Full-state control method for dynamic drifting of distributed driving vehicle

Abstract

The invention is suitable for the technical field of automatic driving vehicle control, and provides a full-state control method for dynamic drifting of a distributed driving vehicle, which comprises the following steps of: tracking a reference path to obtain an expected state quantity; establishing a dynamic error of a full state to obtain expected state derivatives of the velocity, the side slip angle and the yaw velocity; calculating a rear tire force weight, and implementing rear tire force constraint of the distributed driving vehicle; inverting the vehicle dynamics model to obtain expected longitudinal force and lateral force of each wheel; performing tire model inversion to obtain an expected front wheel slip angle and wheel speed of each wheel; and according to the expected front wheel slip angle and the wheel speed of each wheel, an actuator is adjusted, and control instructions of the steering wheel angle and the torque of each wheel are generated. The state of the vehicle can be accurately controlled, the maneuverability of the vehicle is good during conventional turning and limit drifting, and the vehicle can flexibly cope with complex working conditions; the stability and safety of the vehicle are enhanced, and the full-state dynamic drift control of the distributed driving vehicle is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving vehicle control, and in particular relates to a full-state control method for dynamic drift of a distributed drive vehicle. Background Art

[0002] Since its development, autonomous driving technology has made great progress. The key to achieving the goal of improving traffic safety is that the autonomous driving system can use the maximum tire force to control the vehicle to the handling limit in an emergency. Professional drivers are a benchmark that is difficult to surpass in terms of fully utilizing the friction limit of road tires. They are good at pushing the vehicle to the dynamic limit, especially in some challenging special road conditions, such as sharp turns (hairpins) and loose roads with poor road adhesion conditions, and shorten the cornering time through drifting maneuvers, which often exceed the stability limit. Therefore, it is of great significance to give autonomous driving vehicles drift capabilities similar to those of professional drivers, which not only enables autonomous driving vehicles to reach the dynamic limit in regular turning driving, but also allows them to drift outside the stability limit, thereby fully utilizing the vehicle's maneuverability and improving the stability and safety of autonomous driving vehicles.

[0003] Rear-wheel drive is the basis for autonomous driving vehicles to achieve continuous and stable drift. Distributed drive vehicles add front-wheel drive force and yaw moment control on the basis of rear-wheel drive, so they have a stronger ability to stabilize the vehicle near the drift equilibrium point. However, the addition of front-wheel longitudinal force and yaw moment makes it present the characteristics of an over-drive system, which increases the difficulty of controller design. In addition, existing research has not yet achieved integrated control of conventional turning and drifting of vehicles under a unified controller, that is, dynamic drift has not been achieved. In view of this, the present invention proposes a full-state control method for dynamic drift of distributed drive vehicles. Dynamic drift refers to the integration of conventional turning maneuvers under normal driving of the vehicle and drift maneuvers under extreme driving. Full-state control refers to the independent control of all vehicle states (including speed, sideslip angle of the center of mass, and yaw angular velocity) during dynamic drift maneuvers. Summary of the invention

[0004] The purpose of the present invention is to provide a full-state control method for dynamic drift of a distributed drive vehicle, aiming to solve the problems raised in the above-mentioned background technology.

[0005] The purpose of the present invention is achieved through the following technical solutions: A full-state control method for dynamic drift of a distributed drive vehicle comprises the following steps: Step S1: tracking a reference path to obtain desired state quantities, including speed, center of mass sideslip angle, and yaw rate; Step S2: Establish the dynamic error of the whole state, so as to obtain the desired state derivatives of speed, center of mass sideslip angle and yaw rate; Step S3: Calculate the rear tire force weight and implement the rear tire force constraint of the distributed drive vehicle; Step S4: Inverse the vehicle dynamics model to obtain the expected longitudinal force and lateral force of each wheel; Step S5: Implement tire model inversion, including front tire model inversion and each wheel thrust angle model inversion, to obtain the desired front wheel slip angle and each wheel speed; Step S6: adjusting the actuator according to the desired front wheel slip angle and the wheel speed of each wheel to generate control instructions for the steering wheel angle and the torque of each wheel.

