper controller receives the sensor signal, the sensor signal is preprocessed by the signal recognition module and passed to the decision-making module. The decision-making module predicts the driver's expected motion parameters according to the sensor signal and sends it to the optimization module. The optimization module uses the expected motion parameters The minimum gap between the actual motion parameters is the control target, the actual motion parameters are used as the upper control object, the wolf pack algorithm is used for optimization calculation, and the optimal solution of the upper control object is sent to the lower controller;

Step 3: The lower-level controller takes the optimal solution of the upper-level control object as the control target, performs optimization calculations on the front wheel rotation angle and the brake hydraulic pressure of the wheels as the lower-level control objects, and outputs control signals to the chassis execution system;

Step 4: The chassis execution system includes an AFS actuator and a DYC actuator. The DYC actuator receives the brake hydraulic control signal output by the lower controller to adjust the wheel brake hydraulic pressure. The AFS actuator receives the front wheel brake output output from the lower controller. The steering angle control signal controls the steering actuator motor.

Preferably, in step 2, the specific steps of the wolf pack algorithm include:

Step 2.1: Value initialization

a) Initialize the number of wolves N and the position information _Xi (ν, β, α _y , γ) of each wolf, i=(1, N), the position information is four-dimensional coordinates, ν, β, α _y , γ respectively represent vehicle speed, side slip angle of center of mass, lateral acceleration and yaw rate;

b) Set the prey odor concentration Y _i corresponding to each wolf, which is the objective function formed by the difference between the driver's expected vehicle position coordinates and the actual vehicle position coordinates:

Among them, (x _t , y _t ) is the actual vehicle position coordinates at time t, (x _e , y _e ) is the vehicle position coordinates expected by the driver at time t, and

x _t ＝x ₀ +ν*cos(β+γ*t)*t

x ₀ and y ₀ are the initial position coordinates of the vehicle;

c) According to the control accuracy and solution speed requirements, set the maximum number of iterations k _max , the maximum number of walks T _max , the wolf detection scale factor α, the distance determination factor ω, the step size factor S and the update scale factor σ;

d) Select the artificial wolf with the highest prey odor concentration as the head wolf s, take its position as X _lead , and the prey odor concentration as Y _lead ;

Step 2.2: Walking Behavior

a) Arrange the prey odor concentrations of N wolves from large to small, and take the artificial wolves corresponding to the first N*α prey odor concentrations except the head wolf as the detection wolf, and all the artificial wolves except the head wolf and the detection wolf as the ferocious wolf Wolf;

b) For wolf j, whose prey odor concentration is Y _j ⁰ , at this time Y _j ⁰ <Y _lead , j=(1,N*α), it performs a wandering behavior, and advances one walk in h directions respectively The step size is step _a , record the perceived prey odor concentration after each step forward and then return to the original position, then the position of the wolf j after moving to the pth (p=1,2,3...,h) direction is X _j ^p , the prey odor concentration is Y _j ^p ;

c) Choose the direction where Y _j ^p is the largest and greater than the prey odor concentration Y _j ⁰ at the current position, advance a walking step step _a , and update the position information X _j of wolf j;

d) Repeat the above walking behavior until the concentration of prey odor perceived by a certain wolf is Y _j > Y _lead , at this time the wolf becomes the leader wolf, Y _lead = Y _j , or the number of wanderings T reaches the maximum wandering The number of times T _max , at this time, the original wolf remains unchanged;

Step 2.3: Summon/Rush Behavior

a) The head wolf initiates the summoning behavior by howling, and calls the surrounding N*(1-α)-1 wolves to quickly approach the head wolf's position, and the wolf quickly approaches the head wolf with a running step of step _b = 2*step _a where the wolf is located;

b) During the attack, if the wolf z (z=(1,N*(1-α)-1)) perceives the prey odor concentration Y _z >Y _lead , then the wolf will transform into a wolf leader and initiate a calling behavior , at this time Y _lead = Y _z ;

c) During the attack, if the prey odor concentration Yz perceived by the wolf _z <Y _lead , then the wolf z will continue to attack until the distance d _zs between it and the wolf s is less than d _near and join the ranks of attacking the prey , that is, turning into siege behavior;

