CN118151531B - Coordinated control method of distributed electric vehicles based on cooperative game - Google Patents

Abstract

The present invention discloses a distributed electric vehicle multi-agent coordinated control method based on cooperative game. First, each subsystem of the distributed electric vehicle is divided into 6 agents, and a dynamic model characterizing the agents of each subsystem is constructed. The vehicle status information is collected in real time in combination with the on-board sensor. The T-S fuzzy theory is used to deal with the system nonlinearity caused by the change of vehicle speed. A distributed electric vehicle subsystem dynamic coordinated control strategy is established based on cooperative game theory. Finally, a robust compensation control strategy is established to reduce the impact of system disturbance and uncertainty on system control, and to achieve intelligent coordinated control of distributed electric vehicles and efficient and safe driving. This method can constrain the agent behavior with global performance indicators. Compared with the traditional hierarchical control method, it is safer and more reliable. It can dynamically adjust the control output of each agent in real time, and provide a new exploration direction for the modular coordinated control of distributed electric vehicles.

Description

Coordinated control method of distributed electric vehicles based on cooperative game

Technical Field

The present invention relates to the field of distributed electric vehicle intelligent interaction and vehicle advanced driver assistance, and in particular to a distributed electric vehicle multi-agent coordinated control method based on cooperative game.

Background technique

At present, distributed electric vehicles with wheel hub motors as drive control units have torque control modes with independent control of multiple actuators, fast response and precise execution. Their simple chassis architecture gives the car greater room for handling stability improvement, and has been considered by domestic and foreign scholars in the automotive field as one of the most promising electric vehicle architectures. However, due to the coupling relationship between wheel hub motors, wheels and steering mechanisms in dynamics and kinematics, when different controls intervene, due to overlapping interference of functions, execution conflicts may be caused or even partial control failures may occur. Especially in the process of drivers participating in the driving control loop, due to the strong subjectivity of drivers, multiple actuators are more likely to cause execution conflicts. Therefore, establishing a reasonable and effective real-time dynamic coordination control scheme for multiple actuators is very important for improving the active safety of distributed electric vehicles and promoting the transformation of the automotive industry. In particular, the actuators of distributed electric vehicles are divided into intelligent bodies according to functional units, the dynamic models of each intelligent body are constructed, and the dynamic coordination control architecture based on cooperative game is established. At the same time, robust compensation control is used to effectively eliminate system disturbances and uncertainties, thereby better ensuring the safe and efficient operation of distributed electric vehicles, providing a theoretical basis and technical support for the research of distributed electric vehicles in my country.

The prior art has discussed a distributed electric vehicle stability control method, which achieves distributed electric vehicle stability control by optimizing the distribution of tire lateral force and longitudinal force, thereby improving the stability and handling performance of the vehicle system. However, the vehicle actuators cannot interact effectively in real time, and only consider the coordination relationship between the front wheel steering and the yaw moment, without considering the coupling relationship of the vehicle under the intervention of the driver. This makes the method unable to better meet the multi-actuator coordinated control function of distributed electric vehicles in various environments, and there is a problem of low intelligence level.

Summary of the invention

The technical problem to be solved by the present invention is to provide a coordinated control method that can dynamically coordinate the control inputs of various intelligent agents in distributed electric vehicles, effectively eliminating the coupling conflicts and even local failures that may be caused by incomplete actuator information interaction and overlapping functional interference, and providing a distributed electric vehicle multi-agent dynamic coordinated control method with a high intelligence level and strong practicality.

In order to solve the above technical problems, the present invention adopts the following technical solutions:

A distributed electric vehicle multi-agent coordinated control method based on cooperative game is provided, which includes the following steps

Step S1: Divide the intelligent control area according to the functions of each subsystem of the distributed electric vehicle, establish the steering system intelligent control area, and divide it into the driver agent and the steering auxiliary control agent; in the drive/brake system intelligent control area, since the distributed electric vehicle uses the wheel hub motor to control the drive/brake of the vehicle, and the four wheels are independently controllable, establish the left front wheel agent, the left rear wheel agent, the right front wheel agent, and the right rear wheel agent;

Step S2: for the six agents defined in step S1, namely, the driver agent, the steering assist control agent, the left front wheel agent, the left rear wheel agent, the right front wheel agent and the right rear wheel agent, a dynamic model representing each agent is constructed, thereby establishing a vehicle system dynamic model;

Step S3: linearize the vehicle system using T-S fuzzy theory and establish the system state equation after linearization;

Step S4: Establish a distributed electric vehicle multi-agent coordinated control strategy based on cooperative game;

Step S5: In order to solve the problem that cooperative game cannot eliminate system disturbances, a robust compensation control strategy is established to achieve efficient and safe dynamic coordinated control of the six intelligent agents of distributed electric vehicles.