[0006] Furthermore, in step S1, the position and state of the vehicle are associated with the reference path information using a vehicle kinematics model, and the vehicle kinematics model is represented by the following formula: ; ; In the formula, The vehicle is in the global coordinate system The rate of change of position in direction; The vehicle is in the global coordinate system The rate of change of position in direction; is the rate of change of the vehicle yaw angle; is the speed of the vehicle; is the vehicle's center of mass sideslip angle; The vehicle's yaw rate; The vehicle kinematics model is used as the prediction model. As a state quantity, As the control quantity, The position of the vehicle in the global coordinate system, is the yaw angle of the vehicle; the interaction between the vehicle and the path is realized through the model predictive control algorithm. The optimal control problem composed of the cost function and constraints of the model predictive control is as follows: ; In the formula, It is the optimization variable in the model predictive control algorithm; is the cost function of the model predictive control algorithm; is the state quantity of the prediction model; for The state quantity of the moment prediction model; for The state quantity of the moment prediction model; for The control amount of the moment prediction model; is the output of the prediction model; They are the state matrix, input matrix and unit output matrix respectively; represents each time step; for the current moment; and They are prediction time domain and control time domain respectively; Recover envelope constraints for linear maximum stability; Finally, after the optimal solution, the desired vehicle state is generated, including the desired speed , expected sideslip angle and the desired yaw rate , the calculation formula is as follows: ; In the formula, is the deviation between the actual control quantity and the reference control quantity obtained by the optimal solution at the current moment; are the reference values ​​of speed, sideslip angle of center of mass and yaw rate respectively.

[0007] Furthermore, in step S2, the vehicle state is stabilized by establishing an error relationship between the desired state and the actual state of the vehicle, and the desired state derivatives of the vehicle's full state, i.e., the sideslip angle, yaw rate, and velocity, are calculated. , the calculation formula is as follows: ; In the formula, is the desired sideslip angle The differential of is the mass center slip angle The differential of is the desired yaw rate The differential of and is an adjustable control parameter.

[0008] Furthermore, in step S3, the total expected acceleration of the rear wheels is obtained by combining the expected state and state derivative of the vehicle, without considering load transfer, and the weight of the total expected acceleration of the rear wheels to the acceleration limit of the rear wheels, that is, the expected rear tire force weight, is calculated, and the calculation formula is as follows: ; In the formula, Between 0 and 1, acting as the rear tire force constraint applied in the inverse dynamics model; is the desired longitudinal acceleration; is the expected lateral acceleration; is the road adhesion coefficient; is the acceleration due to gravity; is the distance from the front axle to the center of mass; is the vehicle wheelbase; is the desired state derivative of velocity.

[0009] Furthermore, in step S4, a cost function is constructed based on the sum of squares of errors between the desired state derivative and the state derivative of vehicle feedback, and an optimal control problem is formed under the constraints of the distributed drive vehicle dynamics model and the rear tire force weights, and then solved using a nonlinear programming method to obtain the desired longitudinal force and lateral force of each wheel of the distributed drive vehicle.

[0010] Furthermore, in the front tire model inversion of step S5, a neural network tire model is established, and the state data of the vehicle driving on high-adhesion road and low-adhesion road are first collected, including the vehicle's speed, center of mass sideslip angle, yaw angular velocity, steering wheel angle and wheel speed; then the sideslip angle, longitudinal slip and lateral force of the front tire are estimated, and the vertical force of the front tire is calculated by the formula calculate, is the vehicle mass, is the distance from the rear axle to the center of mass; the data of high-adhesion road and low-adhesion road are mixed to form a data set and input into the neural network for training. The neural network contains three hidden layers with 64 neurons; finally, the neural network tire model of the front wheel is inverted by the following formula to obtain the desired front wheel side slip angle : ; In the formula, Front axle lateral forces calculated for the neural network tire model; are the expected lateral forces of the front left wheel and the front right wheel respectively; is the neural network tire model; It is longitudinal slip; The tire's side slip angle limit; In the inversion of the thrust angle model of each wheel, based on the expected longitudinal force and lateral force of each wheel obtained in step S4, the expected wheel speed of each wheel is calculated by inversion using the wheel thrust angle model. , the specific formula is as follows: ; In the formula, are the expected wheel speeds of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; is the front wheel turning angle of the vehicle; is the wheel radius; is the wheelbase of the left and right wheels of the vehicle; are the expected longitudinal forces of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; are the expected lateral forces on the front left wheel, front right wheel, rear left wheel and rear right wheel respectively.

