Non-inflatable electric wheel matching vehicle control method based on multi-agent graph theory

Abstract

The invention discloses a non-inflatable electric wheel matching vehicle control method based on a multi-agent graph theory, and the method comprises the following steps: 1, building a mathematical model of a non-inflatable electric wheel according to the longitudinal, lateral and vertical mechanical characteristics of the non-inflatable electric wheel; step 2, based on the mathematical model of the non-inflatable electric wheel established in the step 1, constructing a non-linear two-degree-of-freedom dynamical model for vehicle control; step 3, based on a multi-agent graph theory, regarding four non-inflatable electric wheels of the vehicle as four agents, considering state coupling among the agents, and designing a corresponding vehicle motion control model; and 4, based on the vehicle motion control model designed in the step 3, performing optimization solution by adopting a model prediction control algorithm to obtain an additional front wheel steering angle and an additional yawing moment for vehicle control. According to the method, the characteristics of the non-inflatable electric wheel in vehicle application are considered, matching of whole vehicle motion control is facilitated, and the maneuvering performance and the stability performance of the vehicle are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electric vehicle chassis control, in particular to a non-inflatable electric wheel matching vehicle control method based on multi-agent graph theory. BACKGROUND

[0002] Four-wheel independent drive-independent steering vehicles based on hub motors and steering motors represent an advanced vehicle architecture that integrates multiple actuators and control degrees of freedom. This design allows each wheel to be independently driven, braked, and steered, offering unique advantages in coordinating and integrating various vehicle active safety technologies. As a key direction for the development of future intelligent electric vehicles, 4WIS-4WID vehicles, through their innovative design philosophy, not only improve the safety and stability of driving, but also bring new possibilities for driving experience.

[0003] However, for independent drive and independent steering vehicles, the requirements for wheel matching chassis control systems are also increasing. Traditional pneumatic tires have inherent limitations such as easy tire blowout, uneven wear, and the need for regular maintenance, which to some extent limit the further improvement of electric vehicle performance. Non-inflatable electric wheels, as an emerging technology, have attracted widespread attention due to their need for inflation, low maintenance costs, and higher safety performance. However, the unique mechanical properties of non-inflatable electric wheels pose new challenges to vehicle control, especially in terms of dynamic stability and maneuvering flexibility. Traditional vehicle dynamics models are usually based on the assumption of pneumatic tires and are difficult to accurately reflect the complex behavior of non-inflatable electric wheels. Multi-agent systems provide an effective framework for handling coordination problems in distributed systems, while graph theory can be used to describe the interaction structure between agents, providing a new approach to solving the integration of non-inflatable electric wheels in vehicle control. SUMMARY

[0004] The purpose of the present application is to provide a non-inflatable electric wheel matching vehicle control method based on multi-agent graph theory, by establishing a high-fidelity wheel model and a nonlinear two-degree-of-freedom dynamics model for vehicle control, and considering the state coupling between agents, a corresponding vehicle motion control model is designed. Finally, the model predictive control algorithm is used to optimize the additional wheel angle and additional yaw moment to maximize the vehicle handling performance and stability performance.

[0005] To achieve the above purpose, the present application adopts the following technical solutions:

[0006] A non-inflatable electric wheel matching vehicle control method based on multi-agent graph theory, comprising the following steps:

[0007] Step 1, according to the longitudinal, lateral and vertical force mechanics characteristics of the non-pneumatic electric wheel, a mathematical model of the non-pneumatic electric wheel is established;

[0008] Step 2, based on the mathematical model of the non-pneumatic electric wheel established in step 1, a nonlinear two-degree-of-freedom dynamics model oriented to vehicle control is constructed;

[0009] Step 3, based on multi-agent graph theory, the four non-pneumatic electric wheels of the vehicle are regarded as four agents, and the state coupling between the agents is considered, and a corresponding vehicle motion control model is designed;

[0010] Step 4, based on the vehicle motion control model designed in step 3, a model predictive control algorithm is used for optimization and solution to obtain additional front wheel steering angle and additional yaw moment for vehicle control.