Preferably, step S2 constructs a dynamic model representing each of the six agents defined in step S1, thereby establishing a vehicle system dynamic model, including the following steps:

Step S21: Establish the dynamic equation that characterizes the driver agent:

The dynamic expression of the hand torque input of the driver agent is:

Among them, τ _h is the steering torque input of the driver agent, J _h is the equivalent inertia of the steering system, B _h is the damping of the steering system, τ _f is the friction torque of the steering system, τ _e is the steering assist torque provided by the steering system motor, δ is the steering wheel angle, and sign() is a sign function used to indicate the positive and negative signs of the parameters;

Step S22: Establish the dynamic equation of the auxiliary steering agent:

The dynamic equation of the auxiliary steering agent is expressed as:

τ _e =I _e K _t η

(Formula 2)

Among them, _τe is the steering torque output of the auxiliary steering intelligent body, _Ie is the real current of the steering motor, _Kt is the torque coefficient of the steering motor, and η is the mechanical efficiency;

Step S23: Establish the dynamic model of the hub motor intelligent body

Since the performance of the four wheel hub motors of distributed electric vehicles is the same, the dynamic model of the wheel hub motor agent can be expressed as;

τ _ij ＝τ _Lij +B _ij +J _ij ω _ij

(Formula 3)

Where, ij = {ff, fr, rf, rr}, respectively represents the left front, left rear, right front, and right rear. ω is the angular velocity of the hub motor, τ, τ _L represent the output torque and load torque of the hub motor, J is the moment of inertia of the hub motor, and B is the damping coefficient;

Step S24: Establishing a distributed electric vehicle system model including the driver agent

Assuming that the tire slip angle of the vehicle is small, the vehicle dynamics model includes longitudinal motion, lateral motion and yaw motion. The distributed electric vehicle system dynamics model can be expressed as:

in, δ is the steering wheel angle, is the steering wheel angular velocity, β is the vehicle center of mass sideslip angle, is the vehicle yaw rate, Y is the vehicle lateral displacement, ω _r is the vehicle yaw angle, v _x is the vehicle longitudinal velocity, _vy is the vehicle lateral velocity, A is the vehicle state system matrix, B is the control input system matrix, B _h is the driver control input system matrix, u = [τ _e τ _ff τ _fr τ _rf τ _rr ], τ _ff is the vehicle left front wheel driving torque, τ _fr is the left rear wheel driving torque, τ _rf is the right front wheel driving torque, τ _rr is the right rear wheel driving torque, H is the system disturbance matrix, w is the system disturbance, z is the system output, and C is the system output matrix.

Preferably, step S3 uses T-S fuzzy theory to perform linear processing on the vehicle system and establishes the system state equation after linear processing, including the following steps:

Step S31: Set the upper limit of the longitudinal speed of the distributed electric vehicle to The lower boundary is

Step S32: Establishing a human-vehicle system model based on T-S fuzzy system:

Taking into account the nonlinear problem caused by the longitudinal speed on the human-vehicle system, the linearized human-vehicle system model is established based on T-S fuzzy theory:

Where k is the current time of the system, υ is the advance variable, ζ _i (υ) is the normalized membership function of the TS fuzzy model, _{Ai_b} is the system state matrix after TS fuzzy processing, B _{h_h} is the driver control matrix after discretization in the system, B _d is the vehicle control matrix after discretization in the system, and H _d is the interference matrix after discretization in the system.

Preferably, step S4 establishes a distributed electric vehicle multi-agent coordinated control strategy based on cooperative game, comprising the following steps:

Step S41: Establish the cost function of each agent:

Since cooperative game cannot handle system disturbance and uncertainty well, we first assume that the system disturbance problem described in step S5 is ignored and establish the human-vehicle dynamics system based on cooperative game as follows:

Among them, Δx is the state error of the system, Δτ _h is the driver's control error, Δu is the vehicle's execution error, and Δz is the system output error;

In game theory, there are six agents as participants, and their utility can be expressed as a cost function. Therefore, the cost function for infinite horizon is defined as:

Wherein, n = {2 3 4 5 6} represents the AFS steering assist motor, the left front wheel hub motor, the left rear wheel hub motor, the right front wheel hub motor, and the right rear wheel hub motor, respectively; Φ _n represents the weight matrix of agent n, r _n is the weight output by agent n, J _n is the cost function of agent n, J _h is the cost function of the driver agent, Φ _h is the weight matrix of the driver agent, and r _h is the weight output by the driver agent;

According to the interactive paradigm of cooperative games, all agents have a common goal, which can be described by a global cost function, which can be expressed as:

Among them, ρ _h is the task weight of the driver agent in the whole game, ρ _n is the task weight of the vehicle execution layer agent in the whole game,

Therefore, the global objective cost function of the six agents is:

Where R _h , R _n and Φ _hk , Φ _nk represent the weight matrix of system control input and state error respectively;

Step S42: Optimization problem of distributed electric vehicle intelligent body:

The optimization problem of 6 agents can be expressed as:

in, _{Pi_h} and _{Pi_n} represent the Lyapunov solution feedback control rate respectively;

Step S43: Solving the dynamic coordination control rate of distributed electric vehicles based on cooperative game

The cooperative game is achieved by solving multiple coupled optimization problems at the same time to obtain the global optimal solution. According to the cost function of the six agents established in step S41 and the expression of the optimization problem of the six agents established in step S42, the expression of the model predictive control algorithm can be established:

in, Φ _k ,Φ _k represents the weight matrix of system state error;

By using the QR algorithm to solve the optimal solution of the system, we can get the optimal solution of the six agents:

Among them, Ψ _h and Ψ _n represent the optimal solution matrices of the driver control input, the auxiliary steering torque control input, the left front wheel control input, the left rear wheel control input, the right front wheel control input, and the right rear wheel control input, respectively.

Preferably, step S5 aims at the problem that cooperative game cannot eliminate system disturbances, establishes a robust compensation control strategy, and realizes safe and efficient dynamic coordinated control of the six intelligent agents of the distributed electric vehicle, including:

The cooperative game theory cannot effectively eliminate the system disturbance and uncertainty problems. In order to ensure the stability of the dynamic coordinated control of distributed electric vehicles, according to the control rate of the intelligent agent system obtained by step (4), a robust H _∞ compensation strategy is further established for the control outputs of the auxiliary steering agent, the left front wheel agent, the left rear wheel agent, the right front wheel agent, and the right rear wheel agent. The following is obtained:

Among them, ε is the stability parameter of the closed-loop system under the given H _∞ performance, and z(k) and w(k) are affected by the system dominance;

Therefore, the optimal control rate of the system can be rewritten as:

Among them, Δτ _rh is the driver's control torque to eliminate system disturbances, Δu _rn is the control torque of the agent to eliminate system disturbances, Δu _cn is the system disturbance compensation control rate of the agent, and Θ _n is the system disturbance compensation matrix of the agent.

The above design fully considers the execution interaction between various intelligent agents and the needs of intelligent coordinated control of distributed electric vehicles. It can make real-time dynamic adjustments to the intelligent agents as needed, effectively ensuring the driving safety of the vehicle.

Compared with the prior art, the present invention has the following beneficial effects:

When the distributed electric vehicle multi-agent coordinated control is performed, the distributed electric vehicle actuators are finely divided into six major agents, namely, the driver, the steering assist control, the left front wheel hub motor, the left rear wheel hub motor, the right front wheel hub motor, and the right rear wheel hub motor, and the six agents are converted into Agent language for description. The system integrated minimum control unit is constructed through dynamic modeling. The vehicle model is linearized based on the T-S fuzzy theory to solve the vehicle nonlinear problem caused by the time-varying longitudinal vehicle speed. The distributed electric vehicle is dynamically coordinated and controlled using cooperative game theory, which has good interaction performance among the agents and strong real-time performance. The coordinated control performance of the vehicle's execution components is safer and more stable than that of the traditional method, and can provide strong technical support for the efficient and intelligent coordinated control of distributed electric vehicles. At the same time, the optimization design based on the robust compensation algorithm greatly eliminates the influence of system disturbance and uncertainty. This design can effectively eliminate the functional coupling conflict of the execution unit and has strong practicality.