Step 2.4: Siege Behavior

a) Consider the position of the wolf closest to the prey, that is, the head wolf, as the moving position of the prey, and the ferocious wolf and the wolf detection will approach the prey closely with the attack step step _c in order to capture it;

b) If after the siege is carried out, the prey odor concentration perceived by an artificial wolf is greater than the prey odor concentration perceived by its original position, then update the position of the artificial wolf, otherwise, the position of the artificial wolf remains unchanged;

c) If after updating the position of the artificial wolf, the concentration of prey odor perceived by an artificial wolf is greater than that of the head wolf, the artificial wolf will be transformed into a head wolf, and the position of the artificial wolf will be regarded as the location of the prey , other artificial wolves continue to besiege around this wolf;

d) After updating the head wolf's position, judge whether the optimization accuracy or the maximum number of iterations k _max is reached, and if so, output the head wolf's position, which is the optimal solution sought, otherwise repeat steps 2.2-2.4.

Preferably, in step 3, the optimization calculation method of the lower controller is a neural network algorithm, the input of the neural network algorithm is the front wheel angle, the brake hydraulic pressure of the wheel; the output is the vehicle speed, the side slip angle of the center of mass, the side Acceleration, yaw rate.

Beneficial effects: 1. The human-vehicle is optimized as a closed-loop system, and the influence of the driver's manipulation on the control target is considered;

2. The wolf pack algorithm has good robustness and global convergence performance for complex functions with different characteristics, which further improves the accuracy and efficiency of the control system;

3. The neural network algorithm improves the adaptability and fault tolerance of the control system due to its unique large-scale parallel structure, distributed storage of information and parallel processing characteristics.

Description of drawings

Fig. 1 is a flow chart of the control method of the present invention;

Fig. 2 is wolf pack algorithm flowchart in the present invention;

Fig. 3 is a flow chart of the neural network algorithm in the present invention.

Detailed ways

The present invention will be further explained below in conjunction with the accompanying drawings.

The invention relates to a vehicle active safety cooperative control system and method based on wolf pack algorithm. First, the sensor system monitors the driver's manipulation signal and the driving status signal. Based on the signal, the upper-level controller performs optimal calculation on the upper-level control object by the built-in algorithm, and the lower-level controller performs optimal calculation on the lower-level control object and converts it into a control command and sends it to the actuator. , in order to realize the adjustment of the driving state.

The present invention will be further described below in conjunction with the accompanying drawings.

As shown in Figure 1, working method of the present invention is:

1) The sensor system monitors the driver's manipulation signals and driving status signals, and transmits them to the upper controller.

2) After the upper controller receives the sensor signal, the sensor signal is preprocessed by the signal recognition module (including filtering, amplification, modulation, etc., the purpose is to extract the content we need in the signal), and the driver's control signal and the current motion state of the vehicle The signal is delivered to the decision-making module, and the decision-making module estimates the driver's expected motion parameters according to the above signals (this is the prior art, so this paper will not repeat them), and input to the optimization module, the optimization module uses the difference between the expected motion parameters and the actual motion parameters The minimum is the control target, and the actual vehicle speed, center of mass side slip angle, lateral acceleration, and yaw rate are the upper-level control objects, and the wolf pack algorithm is used for optimization calculation, and the optimal solution is output to the lower-level controller.

3) In the lower-layer controller, the optimal solution of the upper-layer control objects (vehicle speed, side slip angle, lateral acceleration, and yaw rate) is the control target, and the front wheel rotation angle and the brake hydraulic pressure of the wheels are the lower-layer control objects. The network algorithm performs optimization calculations and outputs control signals to the chassis execution system.

4) The chassis execution system executes actions according to the corresponding control signals, and cooperates to realize the adjustment of the driving state.

5) At this time, the driving status signal and the subsequent driver's manipulation signal will change, and the sensor system will continuously monitor the above signals and transmit them to the upper controller, and the cycle will repeat.

As shown in Figure 2, the optimization process of the wolf pack algorithm is:

Step 2.1: Value initialization

a) Initialize the number of wolves N and the position information _Xi (ν, β, α _y , γ) of each wolf, i=(1, N), the position information is four-dimensional coordinates, ν, β, α _y , γ respectively represent vehicle speed, side slip angle of center of mass, lateral acceleration and yaw rate;

b) Set the prey odor concentration Y _i corresponding to each wolf, which is the objective function formed by the difference between the driver's expected motion and the actual motion