[0011] Furthermore, the adjusting actuator link in step S6 includes wheel speed control and steering system control; In the wheel speed control, based on the wheel speed dynamics model, a feedforward plus feedback closed-loop control method is used to generate the control command of each wheel torque to achieve the control of the desired wheel speed of each wheel in step S5. The formula is as follows: ; In the formula, is the desired wheel torque; is the adjustable control gain; is the actual wheel speed; is the expected wheel speed of each wheel; is the wheel moment of inertia; is the desired longitudinal force; where Indicates the front axle and rear axle. Indicates left wheel and right wheel; In the steering system control, the goal is to indirectly control the front wheel slip angle of the vehicle by controlling the steering wheel. The steering wheel angle is calculated based on the desired front wheel slip angle output by the neural network tire model inversion of the front wheel in step S5. The specific formula is as follows: ; In the formula, is the desired steering wheel angle; is the transmission ratio of the steering system; is an adjustable control gain; is the desired front wheel slip angle.

[0012] Compared with the prior art, the present invention has the following beneficial effects: Improve vehicle maneuverability: The present invention gives autonomous vehicles drifting capabilities similar to those of professional drivers, allowing the vehicle to not only reach the dynamic limit in conventional turning, but also drift outside the stability limit, giving full play to the vehicle's maneuverability. Through precise control of the vehicle state at each step, such as controlling wheel force and steering according to the desired state quantity and state derivative, the vehicle can be flexibly controlled under complex working conditions.

[0013] Enhance vehicle stability and safety: By establishing full-state dynamic errors, stabilizing the vehicle state, and implementing rear tire force constraints and precise wheel force control, the vehicle's speed, center of mass sideslip angle, and yaw rate are adjusted in real time during vehicle movement to ensure that the vehicle remains stable in various driving conditions and avoids loss of control, thereby improving the stability and safety of autonomous vehicles.

[0014] Optimizing control accuracy: The present invention adopts a model predictive control algorithm for reference path tracking, and combines it with vehicle dynamics model inversion and tire model inversion techniques to accurately calculate the expected quantities of each vehicle state and the expected force and wheel speed of each wheel. By adjusting the actuator to generate precise control instructions, precise control of the vehicle is achieved, which improves control accuracy compared to traditional control methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flow chart of the method.

[0016] Figure 2 This is a framework diagram of the method of the present invention.

[0017] Figure 3 For the test scene.

[0018] Figure 4 The vehicle status. DETAILED DESCRIPTION

[0019] In order to have a clearer understanding of the technical features, purposes and beneficial effects of the present invention, the technical solution of the present invention is now described in detail below, but it should not be construed as limiting the applicable scope of the present invention.

[0020] The present invention provides a full-state control method for dynamic drift of a distributed drive vehicle. The flow chart and framework diagram of the method are respectively as follows: Figure 1 and Figure 2 As shown, the specific steps include: Step S1: Track the reference path to obtain the desired state quantities, including speed, center of mass sideslip angle, and yaw rate.

[0021] In step S1, the position and state of the vehicle are associated with the reference path information using the vehicle kinematics model, and the vehicle kinematics model is represented by the following formula: ; ; In the formula, The vehicle is in the global coordinate system The rate of change of position in direction; The vehicle is in the global coordinate system The rate of change of position in direction; is the rate of change of the vehicle yaw angle; is the speed of the vehicle; is the vehicle's center of mass sideslip angle; is the yaw angular velocity of the vehicle.

[0022] The vehicle kinematics model is used as the prediction model. As a state quantity, As the control quantity, The position of the vehicle in the global coordinate system, is the yaw angle of the vehicle; the interaction between the vehicle and the path is realized through the model predictive control algorithm, and then the reference path is tracked. The optimal control problem composed of the cost function and constraints of the model predictive control is as follows: ; In the formula, It is the optimization variable in the model predictive control algorithm; is the cost function of the model predictive control algorithm; is the state quantity of the prediction model; for The state quantity of the moment prediction model; for The state quantity of the moment prediction model; for The control amount of the moment prediction model; is the output of the prediction model; They are the state matrix, input matrix and unit output matrix respectively; represents each time step; for the current moment; and They are prediction time domain and control time domain respectively; Recover envelope constraints for linear maximum stability.