[0011] Further, the non-pneumatic electric wheel comprises an outer wheel set, an inner wheel set and a driving motor;

[0012] The outer wheel set comprises an elastic ring, a buckle, a pin ear and a rubber layer;

[0013] The inner wheel set comprises a hinge connected with the pin ear and a hub;

[0014] The driving motor is a hub motor embedded in the hub and connected with the inner wheel set.

[0015] Further, in step 1, the mathematical model of the non-pneumatic electric wheel is represented as follows:

[0016] A function relationship between the vertical force of the non-pneumatic electric wheel and the loading deflection is established:

[0017] F z =k z (x z -λ)=k z x z -k z λ

[0018] Wherein, F z is the vertical force of the non-pneumatic electric wheel, k z is the fitted vertical stiffness, x z is the wheel deflection, and λ is the offset;

[0019] The longitudinal force F x of the non-pneumatic electric wheel is fitted by a BP neural network model, and a nonlinear relationship between the tire longitudinal force and the slip ratio and the vertical load is established as:

[0020]

[0021] Wherein, Net_F x(*) is the BP neural network model of longitudinal force of non-pneumatic electric wheel, W x1 is the weight matrix between input layer and hidden layer of network, b x1 is the bias vector between input layer and hidden layer of network, W x2 is the weight matrix between hidden layer and output layer of network, b x2 is the bias vector between hidden layer and output layer of network; s is the slip ratio of non-pneumatic electric wheel, which is expressed as:

[0022]

[0023] wherein, i represents the serial number of four non-pneumatic electric wheels, i = 1, 2, 3, 4, respectively representing the left front wheel, the right front wheel, the left rear wheel and the right rear wheel of the vehicle; v i is the longitudinal velocity at the center of the i-th wheel; ω i is the angular velocity of the i-th wheel; R is the rolling radius of the wheel;

[0024] The lateral force of the non-pneumatic electric wheel is fitted by the neural network model, and the nonlinear relationship of the tire lateral force and the side slip angle and the vertical load is obtained as:

[0025]

[0026] wherein, Net_F y (*) is the BP neural network model of lateral force of non-pneumatic electric wheel, W y1 is the weight matrix between input layer and hidden layer of network, b y1 is the bias vector between input layer and hidden layer of network, W y2 is the weight matrix between hidden layer and output layer of network, b y2 is the bias vector between hidden layer and output layer of network; α is the side slip angle of non-pneumatic electric wheel, which is expressed as:

[0027]

[0028] wherein, δ i is the steering angle of the i-th non-pneumatic electric wheel, β is the side slip angle of the vehicle mass center, l f is the distance from the vehicle mass center to the front axle, ω is the yaw angular velocity of the vehicle, v x is the longitudinal velocity of the vehicle.

[0029] Further, in step 2, the nonlinear two-degree-of-freedom dynamics model facing the whole vehicle control is expressed as:

[0030]

[0031] wherein, m is the vehicle body mass, a yis the lateral acceleration of the vehicle; F yi where i = 1, 2, 3, 4, are the lateral forces of the vehicle's front left wheel, front right wheel, rear left wheel and rear right wheel, respectively; F xi where i = 1, 2, 3, 4, are the longitudinal forces of the vehicle's front left wheel, front right wheel, rear left wheel and rear right wheel, respectively; I z is the moment of inertia of the vehicle's yaw motion, ω is the yaw angular velocity of the vehicle, l f and l r are the distances from the front / rear axles to the center of mass, R is the rolling radius of the wheels;

[0032] The lateral acceleration of the vehicle is expressed as:

[0033]

[0034] where, is the lateral velocity change rate of the vehicle, ω is the yaw angular velocity of the vehicle, v x is the longitudinal velocity of the vehicle.