The six agents set in the distributed electric vehicle multi-agent coordinated control method based on cooperative game designed by the present invention can all independently and controllably complete the corresponding instruction operations, among which a single agent is the micro level of the multi-agent system, which is responsive, autonomous and flexible; and the relationship between agents constitutes the macro level of the multi-agent system, which can achieve higher flexibility, environmental adaptability and scalable comprehensive functions through organization and collaboration at all levels. This coordinated control idea is very suitable for the intelligent control of distributed electric vehicles.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of an intelligent assisted steering system of the present invention;

FIG2 is a logic diagram of information interaction among six intelligent agents based on cooperative game of the present invention;

FIG3 is a diagram showing the architecture of a distributed electric vehicle multi-agent coordination control method based on cooperative game according to the present invention;

Detailed ways

The specific implementation of the present invention is described below in conjunction with the accompanying drawings so that those skilled in the art can better understand the present invention. It should be noted that in the description of the present invention, when the current known functions and actual detailed descriptions may dilute the main content of the present invention, these descriptions will be ignored here, and the distributed electric vehicle in this application refers to a distributed drive electric vehicle.

As shown in FIG3 , a distributed electric vehicle multi-agent coordinated control method based on cooperative game includes the following steps:

Step S1: Divide the distributed electric vehicle agent with the actuator as the smallest unit, divide the functions of each subsystem of the distributed electric vehicle into intelligent control areas, establish the steering system intelligent control area, and divide it into the driver agent and the steering auxiliary control agent; in the drive/brake system intelligent control area, since the distributed electric vehicle uses the wheel hub motor to control the vehicle's drive/brake, and the four wheels are independently controllable, establish the left front wheel agent, the left rear wheel agent, the right front wheel agent, and the right rear wheel agent;

Use Agent language to describe the minimum execution unit:

Combined with the intelligent control area established in step S1, the six intelligent agents are described in detail, among which the driver intelligent agent can accurately characterize the driver's behavioral characteristics, including the driver's thinking characteristics and manipulation characteristics; the auxiliary steering intelligent agent can autonomously and intelligently steer, and has active steering control functions as well as high execution accuracy and fast response characteristics; the drive/brake system control area can perform distributed collaborative control functions on the vehicle, and each intelligent agent is independently controllable, with efficient, intelligent and collaborative control characteristics; each intelligent agent is a micro-level of multi-agents, with responsiveness, autonomy and flexibility; the relationship between the six intelligent agents established in this application constructs a macro-level of multi-agents, which can realize the intelligent and scalable comprehensive functions of distributed electric vehicles through organizational collaboration at all levels. This intelligent coordinated control idea has a good promoting effect on the development of distributed electric vehicles.

Step S2: Describe the six constructed agents using dynamics language.

For the six agents defined in step S1, namely, the driver agent, the steering assist control agent, the left front wheel agent, the left rear wheel agent, the right front wheel agent and the right rear wheel agent, a dynamic model representing each agent is constructed to establish a vehicle system dynamic model, including:

Step S21: Establish the dynamic equation that characterizes the driver agent:

The dynamic expression of the hand torque input of the driver agent is:

Among them, τ _h is the steering torque input of the driver agent, J _h is the equivalent inertia of the steering system, B _h is the damping of the steering system, τ _f is the friction torque of the steering system, τ _e is the steering assist torque provided by the steering system motor, δ is the steering wheel angle, and sign() is a sign function used to indicate the positive and negative signs of the parameters;

Step S22: Establish the dynamic equation of the auxiliary steering agent. As shown in Figure 1, there is a mechanism coupling between the auxiliary steering agent and the driver agent. The driver agent can directly affect the auxiliary steering agent through control parameters, and the auxiliary steering agent can also directly affect the driver agent. For the distributed electric vehicle steering system, the dynamic characterization of its auxiliary steering agent can use the control torque input of the steering actuator motor as the key parameter to characterize the auxiliary steering agent, and its dynamic equation is expressed as:

τ _e =I _e K _t η

(Formula 2)

Among them, _τe is the steering torque output of the auxiliary steering intelligent body, _Ie is the real current of the steering motor, _Kt is the torque coefficient of the steering motor, and η is the mechanical efficiency;

S23: Establish the dynamic model of the hub motor intelligent agent. Since the performance of the four hub motors of the distributed electric vehicle is the same, the dynamic model of the hub motor intelligent agent can be expressed as;

τ _ij ＝τ _Lij +B _ij +J _ij ω _ij

(Formula 3)

Where, ij = {ff, fr, rf, rr}, respectively represents the left front, left rear, right front, and right rear. ω is the angular velocity of the hub motor, τ, τ _L represent the output torque and load torque of the hub motor, J is the moment of inertia of the hub motor, and B is the damping coefficient;

Distributed electric vehicle experimental data collection and experimental data processing:

(I) Experimental data collection. Using on-board sensors, the experimental data of the six intelligent bodies of the distributed electric vehicle are collected in real time, including the driver's hand torque, steering control torque, drive pedal travel, brake pedal travel, steering wheel angle, steering wheel speed, wheel hub motor output torque, vehicle speed, center of mass sideslip angle, and yaw angular velocity;

(II) Experimental data processing. Since data is collected through different sensors, some of the collected data is not easy to understand and observe intuitively, so it is necessary to convert the data units, convert the steering wheel angle and speed from radians to degrees, and convert the speed from m/s to km/h. The data is divided into three categories: driver agent data, auxiliary steering agent data, and hub motor agent data, so as to facilitate the input into the six agents for identification. The data in each group of data is segmented, and each time period represents the manipulation behavior of the functional agent within a period of time. For each data segment, an adaptive Kalman filter algorithm is used to remove the outliers in each data segment; when the prediction result is that the error increases, the Taylor series expansion algorithm under the current number of iterations is determined to be a divergent state, and the measured value of the current state is regarded as a bad point and removed.

Step S24: Establishing a distributed electric vehicle system model including the driver agent

Assuming that the tire slip angle of the vehicle is small, the vehicle dynamics model includes longitudinal motion, lateral motion and yaw motion. The distributed electric vehicle system dynamics model can be expressed as:

in, δ is the steering wheel angle, is the steering wheel angular velocity, β is the vehicle center of mass sideslip angle, is the vehicle yaw rate, Y is the vehicle lateral displacement, ω _r is the vehicle yaw angle, v _x is the vehicle longitudinal velocity, _vy is the vehicle lateral velocity, A is the vehicle state system matrix, B is the control input system matrix, B _h is the driver control input system matrix, u = [τ _e τ _ff τ _fr τ _rf τ _rr ], τ _ff is the vehicle left front wheel driving torque, τ _fr is the left rear wheel driving torque, τ _rf is the right front wheel driving torque, τ _rr is the right rear wheel driving torque, H is the system disturbance matrix, w is the system disturbance, z is the system output, and C is the system output matrix.

Step S3: Use T-S fuzzy theory to linearize the vehicle system and establish the system state equation after linearization, including:

S31: Set the upper limit of the longitudinal speed of distributed electric vehicles to The lower boundary is

S32: Establish a human-vehicle system model based on T-S fuzzy system:

Taking into account the nonlinear problem caused by the longitudinal speed on the human-vehicle system, the linearized human-vehicle system model is established based on T-S fuzzy theory:

Where k is the current time of the system, υ is the advance variable, ζ _i (υ) is the normalized membership function of the TS fuzzy model, _{Ai_b} is the system state matrix after TS fuzzy processing, B _{h_h} is the driver control matrix after discretization in the system, B _d is the vehicle control matrix after discretization in the system, and H _d is the interference matrix after discretization in the system.

The distributed electric vehicle multi-agent coordinated control strategy based on robust cooperative game includes the following contents:

Step S4: Establish a distributed electric vehicle multi-agent coordinated control strategy based on cooperative game, and obtain the reference yaw rate and reference center of mass sideslip angle required by the patent by linearizing the vehicle model; the reference yaw rate dynamics is represented as:

in, is the reference yaw rate required by the patent, L is the vehicle wheelbase length, K is the vehicle stability coefficient, μ is the road adhesion coefficient, and g is the gravity parameter;

The reference center of mass sideslip angle dynamics is characterized by:

Among them, β _tar is the vehicle reference center of mass sideslip angle, b is the distance from the vehicle's rear axle to the center of mass, a is the distance from the vehicle's front axle to the center of mass, and _{Cr is} the cornering stiffness of the vehicle's rear wheel.

The reference yaw angle isomechanical characterization is:

Among them, _yp is the lateral deviation at the preview point, and _xp is the longitudinal deviation at the preview point.

The reference steering wheel angle dynamics is characterized as:

Where δ _tar is the reference steering wheel angle, l is the preview distance based on the current speed, N _ss is the steering system transmission ratio, t _pr is the preview time, and y is the lateral displacement of the current vehicle.

The reference steering wheel angle angular velocity dynamics is characterized by:

in, is the reference steering wheel angular velocity, _Ksr is the steering column drag coefficient, and _Kp is the active steering torque gain coefficient.