[0023] Finally, after the optimal solution, the desired vehicle state is generated, including the desired speed , expected sideslip angle and the desired yaw rate , the calculation formula is as follows: ; In the formula, is the deviation between the actual control quantity and the reference control quantity obtained by the optimal solution at the current moment; are the reference values ​​of speed, sideslip angle of center of mass and yaw rate respectively.

[0024] Step S2: Establish the dynamic error of the whole state, so as to obtain the desired state derivatives of velocity, sideslip angle of center of mass and yaw rate.

[0025] In step S2, the vehicle state is stabilized by establishing the error relationship between the desired state and the actual state of the vehicle, and the desired state derivatives of the vehicle's full state, namely the sideslip angle, yaw rate and velocity, are calculated. , the calculation formula is as follows: ; In the formula, is the desired sideslip angle of the center of mass; is the vehicle's center of mass sideslip angle; is the desired sideslip angle The differential of is the mass center slip angle The differential of is the desired yaw rate; is the yaw rate of the vehicle; is the desired yaw rate The differential of is the expected speed; is the speed of the vehicle; and is an adjustable control parameter.

[0026] Step S3: Calculate the rear tire force weight and implement the rear tire force constraint of the distributed drive vehicle.

[0027] In step S3, the total expected acceleration of the rear wheels is obtained by combining the expected state and state derivative of the vehicle, without considering load transfer, and the weight of the total expected acceleration of the rear wheels to the acceleration limit of the rear wheels is calculated. This weight is the expected rear tire force weight, which is used to implement the rear tire force constraint of the distributed drive vehicle. The calculation formula is as follows: ; In the formula, Between 0 and 1, it is the rear tire force constraint applied in the inverse dynamics model; is the desired longitudinal acceleration; is the expected lateral acceleration; is the road adhesion coefficient; is the acceleration due to gravity; is the distance from the front axle to the center of mass; is the vehicle wheelbase; is the desired state derivative of velocity; is the desired sideslip angle of the center of mass; is the desired yaw rate; is the expected speed.

[0028] Step S4: Inverse the vehicle dynamics model to obtain the expected longitudinal force and lateral force of each wheel.

[0029] Since the distributed drive vehicle has the characteristics of four-wheel independent drive, it can generate additional front wheel drive force and yaw torque compared to traditional rear-wheel drive vehicles. This allows the vehicle to achieve state decoupling during drift maneuvers, that is, the vehicle's speed, center of mass sideslip angle and yaw rate can be independently controlled, thereby achieving full state control of the vehicle during conventional turning and drift maneuvering integrated control, that is, dynamic drift control.

[0030] In the inversion process of the distributed drive vehicle dynamics model, the complete state control of speed, center of mass sideslip angle and yaw rate is realized synchronously. The specific implementation method is: based on the square sum of the error between the desired state derivative and the state derivative of the vehicle feedback, the cost function is constructed, and under the constraints of the distributed drive vehicle dynamics model and the rear tire force weight, the optimal control problem is formed, and then the nonlinear programming method is used to solve it to obtain the desired longitudinal force and lateral force of each wheel of the distributed drive vehicle. The relevant formula is as follows: ; In the formula, is the expected longitudinal force of each wheel, is the desired lateral force of each wheel, where Indicates the front axle and rear axle. Indicates left wheel and right wheel; is the differential of velocity; is the desired state derivative of velocity; is the differential of the sideslip angle of the center of mass; is the yaw rate of the vehicle; is the desired state derivative of the sideslip angle of the center of mass; is the desired yaw rate; is the differential of the yaw rate; is the desired state derivative of the yaw rate; They are front wheel longitudinal force, front wheel lateral force, rear wheel longitudinal force and rear wheel lateral force respectively; They are the front left wheel longitudinal force, the front right wheel longitudinal force, the rear left wheel longitudinal force and the rear right wheel longitudinal force; They are the front left wheel lateral force, the front right wheel lateral force, the rear left wheel lateral force and the rear right wheel lateral force respectively; is the front wheel turning angle of the vehicle; is the vehicle's center of mass sideslip angle; is the vehicle mass; is the speed of the vehicle; is the yaw rate of the vehicle; are the distances from the front axle and rear axle to the center of mass respectively; is the moment of inertia of the vehicle; is the yaw moment; is the wheelbase of the left and right wheels of the vehicle; are the expected longitudinal forces of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; are the expected lateral forces of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; are the vertical forces of the front and rear tires respectively; The load transfer between the front and rear axles is shown in Figure 2. The calculation formula is as follows: ; ; In the formula, is the acceleration due to gravity; is the vehicle wheelbase; is the height of the vehicle's center of mass; are the lateral load transfer coefficients of the front and rear axles, respectively.