[0035] Further, in step 3, the four non-inflatable electric wheels are regarded as 4 multi-agents, i.e. 4 vertices, to form a graph G4; the adjacency matrix A, the degree matrix D, the Laplacian matrix L, and the incidence matrix B are introduced;

[0036] The relevant matrices in the established graph are:

[0037]

[0038] Thus, the side slip angle of the wheel is expressed as:

[0039]

[0040] where Δε i The state of the other wheel agents coupled to the wheel agent i,

[0041]

[0042] Let β i , ω i be the lateral force F yi of the i-th wheel acting alone, the center of mass side slip angle and yaw angular velocity generated at the center of mass of the whole vehicle, through vector transformation, satisfy β = β1+ β2+ β3+ β4, ω = ω1+ ω2+ ω3+ ω4, and their values are:

[0043]

[0044] Thus, the nonlinear two-degree-of-freedom dynamic model is expressed as:

[0045]

[0046] Where, k i For the i-th non-pneumatic electric wheel in a y Lateral stiffness under conditions <0.4g, l1=l2=l f l3 = l4 = l r , e1=e2=1, e3=e4=-1, T i δ is the driving torque of the i-th non-pneumatic electric wheel; i Let be the steering angle of the i-th non-pneumatic electric wheel;

[0047] Therefore, the vehicle control model based on multi-agent graph theory for matching non-pneumatic electric wheels with the overall vehicle control is obtained as follows:

[0048]

[0049] The state-space equation of the model is:

[0050]

[0051] in u * =Δε i ,

[0052] Furthermore, in step 4, the state-space equation from step 3 is discretized:

[0053]

[0054] in,

[0055] Let N c To control the time domain, N p To predict the time domain, the state variables and control variables recursively derived to time N in the future are:

[0056] {x i,k (t)=x i (t+k|t,k=1,2,...N}

[0057] {u i,k (t)=u i (t+k|t),k=0,...,N-1}

[0058] The predicted output is:

[0059] Y i (k+1|k)=S x x i (k)+S u u i (k)

[0060] wherein,

[0061] The intention of the whole vehicle control is to maintain the maneuverability and stability, and the control target function is designed as follows:

[0062]

[0063] wherein, Q1 is a weighted matrix of the prediction output and the reference output error; R1 is a weighted matrix of the control amount; is a relaxation factor; and p is a relaxation factor penalty coefficient, used to constrain the relaxation degree and avoid the infinite increase of the relaxation factor;

[0064] Considering the constraints of the actuator and the requirements for the smoothness of the control amount, the control amount, the control increment and the output amount are limited, and at the same time, the relaxation factor is added to soften the constraints to ensure that the optimization problem has a feasible solution:

[0065]

[0066] A quadratic programming algorithm is used to calculate the control sequence of each control period, wherein the first element is the output of the feedback control, and the final required additional steering angle and additional yaw moment are obtained.

[0067] Beneficial effects: the application provides a non-inflatable electric wheel matching whole vehicle control method based on multi-agent graph theory, a high-fidelity wheel model and a nonlinear two-degree-of-freedom dynamics model for whole vehicle control are established, each non-inflatable electric wheel is regarded as an independent agent, the state coupling between the agents is considered, and a corresponding vehicle motion control model is designed. Finally, a model predictive control algorithm is used to optimize and solve the additional front wheel steering angle and the additional yaw moment, so as to maximize the vehicle handling performance and stability performance. The application fully utilizes the unique advantages of the four-wheel independent drive-independent steering vehicle, considers the characteristics of the non-inflatable electric wheel in vehicle application, is beneficial to matching the whole vehicle motion control, and improves the safety and driving experience of the vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0068] Figure 1 is a control method flowchart of the application. DETAILED DESCRIPTION

[0069] The application will be further described below in combination with the drawings.