Step S41: Establish the cost function of each agent. Since cooperative game cannot handle system disturbance and uncertainty well, first assume that the system disturbance problem described in step S5 is ignored, and establish the human-vehicle dynamics system based on cooperative game as follows:

Among them, Δx is the state error of the system, Δτ _h is the driver's control error, Δu is the vehicle's execution error, and Δz is the system output error;

In game theory, there are six agents as participants, and their utility can be expressed as a cost function. Therefore, the cost function for infinite horizon is defined as:

Among them, n = {2 3 4 5 6} represents the AFS steering assist motor, the left front wheel hub motor, the left rear wheel hub motor, the right front wheel hub motor, and the right rear wheel hub motor respectively, _Φn represents the weight matrix of agent n, _rn is the weight output by agent n, _Jn is the cost function of agent n, _Jh is the cost function of the driver agent, _Φh is the weight matrix of the driver agent, and _rh is the weight output by the driver agent.

As shown in Figure 2, according to the interactive paradigm of cooperative games, all agents have a common goal, which can be described by a global cost function, which can be expressed as:

Among them, ρ _h is the task weight of the driver agent in the whole game, ρ _n is the task weight of the vehicle execution layer agent in the whole game,

Therefore, the global objective cost function of the six agents is:

Where R _h , R _n and Φ _hk , Φ _nk represent the weight matrix of system control input and state error respectively;

Step S42: Optimization problem of distributed electric vehicle agents. The optimization problem of 6 agents can be expressed as:

in, _{Pi_h} and _{Pi_n} represent the Lyapunov solution feedback control rate respectively.

Step S43: Solving the dynamic coordination control rate of distributed electric vehicles based on cooperative game. Cooperative game is achieved by solving multiple coupled optimization problems at the same time to obtain the global optimal solution. According to the cost function of the established 6 intelligent agents and the expression form of the optimization problem of the established 6 intelligent agents, the expression form of the model predictive control algorithm can be established:

in, Φ _k ,Φ _k represents the weight matrix of system state error;

By using the QR algorithm to solve the optimal solution of the system, we can get the optimal solution of the six agents:

Among them, Ψ _h and Ψ _n represent the optimal solution matrices of the driver control input, the auxiliary steering torque control input, the left front wheel control input, the left rear wheel control input, the right front wheel control input, and the right rear wheel control input, respectively.

Step S5 aims to solve the problem that cooperative game cannot eliminate system disturbances, establish a robust compensation control strategy, and realize efficient and safe dynamic coordinated control of the six intelligent agents of distributed electric vehicles, including:

The cooperative game theory cannot effectively eliminate the system disturbance and uncertainty problems. In order to ensure the stability of the dynamic coordinated control of distributed electric vehicles, according to the control rate of the multi-agent obtained by step (7), a robust H _∞ compensation strategy is further established for the control output of the auxiliary steering agent, the left front wheel agent, the left rear wheel agent, the right front wheel agent, and the right rear wheel agent. The following is obtained:

Among them, ε is the stability parameter of the closed-loop system under a given _H∞ performance, and z(k) and w(k) are affected by the system.

Therefore, the optimal control rate of the system can be rewritten as:

Among them, Δτ _rh is the driver's control torque to eliminate system disturbances, Δu _rn is the vehicle agent's control torque to eliminate system disturbances, Δu _cn is the system disturbance compensation control rate of the vehicle agent, and Θ _n is the system disturbance compensation matrix of the vehicle agent, completing the dynamic coordination of distributed electric vehicle multi-agents.

As shown in Figure 3, this example designs a distributed electric vehicle multi-agent coordinated control method based on cooperative game. In the process of multi-agent dynamic coordinated control, the Agent language is first used to intelligently divide the functional areas of the distributed electric vehicle. The interactivity and control accuracy of the multi-agent coordinated control can be guaranteed based on the real-time collected data. The control input of each agent is intelligently coordinated through the cooperative game control algorithm, and the disturbance and uncertainty in the distributed electric vehicle control process are eliminated based on the robust compensation control strategy. This design effectively avoids the problem of distributed electric vehicles being difficult to coordinate and control when the interaction between vehicle agents is imperfect and each agent acts independently.

The advantages of this example are:

The distributed electric vehicle dynamic coordination control method adopted in this method can effectively identify the intentions of each intelligent agent and has perfect information interaction, which is of great significance for the intelligent, safe and efficient driving of distributed electric vehicles. The coordination control algorithm based on cooperative game has the advantages of intelligent agent interactive control coupling and convenient real-time control. The robust compensation control method eliminates the influence of system disturbances and interference on control. The coordinated control output based on robust cooperative game is safer, more reliable and applicable to various driving conditions than traditional methods, and can meet the needs of intelligent coordination control of distributed electric vehicles.