Among them, (x _t , y _t ) is the actual vehicle position coordinates at time t, (x _e , y _e ) is the driver’s expected vehicle position coordinates at time t, and

x _t ＝x ₀ +ν*cos(β+γ*t)*t

x ₀ and y ₀ are the initial position coordinates of the vehicle;

c) According to the control accuracy and solution speed requirements, set the maximum number of iterations k _max , the maximum number of walks T _max , the wolf detection scale factor α, the distance determination factor ω, the step size factor S and the update scale factor σ;

d) Select the artificial wolf with the highest prey odor concentration as the head wolf s, take its position as X _lead , and the prey odor concentration as Y _lead ;

Step 2.2: Walking Behavior

a) Take the N*α artificial wolves with better prey odor concentration in the initial solution space as the wolves, and all the artificial wolves except the head wolf and the wolves are ferocious wolves;

b) For the wolf detective j, the prey odor concentration Y _j ⁰ at the current position, at this time Y _j ⁰ <Y _lead , j=(1,N*α), it performs a wandering behavior, and advances one swim in h directions respectively. Walk step length step _a , record the perceived prey odor concentration after each step forward and then return to the original position, then the position of wolf j after moving to the pth (p=1,2,3...,h) direction is The prey odor concentration is

c) Choose the direction where Y _j ^p is the largest and greater than the prey odor concentration Y _j ⁰ at the current position, advance a walking step step _a , and update the position information X _j of wolf j;

d) Repeat the above walking behavior until the concentration of prey odor perceived by a certain wolf is Y _j > Y _lead , at this time the wolf becomes the leader wolf, Y _lead = Y _j , or the number of wanderings T reaches the maximum wandering The number of times T _max , at this time, the original wolf remains unchanged;

Step 2.3: Summon/Rush Behavior

a) The head wolf initiates the summoning behavior by howling, and calls the surrounding N*(1-α)-1 wolves to quickly approach the head wolf's position, and the wolf quickly approaches the head wolf with a running step of step _b = 2*step _a where the wolf is located;

b) During the attack, if the wolf z (z=(1,N*(1-α)-1)) perceives the prey odor concentration Y _z >Y _lead , then the wolf will transform into a wolf leader and initiate a calling behavior , at this time Y _lead = Y _z ;

c) During the attack, if the prey odor concentration Yz perceived by the wolf _z <Y _lead , then the wolf z will continue to attack until the distance d _zs between it and the wolf s is less than d _near (by the distance determination factor ω and the optimization When the value range of the object is determined), join the ranks of attacking the prey, that is, turn into the siege behavior;

Step 2.4: Siege Behavior

a) Consider the position of the wolf closest to the prey, that is, the head wolf, as the moving position of the prey, and the ferocious wolf should cooperate with the wolf detection to approach the prey closely with the attack step _c in order to capture it;

b) If after the siege is carried out, the prey odor concentration perceived by an artificial wolf is greater than the prey odor concentration perceived by its original position, then update the position of the artificial wolf, otherwise, the position of the artificial wolf remains unchanged;

c) If after updating the position of the artificial wolf, the concentration of prey odor perceived by an artificial wolf is greater than that of the head wolf, the artificial wolf will be transformed into a head wolf, and the position of the artificial wolf will be regarded as the location of the prey , other artificial wolves continue to besiege around this wolf;

d) After updating the position of the wolf, judge whether the optimization accuracy or the maximum number of iterations k _max is reached, and if so, output the position of the wolf, which is the optimal solution sought, otherwise go to step 2.2.

The wolf pack algorithm applies two mechanisms:

1. The wolves update mechanism of "survival of the strong"

a) The prey is distributed according to the principle of "from strong to weak", resulting in weak wolves starving to death, that is, it is necessary to remove N*σ artificial wolves with the lowest prey odor concentration in the wolf pack;

b) At the same time, in order to maintain the individual diversity of wolves, it is necessary to randomly generate N*σ new artificial wolves, that is, to open up a new solution space;

2. The "winner takes the king" alpha wolf production mechanism

a) In order to ensure that the final solution is optimal, it is necessary to continuously carry out the survival of the fittest for the leader wolf. When there is an artificial wolf whose prey odor concentration is greater than Y _lead , the wolf is updated as the leader wolf;

b) After updating the position of the wolf, judge whether the optimization accuracy or the maximum number of iterations k _max is reached, and if so, output the position of the wolf, which is the optimal solution sought, otherwise go to step 2).