[0070] As shown in the drawings, Figure 1 a non-inflatable electric wheel matching whole vehicle control method based on multi-agent graph theory, comprising the following steps:

[0071] Step 1: according to the longitudinal, lateral and vertical mechanical properties of the non-inflatable electric wheel, a mathematical model of the non-inflatable electric wheel is established;

[0072] Wherein, the non-pneumatic electric wheel comprises an outer wheel set, an inner wheel set and a driving motor; the outer wheel set comprises an elastic ring, a buckle, a pin ear and a rubber layer; the inner wheel set comprises a hinge connected with the pin ear and a hub; the driving motor is a hub motor embedded in the hub and connected with the inner wheel set.

[0073] The mathematical model of the non-pneumatic electric wheel is expressed as follows:

[0074] The functional relationship between the vertical force and the load deflection of the non-pneumatic electric wheel is established:

[0075] F z =k z (x z -λ)=k z x z -k z λ

[0076] Wherein, F z is the vertical force of the non-pneumatic electric wheel, k z is the fitted vertical stiffness, x z is the wheel deflection, and λ is the offset;

[0077] The longitudinal force F x of the non-pneumatic electric wheel is fitted by the BP neural network model, and the nonlinear relationship between the tire longitudinal force and the slip ratio and the vertical load is established as:

[0078]

[0079] Wherein, Net_F x (*) is the BP neural network model of the non-pneumatic electric wheel longitudinal force, W x1 is the weight matrix between the network input layer and the hidden layer, b x1 is the bias vector between the network input layer and the hidden layer, W x2 is the weight matrix between the network hidden layer and the output layer, b x2 is the bias vector between the network hidden layer and the output layer; s is the slip ratio of the non-pneumatic electric wheel, which is expressed as:

[0080]

[0081] Wherein, i represents the serial number of the four non-pneumatic electric wheels, i = 1, 2, 3, 4, respectively representing the left front wheel, the right front wheel, the left rear wheel and the right rear wheel of the vehicle; v i is the longitudinal speed at the center of the i-th wheel; ω i is the angular speed of the i-th wheel; R is the rolling radius of the wheel;

[0082] The lateral force of the non-pneumatic electric wheel is fitted by a neural network model, and the nonlinear relationship between the tire lateral force, the side slip angle and the vertical load is obtained as follows:

[0083]

[0084] where Net_F y is the lateral force of the non-pneumatic electric wheel, W y1 is the weight matrix between the network input layer and the hidden layer, b y1 is the bias vector between the network input layer and the hidden layer, W y2 is the weight matrix between the network hidden layer and the output layer, b y2 is the bias vector between the network hidden layer and the output layer; α is the side slip angle of the non-pneumatic electric wheel, and is expressed as:

[0085]

[0086] where δ i is the steering angle of the i-th non-pneumatic electric wheel, β is the side slip angle of the vehicle mass center, l f is the distance from the vehicle mass center to the front axle, ω is the yaw angular velocity of the vehicle, and v x is the longitudinal speed of the vehicle.

[0087] Step 2, based on the mathematical model of the non-pneumatic electric wheel established in step 1, a nonlinear two-degree-of-freedom dynamic model oriented to vehicle control is constructed.

[0088] The nonlinear two-degree-of-freedom dynamic model oriented to vehicle control is expressed as follows:

[0089]

[0090] where m is the vehicle body mass, a y is the lateral acceleration of the vehicle; F yi , where i = 1, 2, 3, 4, are the lateral forces of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel of the vehicle, respectively, F xi , where i = 1, 2, 3, 4, are the longitudinal forces of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel of the vehicle, respectively; I z is the yaw motion moment of inertia of the vehicle, ω is the yaw angular velocity of the vehicle, l f and l r are the distances from the front axle and the rear axle to the mass center, and R is the rolling radius of the wheel.