The above description is only a preferred feasible implementation example of the present invention, and does not limit the scope of rights of the present invention. All equivalent structural changes made using the contents of the present invention and the drawings are included in the scope of rights of the present invention.

Claims (4)

1. A distributed electric automobile multi-agent cooperative control method based on cooperative game is characterized by comprising the following steps:

Step S1: aiming at the functions of each subsystem of the distributed electric automobile, an intelligent control area of a steering system is established, and the intelligent control area is divided into a driver intelligent body and a steering auxiliary control intelligent body; in the intelligent control area of the driving/braking system, the distributed electric automobile controls the driving/braking of the automobile for an in-wheel motor, and four wheels are independently controllable, so that a left front wheel intelligent body, a left rear wheel intelligent body, a right front wheel intelligent body and a right rear wheel intelligent body are established;

Step S2: aiming at the driver intelligent agent, the steering auxiliary control intelligent agent, the left front wheel intelligent agent, the left rear wheel intelligent agent, the right front wheel intelligent agent and the right rear wheel intelligent agent which are defined in the step S1, 6 intelligent agents are combined, and a dynamics model for representing each intelligent agent is built, so that a vehicle system dynamics model is built;

step S3: carrying out linearization treatment on a vehicle system by adopting a T-S fuzzy theory, and establishing a system state equation after linearization treatment;

Step S4: establishing a distributed electric vehicle multi-agent coordination control strategy based on cooperative game;

Step S5: aiming at the problem that the disturbance of the system cannot be eliminated in the cooperative game, a robust compensation control strategy is established, and the efficient and safe dynamic coordination control of 6 intelligent bodies of the distributed electric automobile is realized;

Step S2 builds a kinetic model characterizing each agent for the 6 agents defined in step S1, thereby building a vehicle system kinetic model, and comprises the following steps:

Step S21: establishing a dynamics equation for representing the intelligent agent of the driver:

Kinetic expression of the hand torque input of the driver agent:

Wherein τ _h is the steering torque input of the driver agent, J _h is the equivalent inertia of the steering system, B _h is the steering system damping, τ _f is the steering system friction torque, τ _e is the steering assist torque provided by the steering system motor, δ is the steering wheel angle, sign () is the sign function for indicating the sign of the parameter;

Step S22: establishing a dynamics equation of the auxiliary steering agent:

The kinetic equation expression of the auxiliary steering agent is:

τ _e＝I_eK_t η equation 2

Wherein τ _e is the steering torque output of the auxiliary steering agent, I _e is the actual current of the steering motor, K _t is the torque coefficient of the steering motor, and η is the mechanical efficiency;

step S23: dynamic model of hub motor intelligent body is established

Because the four hub motors of the distributed electric automobile have the same performance, the dynamic model of the hub motor intelligent body can be expressed as follows;

τ _ij＝τ_Lij+B_ij+J_ijω_ij equation 3

Wherein ij= { ff, fr, rf, rr }, respectively represent left front, left back, right front, right back, ω is the hub motor angular velocity, τ, τ _L are respectively represented as the hub motor output torque and the load torque, J is the hub motor moment of inertia, and B is the damping coefficient;

Step S24: establishing a distributed electric automobile system model covering a driver agent

Setting the tire slip angle of the vehicle to be small, the vehicle dynamics model includes longitudinal movement, transverse movement and yaw movement, and the distributed electric vehicle system dynamics model can be expressed as:

Wherein, Delta is the steering wheel angle of the steering wheel,Is the steering wheel turning speed, beta is the vehicle mass center slip angle,The yaw rate of the vehicle, Y is the lateral displacement of the vehicle, omega _r is the yaw angle of the vehicle, v _x is the longitudinal speed of the vehicle, v _y is the lateral speed of the vehicle, A is the vehicle state system matrix, B is the control input system matrix, B _h is the driver control input system matrix, u= [ tau _e τ_ff τ_fr τ_rf τ_rr],τ_ff ] is the left front wheel driving moment of the vehicle, tau _fr is the left rear wheel driving moment, tau _rf is the right front wheel driving moment, tau _rr is the right rear wheel driving moment, H is the system disturbance matrix, w is the system disturbance, z is the system output, and C is the system output matrix.