[0031] Step S5: Implement tire model inversion, including front tire model inversion and each wheel thrust angle model inversion, to obtain the desired front wheel slip angle and each wheel speed.

[0032] In the front tire model inversion, a neural network tire model is established. First, the state data of the vehicle driving on high-adhesion and low-adhesion roads are collected, including the vehicle's speed, center of mass sideslip angle, yaw angular velocity, steering wheel angle, and wheel speed; then the sideslip angle, longitudinal slip, and lateral force of the front tire are estimated. The vertical force of the front tire can be calculated by the formula calculate, is the vehicle mass, is the acceleration due to gravity, is the distance from the rear axle to the center of mass, is the vehicle wheelbase; the data of high-adhesion road and low-adhesion road are mixed to form a data set and input into the neural network for training. The neural network contains three hidden layers with 64 neurons; finally, the neural network tire model of the front wheel is inverted by the following formula to obtain the desired front wheel side slip angle : ; In the formula, Front axle lateral forces calculated for the neural network tire model; are the expected lateral forces of the front left wheel and the front right wheel respectively; is the neural network tire model; It is longitudinal slip; is the road adhesion coefficient; is the vertical force of the front tire; It is the side slip angle limit of the tire. Once this limit is exceeded, the lateral force of the tire will reach saturation.

[0033] In the inversion of the thrust angle model of each wheel, based on the expected longitudinal force and lateral force of each wheel obtained in step S4, the expected wheel speed of each wheel is calculated by inversion using the wheel thrust angle model. , the specific formula is as follows: ; In the formula, are the expected wheel speeds of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; is the front wheel turning angle of the vehicle; is the speed of the vehicle; is the vehicle's center of mass sideslip angle; is the wheel radius; is the yaw rate of the vehicle; and are the distances from the front axle and rear axle to the center of mass respectively; is the wheelbase of the left and right wheels of the vehicle; are the expected longitudinal forces of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; are the expected lateral forces on the front left wheel, front right wheel, rear left wheel and rear right wheel respectively.

[0034] Step S6: adjusting the actuator according to the desired front wheel slip angle and the wheel speed of each wheel to generate control instructions for the steering wheel angle and the torque of each wheel.

[0035] The adjustment actuator link in step S6 includes wheel speed control and steering system control.

[0036] Wheel speed control: Based on the wheel speed dynamics model, a closed-loop control method of feedforward plus feedback is used to generate control instructions for the torque of each wheel to achieve control of the desired wheel speed of each wheel in step S5. The formula is as follows: ; In the formula, is the desired wheel torque; is the adjustable control gain; is the actual wheel speed; is the expected wheel speed of each wheel; is the wheel moment of inertia; is the wheel radius; is the desired longitudinal force; where Indicates the front axle and rear axle. Indicates left wheel and right wheel.

[0037] Steering system control: Its goal is to indirectly control the front wheel slip angle of the vehicle by controlling the steering wheel. The steering wheel angle is calculated based on the desired front wheel slip angle output by the neural network tire model inversion of the front wheel in step S5. The specific formula is as follows: ; In the formula, is the desired steering wheel angle; is the transmission ratio of the steering system; is an adjustable control gain; is the speed of the vehicle; is the vehicle's center of mass sideslip angle; is the distance from the front axle to the center of mass; is the yaw rate of the vehicle; is the front wheel turning angle of the vehicle; is the desired front wheel slip angle.

[0038] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0039] Example 1: In order to verify the feasibility of the method of the present invention, a dynamic drift control test was carried out. The test scenario is as follows: Figure 3 The purpose is to make the vehicle track the reference trajectory in an integrated form of conventional turning maneuvers and drifting maneuvers. At the same time, in order to highlight the full-state control advantage of distributed drive vehicles in the dynamic drift process, the dynamic drift control performance of the distributed drive vehicle is compared with that of the rear-wheel drive vehicle in the same test scenario. The comparison results are shown in Figure 4 As shown: In terms of speed control performance, the rear-wheel drive vehicle had the largest speed error, reaching 1.45m / s when driving around 210m, while the speed control of the distributed drive vehicle was more stable, with a speed error of only 0.48m / s. This shows that the addition of front-wheel drive significantly improves the speed control stability of the distributed drive electric vehicle.