[0091] The lateral acceleration of the vehicle is expressed as:

[0092]

[0093] where, is the lateral velocity rate of the vehicle, ω is the yaw rate of the vehicle, v x is the longitudinal velocity of the vehicle.

[0094] Step 3, based on multi-agent graph theory, four non-inflatable electric wheels of the vehicle are regarded as four agents, and a corresponding vehicle motion control model is designed considering the state coupling between the agents;

[0095] Let G = (V, E) represent a group system network composed of n agents that can be regarded as nodes, where V = {v1, v2,... v n} represents the set of individual nodes of the agents in the group, E = {e1, e2,... e n} represents the edge set of mutual communication between the agents, and each edge can be represented in the form of node e ij = (v i , v j ), v i is the starting point, and v j is the end point.

[0096] After the multi-agent system graph G is established, the next step is to modify the structure of the system through matrix and further analysis. Adjacency matrix A, degree matrix D, Laplacian matrix L, and incidence matrix B are introduced.

[0097] The four wheels are regarded as four multi-agents, i.e. four vertices, to form a graph G4. The adjacency matrix A is an n-order square matrix, and if there is an edge between vertex i and vertex j, then the element a ij = 1, and if there is no edge connecting vertex i and vertex j, then the element a ij = 0; the degree matrix D is an n-order diagonal matrix, and the diagonal elements d ii represent the degree of vertex i, and the non-diagonal elements are all 0; the Laplacian matrix L is obtained by subtracting the adjacency matrix A from the degree matrix D, denoted as L = D-A; the incidence matrix B describes the relationship between the vertices and edges in the graph, and if vertex i is associated with edge j, then b ij = 1, otherwise b ij = 0.

[0098] The relevant matrices in the established graph are:

[0099]

[0100] The side slip angle of the wheel is represented as:

[0101]

[0102] where Δε i Other wheel agents couple the state of the wheel agent i,

[0103]

[0104] Let β i , ω i be the i-th wheel side force F yi When the individual action, the vehicle mass center generated by the mass center side angle and yaw rate, through the vector transformation, meet β = β1 + β2 + β3 + β4, ω = ω1 + ω2 + ω3 + ω4, the value is:

[0105]

[0106] Thus the nonlinear two degree of freedom dynamic model is represented as:

[0107]

[0108] Where, k i is the i-th non-pneumatic electric wheel in a y <0.4g condition of side stiffness, l1 = l2 = l f , l3 = l4 = l r , e1 = e2 = 1, e3 = e4 = -1, T i is the driving torque of the i-th non-pneumatic electric wheel; δ i is the steering angle of the i-th non-pneumatic electric wheel;

[0109] Thus the vehicle control model based on multi-agent graph theory for non-pneumatic electric wheel matching vehicle control is obtained as:

[0110]

[0111] The state space equation of the model is:

[0112]

[0113] Where u * = Δε i ,

[0114] Step 4, based on the vehicle motion control model designed in step 3, the model predictive control algorithm is used for optimization solution, and the additional front wheel angle and additional yaw moment for vehicle control are obtained;

[0115] Discretize the state space equation of step 3:

[0116]

[0117] Where,

[0118] Let Nc For control horizon, N p For prediction horizon, the state variable and control variable recursively to future N time are:

[0119] {x i,k (t)=x i (t+k|t,k=1,2,...N}

[0120] {u i,k (t)=u i (t+k∣t),k=0,...,N-1}

[0121] The predicted output is:

[0122] Y i (k+1|k)=S x x i (k)+S u u i (k)

[0123] Wherein,

[0124] The intention of the whole vehicle control is to maintain the maneuverability and stability, and the control target function is designed as follows:

[0125]

[0126] Wherein, Q1 is the weighted matrix of the error between the predicted output and the reference output; R1 is the weighted matrix of the control amount; Is the relaxation factor; ρ is the relaxation factor penalty coefficient, which is used to constrain the relaxation degree and avoid the infinite increase of the relaxation factor;

[0127] Considering the constraints of the actuator and the requirements for the smoothness of the control amount, the control amount, the control increment and the output are limited, and at the same time, the relaxation factor is added to soften the constraints to ensure that the optimization problem has a feasible solution:

[0128]

[0129] The quadratic programming algorithm is used to calculate the control sequence of each control period, wherein the first element is the output of the feedback control, and the final required additional corner and additional yaw moment are obtained.