2. The cooperative game-based distributed electric vehicle multi-agent cooperative control method according to claim 1 is characterized in that the step S3 adopts a T-S fuzzy theory to carry out linearization treatment on a vehicle system, and establishes a linearized system state equation, and the method comprises the following steps:

step S31: setting the upper boundary of the longitudinal speed of the distributed electric automobile as Lower boundary is

Step S32: establishing a human-vehicle system model based on a T-S fuzzy system:

Comprehensively considering the nonlinear problem caused by the longitudinal speed to the human-vehicle system, and establishing a linearized human-vehicle system model based on the T-S fuzzy theory as follows:

Wherein k is the current moment of the system, v is an advance variable, ζ _i (v) is a normalized membership function of a T-S fuzzy model, A _{i_b} is a system state matrix after T-S fuzzy processing, B _{h_h} is a driver control matrix after discretization in the system, B _d is a vehicle control matrix after discretization in the system, and H _d is an interference matrix after discretization in the system.

3. The distributed electric vehicle multi-agent coordination control method based on the cooperative game according to claim 1, wherein step S4 establishes a distributed electric vehicle multi-agent coordination control strategy based on the cooperative game, and the method comprises the following steps:

step S41: establishing a cost function of each agent:

Aiming at the problem that the cooperative game cannot well process system interference and uncertainty, firstly, the system disturbance problem described in the step S5 is ignored, and a man-vehicle dynamics system based on the cooperative game is established as follows:

Wherein Δx is the state error amount of the system, Δτ _h is the manipulation error amount of the driver, Δu is the execution error amount of the vehicle, and Δz is the error amount of the system output;

in game theory 6 agents are participants whose utility can take the form of a cost function, thus defining an infinite field of view as:

wherein n= { 23 45 6} represents an AFS steering auxiliary motor, a left front wheel hub motor, a left rear wheel hub motor, a right front wheel hub motor, a right rear wheel hub motor, Φ _n represents a weight matrix of the agent n, r _n represents a weight output by the agent n, J _n represents a cost function of the agent n, J _h represents a cost function of the driver agent, Φ _h represents a weight matrix of the driver agent, and r _h represents a weight output by the driver agent;

according to the interactive paradigm of cooperative gaming, all agents have a common goal, which can be described by a global cost function, which can be expressed as:

Wherein ρ _h is the task weight of the driver agent in the whole game, ρ _n is the task weight of the vehicle execution layer agent in the whole game,

Thus, the global objective cost function for 6 agents is:

Wherein R _h,R_n and phi _hk,Φ_nk are respectively represented as a weight matrix of system control input weights and state errors;

step S42: optimization problem of distributed electric automobile agent:

The optimization problem for 6 agents can be expressed as:

Wherein, P _{i_h} and P _{i_n} represent lyapunov solution feedback control rate, respectively;

step S43: dynamic coordination control rate solving method for distributed electric automobile based on cooperative game

The cooperative game is implemented by simultaneously solving a plurality of coupling optimization problems to obtain a global optimal solution, and according to the cost functions of the 6 agents established in the step S41 and the expression forms of the optimization problems of the 6 agents established in the step S42, the expression forms of the model predictive control algorithm can be established and calculated:

Wherein, Phi _h,Φ_n is represented as a weight matrix of system state errors;

The optimal solution of the system is solved through the QR algorithm, and the optimal solution of 6 intelligent agents can be obtained as follows:

Wherein, ψ _h and ψ _n represent the optimal solution matrices of the driver control input, the auxiliary steering torque control input, the left front wheel control input, the left rear wheel control input, the right front wheel control input, the right rear wheel control input, respectively.

4. The distributed electric vehicle multi-agent cooperative control method based on the cooperative game according to claim 1, wherein step S5 establishes a robust compensation control strategy for the problem that the cooperative game cannot eliminate system disturbance, and realizes efficient and safe dynamic cooperative control of 6 agents of the distributed electric vehicle, and the method comprises the following steps:

The problem of system disturbance and uncertainty can not be effectively eliminated by the cooperative game theory, and in order to ensure the stability of dynamic coordination control of the distributed electric automobile, the control rate of the multi-agent obtained by solving according to claim 3 further establishes a robust H _∞ compensation strategy for the control output of the auxiliary steering agent, the left front wheel agent, the left rear wheel agent, the right front wheel agent and the right rear wheel agent, and can obtain:

Where ε is the stability parameter of the closed loop system given H _∞ performance, z (k) and w (k) are affected by the system;

therefore, the optimal control rate of the system can be rewritten as:

Wherein, deltaτ _rh is the control moment of the driver for eliminating the system disturbance, deltau _rn is the control moment of the agent for eliminating the system disturbance, deltau _cn is the system disturbance compensation control rate of the agent, and Θ _n is the system disturbance compensation matrix of the agent.