[0040] In terms of center of mass slip angle control, when the vehicle is in the steady-state drift stage (such as the 400m-450m range), the center of mass slip angle of the distributed drive vehicle is closest to the reference value. Further comparison of the root mean square error value shows that the root mean square error of the center of mass slip angle of the distributed drive vehicle is the smallest, at 10.6°, while the rear-wheel drive vehicle has the largest value. This shows that in the integrated control of drifting and conventional turning, distributed drive electric vehicles have obvious advantages over rear-wheel drive vehicles.

[0041] When the vehicle transitions between drifting and conventional turning maneuvers, the sideslip angle at the center of mass changes dramatically. At this time, the vehicle needs to meet the yaw rate required for the tracking path while also requiring an additional yaw rate to track the sideslip angle at the center of mass. From the yaw rate curve, the yaw rates of both the distributed drive vehicle and the rear-wheel drive vehicle can be effectively controlled during dynamic drift. Based on the above results, it can be seen that during dynamic drift, the distributed drive vehicle can better control the speed and the sideslip angle at the center of mass while ensuring stable control of the yaw rate; while the rear-wheel drive vehicle is significantly weaker in controlling the speed and the sideslip angle at the center of mass. The results fully reflect the characteristics of the distributed drive vehicle in full-state control of dynamic drift, and also strongly demonstrate the significant advantages of the method of the present invention.

[0042] The above are only preferred embodiments of the present invention. It should be pointed out that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention. These should also be regarded as the protection scope of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent.

Claims

1. A full-state control method for dynamic drift of a distributed drive vehicle, characterized in that: The following steps are involved: Step S1: tracking a reference path to obtain desired state quantities, including speed, center of mass sideslip angle, and yaw rate; Step S2: Establish the dynamic error of the whole state, so as to obtain the desired state derivatives of speed, center of mass sideslip angle and yaw rate; Step S3: Calculate the rear tire force weight and implement the rear tire force constraint of the distributed drive vehicle; Step S4: Inverse the vehicle dynamics model to obtain the expected longitudinal force and lateral force of each wheel; Step S5: Implement tire model inversion, including front tire model inversion and each wheel thrust angle model inversion, to obtain the desired front wheel slip angle and each wheel speed; Step S6: adjusting the actuator according to the desired front wheel slip angle and the wheel speed of each wheel to generate control instructions for the steering wheel angle and the torque of each wheel.

2. The full-state control method for dynamic drift of a distributed drive vehicle according to claim 1, characterized in that: In step S1, the position and state of the vehicle are associated with the reference path information using a vehicle kinematics model, and the vehicle kinematics model is represented by the following formula: ； ； In the formula, The vehicle is in the global coordinate system The rate of change of position in direction; The vehicle is in the global coordinate system The rate of change of position in direction; is the rate of change of the vehicle yaw angle; is the speed of the vehicle; is the vehicle's center of mass sideslip angle; The vehicle's yaw rate; The vehicle kinematic model is used as the prediction model. As a state quantity, As the control quantity, The position of the vehicle in the global coordinate system, is the yaw angle of the vehicle; the interaction between the vehicle and the path is realized through the model predictive control algorithm. The optimal control problem composed of the cost function and constraints of the model predictive control is as follows: ； In the formula, It is the optimization variable in the model predictive control algorithm; is the cost function of the model predictive control algorithm; is the state quantity of the prediction model; for The state quantity of the moment prediction model; for The state quantity of the moment prediction model; for The control amount of the moment prediction model; is the output of the prediction model; They are the state matrix, input matrix and unit output matrix respectively; represents each time step; for the current moment; and They are prediction time domain and control time domain respectively; Recover envelope constraints for linear maximum stability; Finally, after the optimal solution, the desired vehicle state is generated, including the desired speed , expected sideslip angle and the desired yaw rate , the calculation formula is as follows: ； In the formula, is the deviation between the actual control quantity and the reference control quantity obtained by the optimal solution at the current moment; are the reference values ​​of speed, sideslip angle of center of mass and yaw rate respectively.