[0130] The above only describes the preferred embodiments of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should be considered as the protection scope of the present application.

Claims

1. A multi-agent graph theory-based non-hydrodynamic electric wheel matching vehicle control method, characterized in that: The method comprises the following steps: Step 1, a mathematical model of the non-pneumatic electric wheel is established according to the longitudinal, lateral and vertical mechanical properties of the non-pneumatic electric wheel; Step 2, a nonlinear two-degree-of-freedom dynamic model facing the vehicle control is constructed based on the mathematical model of the non-pneumatic electric wheel established in step 1; Step 3, based on multi-agent graph theory, the four non-pneumatic electric wheels of the vehicle are regarded as four agents, and a corresponding vehicle motion control model is designed by considering the state coupling between the agents; Step 4, based on the vehicle motion control model designed in step 3, a model predictive control algorithm is used for optimization and solution to obtain additional front wheel turning angles and additional yaw moments for vehicle control.

2. The multi-agent graph theory based non-hydrodynamic electric wheel matching vehicle control method according to claim 1, characterized in that: The non-pneumatic electric wheel comprises an outer wheel set, an inner wheel set and a driving motor; The outer wheel set comprises an elastic ring, a buckle, a pin ear and a rubber layer; The inner wheel set comprises a hinge connected with the pin ear and a hub; The driving motor is a hub motor embedded in the hub and connected with the inner wheel set.

3. The multi-agent graph theory based non-hydrodynamic electric wheel matching vehicle control method according to claim 1 or 2, characterized in that: In step 1, the mathematical model of the non-pneumatic electric wheel is represented as follows: A function relationship between the vertical force and the loading deflection of the non-pneumatic electric wheel is established: F z = k z (x z - λ) = k z x z - k z λ where F z is the non-pneumatic electric wheel vertical force, k z is the fitted vertical stiffness, x z is the wheel sinkage, and λ is the offset. The longitudinal force F of the non-pneumatic electric wheel is fitted by a BP neural network model x The nonlinear relationship between the tire longitudinal force and the slip ratio and the vertical load is established as where Net_F x (*) is the longitudinal force BP neural network model of the non-pneumatic electric wheel, W x1 is the weight matrix between the network input layer and the hidden layer, b x1 is the bias vector between the network input layer and the hidden layer, W x2 is the weight matrix between the network hidden layer and the output layer, b x2 is the bias vector between the network hidden layer and the output layer; s is the slip ratio of the non-pneumatic electric wheel, which is expressed as: where i represents the serial number of the four non-inflatable electric wheels, i = 1, 2, 3, 4, representing the left front wheel, the right front wheel, the left rear wheel and the right rear wheel of the vehicle respectively; v i is the longitudinal speed at the center of the i-th wheel; ω i is the angular speed of the i-th wheel; R is the rolling radius of the wheel; The lateral force of the non-pneumatic electric wheel is fitted through a neural network model, and the nonlinear relationship between the tire lateral force, the side slip angle and the vertical load is obtained as follows: where Net_F y (*) is the BP neural network model of the lateral force of the non-pneumatic electric wheel, W y1 is the weight matrix between the input layer and the hidden layer of the network, b y1 is the bias vector between the input layer and the hidden layer of the network, W y2 is the weight matrix between the hidden layer and the output layer of the network, b y2 is the bias vector between the hidden layer and the output layer of the network; and α is the side slip angle of the non-pneumatic electric wheel, and is expressed as: where δ i is the steering angle of the ith non-pneumatic electric wheel, β is the vehicle's center of mass side slip angle, l f is the distance from the vehicle's center of mass to the front axle, ω is the vehicle's yaw angular velocity, v x is the vehicle's longitudinal velocity.