As shown in Figure 3, the training process of the neural network algorithm is:

The input of the neural network system is selected as X _i (front wheel angle, front left wheel brake hydraulic pressure, front right wheel brake hydraulic pressure, rear left wheel brake hydraulic pressure, rear right wheel brake hydraulic pressure), and the system output is Y _i (vehicle speed, center of mass side deflection angle, lateral acceleration, and yaw rate), that is, the number of nodes in the input layer is n=5, the number of nodes in the output layer is m=4, and the number of nodes in the hidden layer is l=3. Take a large amount of real vehicle test data, that is, measure the input and output signals of the real vehicle when driving, and use it as the sample data for neural network training to train the neural network. After iterating the results, take some data to test the neural network. When the output results meet the accuracy When required, it can be used for parameter optimization of the control system. At this point, given a set of outputs (vehicle speed, side slip angle, lateral acceleration, yaw rate), the corresponding set of inputs (front wheel rotation angle, front left wheel brake pressure, front right wheel brake pressure, rear left wheel brake pressure, rear right wheel brake pressure), that is, the optimal solution for the control object under the given control objective has been completed. Then, the lower layer controller sends the optimal solution of the control object to the chassis execution system.

The above is only a preferred embodiment of the present invention, it should be pointed out that, for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made, and these improvements and modifications can also be made. It should be regarded as the protection scope of the present invention.

Claims (6)

1. a kind of automobile active safety cooperative control system based on wolf pack algorithm, which is characterized in that including sequentially connected biography Sensor system, top level control device, lower layer's controller, chassis execute system；

The sensing system includes for acquiring driver's sensing module of driver actions' signal, current for recording automobile The vehicle sensing module of driving condition；Driver's sensing module and vehicle sensing module are transmitted to upper layer controller respectively to be passed Sensor signal；

The top level control device includes signal identification module, decision-making module, optimization module；The signal identification module receives sensing Device signal is sent to decision-making module after pretreatment；The decision-making module predicts driver's desired movement ginseng according to sensor signal Number, and it is sent to optimization module, the optimization module is minimum with the gap between actual motion parameter with desired movement parameter Control targe optimizes calculating using actual motion parameter as top level control object using wolf pack algorithm, and by top level control pair The optimal solution of elephant is sent to lower layer's controller；

Lower layer's controller is with the optimal solution of upper layer control object target in order to control, with front wheel angle, the brake fluid pressure of wheel Calculating is optimized for lower layer's control object, and exports optimal solution and executes system to chassis；

The chassis executes the control signal that system is used to receive lower layer's controller output, control turn to actuating motor rotation and Adjust wheel braking hydraulic pressure.

2. a kind of automobile active safety cooperative control system based on wolf pack algorithm according to claim 1, feature exist In the kinematic parameter includes speed, side slip angle, side acceleration, yaw velocity.

3. a kind of automobile active safety cooperative control system based on wolf pack algorithm according to claim 1, feature exist In driver's sensing module includes steering wheel angle sensor, brake pedal displacement sensor, for acquiring driver actions Signal；The vehicle sensing module includes wheel detector, sideslip sensor, gyro sensor, current for recording automobile Driving condition.

4. a kind of automobile active safety cooperative control method based on wolf pack algorithm based on control system described in claim 1, It is characterised in that it includes following steps：

Step 1：Sensing system acquires driver actions' signal and automobile current driving situation, and passes it to top level control Device；

Step 2：After top level control device receiving sensor signal, by signal identification module pre-processing sensor signals, and pass to Decision-making module, decision-making module predicts driver's desired movement parameter according to sensor signal, and is sent to optimization module, optimizes mould Block, with the minimum control targe of gap between actual motion parameter, is controlled with desired movement parameter by upper layer of actual motion parameter Object processed optimizes calculating using wolf pack algorithm, and the optimal solution of top level control object is sent to lower layer's controller；

Step 3：Lower layer's controller is with the optimal solution of upper layer control object target in order to control, with front wheel angle, the brake fluid of wheel Pressure is that lower layer's control object optimizes calculating, calculates the optimal solution of lower layer's control object, and output a control signal to chassis Execution system；

Step 4：It includes AFS actuators, DYC actuators that the chassis, which executes system, and the DYC actuators receive lower layer's controller The brake fluid pressure of output controls Signal Regulation wheel braking hydraulic pressure, and AFS actuators receive the front wheel angle of lower layer's controller output It controls signal control and turns to actuating motor.