3. The full-state control method for dynamic drift of a distributed drive vehicle according to claim 2, characterized in that: In step S2, the vehicle state is stabilized by establishing an error relationship between the desired state and the actual state of the vehicle, and the desired state derivatives of the vehicle's full state, i.e., the sideslip angle, yaw rate, and velocity, are calculated. , the calculation formula is as follows: ； In the formula, is the desired sideslip angle The differential of is the mass center slip angle The differential of is the desired yaw rate The differential of and is an adjustable control parameter.

4. The full-state control method for dynamic drift of a distributed drive vehicle according to claim 3, characterized in that: In step S3, the total expected acceleration of the rear wheels is obtained by combining the expected state and state derivative of the vehicle, without considering load transfer, and the weight of the total expected acceleration of the rear wheels to the acceleration limit of the rear wheels, that is, the expected rear tire force weight, is calculated, and the calculation formula is as follows: ； In the formula, Between 0 and 1, acting as the rear tire force constraint applied in the inverse dynamics model; is the desired longitudinal acceleration; is the expected lateral acceleration; is the road adhesion coefficient; is the acceleration due to gravity; is the distance from the front axle to the center of mass; is the vehicle wheelbase; is the desired state derivative of velocity.

5. The full-state control method for dynamic drift of a distributed drive vehicle according to claim 4, characterized in that: In step S4, a cost function is constructed based on the sum of squares of the errors between the desired state derivative and the state derivative of vehicle feedback, and an optimal control problem is formed under the constraints of the distributed drive vehicle dynamics model and the rear tire force weights, and then solved using a nonlinear programming method to obtain the desired longitudinal force and lateral force of each wheel of the distributed drive vehicle.

6. The full-state control method for dynamic drift of a distributed drive vehicle according to claim 5, characterized in that: In the front tire model inversion of step S5, a neural network tire model is established. First, the state data of the vehicle during driving on high-adhesion road surfaces and low-adhesion road surfaces are collected, including the vehicle's speed, center of mass sideslip angle, yaw angular velocity, steering wheel angle and wheel speed; then the sideslip angle, longitudinal slip and lateral force of the front tire are estimated, and the vertical force of the front tire is calculated by the formula calculate, is the vehicle mass, is the distance from the rear axle to the center of mass; The data of high-adhesion road and low-adhesion road are mixed to form a data set and input into the neural network for training. The neural network contains three hidden layers with 64 neurons. Finally, the neural network tire model of the front wheel is inverted by the following formula to obtain the desired front wheel side slip angle : ； In the formula, Front axle lateral forces calculated for the neural network tire model; are the expected lateral forces of the front left wheel and the front right wheel respectively; is the neural network tire model; It is longitudinal slip; The tire's side slip angle limit; In the inversion of the thrust angle model of each wheel, based on the expected longitudinal force and lateral force of each wheel obtained in step S4, the expected wheel speed of each wheel is calculated by inversion using the wheel thrust angle model. , the specific formula is as follows: ； In the formula, are the expected wheel speeds of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; is the front wheel turning angle of the vehicle; is the wheel radius; is the wheelbase of the left and right wheels of the vehicle; are the expected longitudinal forces of the front left wheel, front right wheel, rear left wheel and rear right wheel respectively; are the expected lateral forces on the front left wheel, front right wheel, rear left wheel and rear right wheel respectively.

7. The full-state control method for dynamic drift of a distributed drive vehicle according to claim 6, characterized in that: The adjusting actuator link in step S6 includes wheel speed control and steering system control; In the wheel speed control, based on the wheel speed dynamics model, a feedforward plus feedback closed-loop control method is used to generate the control command of each wheel torque to achieve the control of the desired wheel speed of each wheel in step S5. The formula is as follows: ； In the formula, is the desired wheel torque; is the adjustable control gain; is the actual wheel speed; is the expected wheel speed of each wheel; is the wheel moment of inertia; is the desired longitudinal force; where Indicates the front axle and rear axle. Indicates left wheel and right wheel; In the steering system control, the goal is to indirectly control the front wheel slip angle of the vehicle by controlling the steering wheel. The steering wheel angle is calculated based on the desired front wheel slip angle output by the neural network tire model inversion of the front wheel in step S5. The specific formula is as follows: ； In the formula, is the desired steering wheel angle; is the gear ratio of the steering system; is an adjustable control gain; is the desired front wheel slip angle.

Images