4. The multi-agent graph theory based non-pneumatic motorized wheel matching whole vehicle control method according to claim 3, characterized in that: In step 2, the nonlinear two-degree-of-freedom dynamic model facing the vehicle control is represented as follows: where m is the vehicle mass, a y is the vehicle lateral acceleration; F yi where i = 1, 2, 3, 4 are the lateral forces of the vehicle front left wheel, front right wheel, rear left wheel and rear right wheel, respectively, F xi where i = 1, 2, 3, 4 are the longitudinal forces of the vehicle front left wheel, front right wheel, rear left wheel and rear right wheel, respectively; I z is the vehicle yaw motion moment of inertia, ω is the vehicle yaw angular velocity, l f and l r are the front / rear axle to center of mass distances, R is the rolling radius of the wheels; The lateral acceleration of the vehicle is represented as follows: wherein, is the lateral velocity of the vehicle, ω is the yaw rate of the vehicle, v x is the longitudinal velocity of the vehicle.

5. The multi-agent graph theory based non-pneumatic motorized wheel matching whole vehicle control method according to claim 4, characterized in that: In step 3, the four non-pneumatic electric wheels are regarded as four multi-agents, i.e. four vertices, to form a graph G4; an adjacency matrix A, a degree matrix D, a Laplacian matrix L and a correlation matrix B are introduced; The related matrices in the established graph are as follows: The side slip angle of the wheel is represented as follows: where Δε i Other wheel agents couple to the state of the ego wheel agent i, Let β i ω i The lateral force F of the i-th wheel is respectively yi When acting alone, the sideslip angle and yaw rate generated at the vehicle's center of gravity, through vector transformation, satisfy β=β1+β2+β3+β4, ω=ω1+ω2+ω3+ω4, with values ​​as follows: The nonlinear two-degree-of-freedom dynamic model is represented as follows: wherein k i is the lateral stiffness of the ith non-pneumatic electric wheel under a y <0.4g condition, li = l2 = l f , l3 = l4 = l r , ei = e2 = 1, e3 = e4 = -1, T i is the driving torque of the ith non-pneumatic electric wheel; δ i is the steering angle of the ith non-pneumatic electric wheel; The vehicle control model based on the multi-agent graph theory for matching the vehicle control of the non-pneumatic electric wheel is obtained as follows: The state space equation of the model is as follows: wherein u * = Δε i , 6. The multi-agent graph theory based non-pneumatic motorized wheel matching whole vehicle control method according to claim 5, characterized in that: In step 4, the state space equation of step 3 is discretized and represented as follows: wherein, Let N c be the control horizon, N p be the prediction horizon, the state and control variables recursively to the future N time instants are: {x i,k (t) = x i (t + k | t, k = 1, 2,... N} {u i,k (t) = u i (t+k | t), k = 0,..., N - 1} The prediction output is as follows: Y i (k+1|k) = S x x i (k) + S u u i (k) wherein The intention of the vehicle control is to maintain the maneuverability and stability, and the control objective function is designed as follows: wherein Q1 is a weighting matrix of prediction output and reference output error; R1 is a weighting matrix of control amount; is a relaxation factor; p is a relaxation factor penalty coefficient, used to constrain the relaxation degree and avoid the relaxation factor from increasing infinitely; Considering the constraints of the actuator and the requirements for the smoothness of the control quantity, the control increment and the output quantity are limited, and a relaxation factor is added to soften the constraints to ensure that the optimization problem has a feasible solution: A quadratic programming algorithm is used to calculate the control sequence of each control period, and the first element is the output of the feedback control to obtain the required additional turning angle and additional yaw moment.

Images