5. a kind of automobile active safety cooperative control method based on wolf pack algorithm according to claim 4, feature exist In in step 2, the specific steps of the wolf pack algorithm include：

Step 2.1：Numerical value initializes

A) the wolf pack quantity N and location information X of every wolf is initialized_i(ν,β,α_y, γ), i=(1, N), the location information is Four-dimensional coordinate, ν, β, α_y, γ respectively represents speed, side slip angle, side acceleration, yaw velocity；

B) the corresponding prey odorousness Y of every wolf is set_i, the as desired vehicle location coordinate of driver and actual vehicle The object function that the difference of position coordinates is constituted：

Wherein, (x_t,y_t) it is t moment vehicle actual position coordinate, (x_e,y_e) it is the desired vehicle location coordinate of t moment driver, And

x_t=x₀+ν*cos(β+γ*t)*t

x₀、y₀For vehicle initial position co-ordinates；

C) according to control accuracy and solving speed demand, setting maximum iteration k_max, maximum migration number T_max, visit wolf ratio Factor-alpha, range estimation factor ω, step factor S and update scale factor σ；

D) it is head wolf s to choose the current highest artificial wolf of prey odorousness, and it is X to take its position_lead, prey odorousness be Y_lead；

Step 2.2：Migration behavior

A) the prey odorousness of N wolf is arranged from big to small, the corresponding people of N* α prey odorousnesses before removing outside a wolf Work wolf is to visit wolf, except head wolf and all artificial wolves in addition to visiting wolf are violent wolf；

B) for visiting wolf j, prey odorousness is Y_j ⁰, Y at this time_j ⁰＜ Y_lead, j=(1, N* α), progress migration behavior, to h Advance a migration step-length step respectively in a direction_a, original is retracted after recording the prey odorousness perceived after often taking a step forward Position, the then position that wolf j is visited after advancing to a direction of pth (p=1,2,3..., h) arePrey odorousness is

C) it selectsPrey odorousness Y maximum and more than current location_j ⁰Direction advance a migration step-length step_a, more The new location information X for visiting wolf j_j；

D) the migration behavior more than repeating, until certain only visits the prey odorousness Y that wolf perceives_j＞ Y_lead, at this time the spy wolf at For head wolf, Y_lead=Y_j, or migration number T reaches maximum migration number T_max, maintain procephalon wolf constant at this time；

Step 2.3：Calling/long-range raid behavior

A) head wolf initiates calling behavior by yelping, and the violent wolves of N* (1- α) -1 of surrounding is convened to be leaned on rapidly to head wolf position Hold together, violent wolf is with long-range raid step-length step_b=2*step_aQuickly approach a wolf position；

B) long-range raid on the way, if the prey odorousness Y that violent wolf z (z=(1, N* (1- α) -1)) is perceived_z＞ Y_lead, then the violent wolf It is converted into a wolf and initiates calling behavior, at this time Y_lead=Y_z；

C) long-range raid on the way, if the prey odorousness Y that violent wolf z is perceived_z＜ Y_lead, then violent wolf z continuation long-range raid is until itself and head wolf The distance between s d_zsLess than d_nearWhen be added to attack ranks to prey, that is, be transferred to jointly attack behavior；

Step 2.4：Jointly attack behavior

A) position of wolf, that is, head wolf nearest from prey at this time is considered as the shift position of prey, violent wolf joint visits wolf to attack step Long step_cTo prey close approximation to being captured；

If after b) implementing jointly attack behavior, the prey odorousness that certain artificial wolf perceives is more than what its original position state was perceived Prey odorousness then updates the position of this artificial wolf, and if not, artificial wolf position is constant；

If after c) updating artificial wolf position, the prey odorousness that certain artificial wolf perceives is more than the prey that head wolf is perceived Odorousness, then the artificial wolf be converted into a wolf, the position of the artificial wolf is considered as prey position, other artificial wolves are with this Continue jointly attack behavior centered on wolf；

D) behind update head wolf position, judge whether to reach optimization precision or maximum iteration k_max, this wolf is exported if reaching Position, i.e., required optimal solution, otherwise repeatedly step 2.2-2.4.

6. a kind of automobile active safety cooperative control method based on wolf pack algorithm according to claim 4, feature exist In, in step 3, lower layer's controller optimization calculate method be neural network algorithm, the neural network algorithm it is defeated Enter for the brake fluid pressure of front wheel angle, wheel；Output is speed, side slip angle, side acceleration, yaw velocity.