CN-121973757-A - Stability undisturbed switching control method for independently-driven electric automobile

CN121973757ACN 121973757 ACN121973757 ACN 121973757ACN-121973757-A

Abstract

An undisturbed switching control method for stability of an independently driven electric automobile belongs to the technical field of electric automobiles. The method comprises the steps of establishing a two-degree-of-freedom dynamics model of a vehicle, determining an ideal centroid slip angle and a yaw rate, designing a model prediction controller, constructing an Actor-Critic deep reinforcement learning framework based on a long-short time memory network, simultaneously designing a state space, an action space and a return function, designing an undisturbed switching control strategy based on return function evaluation, smoothly switching between model prediction control and deep reinforcement learning control, and finally distributing the obtained additional yaw moment to each driving wheel according to the tire attachment utilization rate. The advantages of model-based control and data-driven control are integrally combined, control quantity jump is avoided through undisturbed switching, and stability control precision and stability of the vehicle under complex working conditions are remarkably improved.

Inventors

HE YILIN
ZHAO XUAN
YU MAN
YUAN XIAOLEI
ZHOU CHENYU
WANG SHU
CHEN FU

Assignees

长安大学

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (10)

1. The undisturbed switching control method for the stability of the independently driven electric automobile is characterized by comprising the following steps of: step one, a two-degree-of-freedom vehicle dynamics model is established, wherein the two-degree-of-freedom vehicle dynamics model comprises a vehicle transverse dynamics equation and a vehicle yaw dynamics equation; Acquiring ideal state information of stability control according to the vehicle motion state and road attachment conditions, wherein the ideal state information comprises an ideal centroid slip angle and an ideal yaw rate; Designing a model prediction stability control method based on a physical model on the basis of the two-degree-of-freedom vehicle dynamics model in the first step, wherein the model prediction stability control method comprises a prediction model, an optimization target and constraint conditions, and the model prediction stability control method is used for inputting actual state information of a vehicle at the current moment and ideal state information obtained in the second step and outputting an additional yaw moment control quantity calculated based on model prediction control; Step four, constructing an Actor-Critic deep reinforcement learning framework, taking actual state information of the vehicle and the additional yaw moment control quantity obtained in the step three as inputs, and outputting the additional yaw moment control quantity based on deep reinforcement learning; The actual state information of the vehicle comprises a centroid slip angle and a yaw rate of the current moment and the past T time steps; The additional yaw moment control quantity obtained in the third step and the additional yaw moment control quantity based on deep reinforcement learning obtained in the fourth step are respectively used as inputs of a model prediction controller and a reinforcement learning controller, and the outputs of the model prediction controller and the reinforcement learning controller are final additional yaw moment control quantity after undisturbed switching; designing a switching control method, wherein the switching control method is used for switching the model predictive controller and the reinforcement learning controller; And step six, distributing the final additional yaw moment control quantity obtained in the step five after undisturbed switching to each driving wheel according to the tire adhesion utilization rate.
2. The method for undisturbed switching control of stability of an independently driven electric vehicle as set forth in claim 1, wherein in step one, a vehicle transverse dynamics equation is as follows: The vehicle yaw dynamics equation is as follows: Wherein, the 、 Representing the cornering stiffness of the front and rear axles; The mass center side deflection angle is represented by u, the longitudinal speed is represented by u, and the distance from the mass center to the front shaft and the rear shaft is represented by a and b; the yaw rate is indicated as such, Representing the yaw rate change rate; indicating the front wheel rotation angle; representing the mass of the vehicle; indicating the lateral velocity of the vehicle, Representing the rate of change of lateral velocity; indicating the winding of the car The moment of inertia of the shaft.
3. The method for undisturbed switching control of stability of an independently driven electric vehicle as set forth in claim 1, wherein in step two, the ideal centroid slip angle is as follows And ideal yaw rate The calculation formula of (2) is as follows: Wherein, the Representing the wheelbase of the automobile; a stability factor is indicated and is indicated, Mu represents road adhesion coefficients under different types of roads; Representing the mass of the car, g representing the gravitational acceleration, sgn representing a sign function.
4. The method for undisturbed switching control of stability of an independently driven electric vehicle as set forth in claim 1, wherein the model predictive stability control method in step three specifically includes: Based on the two-degree-of-freedom vehicle dynamics model in the first step, a nonlinear state space prediction model of the control system is established, and a nonlinear state space equation is as follows: Wherein, the , , Representing the state change amount; The constraint condition is that Wherein And Respectively a minimum allowable value and a maximum allowable value of the additional yaw moment; The optimization target is Is in actual state Tracking ideal conditions At the cost of Q is a weight matrix of the tracking target, Representing the control input, W represents the weight matrix of the control input.
5. The method for controlling stability undisturbed switching of an independently driven electric vehicle as described in claim 1, wherein in step four, the Actor-Critic deep reinforcement learning framework comprises an Actor network and a Critic network; The Actor network comprises a first input layer, a first long-short-time memory network, a first full-connection layer, a second full-connection layer and a first output layer which are sequentially connected in series; The Critic network comprises a second input layer, a third long-time memory network, a third full-connection layer, a fourth full-connection layer and a second output layer which are sequentially connected in series, and the first output layer is also connected in series with the third full-connection layer; the actual state information of the vehicle is used as the input of a first input layer and a second input layer, and the additional yaw moment control quantity is used as the input of a third full connection layer.
6. The method for controlling stability undisturbed switching of an independently driven electric vehicle as described in claim 5, wherein said long-short-term memory network comprises an LSTM input layer, an LSTM hidden layer, an LSTM full connection layer and an LSTM output layer which are connected in sequence; the LSTM input layer is used for receiving actual state information of the vehicle; the LSTM hidden layer is used for extracting the time sequence characteristics of the actual state information of the vehicle; the LSTM full-connection layer adopts a ReLU as an activation function to realize feature mapping; the LSTM output layer is used to output the additional yaw moment control quantity or the corresponding cost function.
7. The method for undisturbed switching control of stability of an independently driven electric vehicle as set forth in claim 1, wherein the switching control method in step five specifically comprises: a control Flag bit is defined, and when flag=1, a reinforcement learning controller is selected, and when flag=0, a model predictive controller is selected: Wherein, the To enhance the instant return under learning stability control, The instantaneous return under stability control is predicted for the model.
8. The method for undisturbed switching control of stability of an independently driven electric vehicle as set forth in claim 1, wherein the sixth step specifically comprises: Step 601, calculating the single tire attachment utilization ratio according to the ratio of the utilization adhesion force of the single wheel to the maximum adhesion force provided by the ground: Wherein i=1, 2,3,4 represent the left front wheel, the right front wheel, the left rear wheel, the right rear wheel, respectively; Represents the ith tire longitudinal force; represents the ith tire lateral force; represents the ith tire vertical force; Represents the road adhesion coefficient; step 602, taking the sum of squares of four tire attachment utilization ratios as an optimization objective function: Wherein i=1, 2,3,4 represent the left front wheel, the right front wheel, the left rear wheel, the right rear wheel, respectively; Represents the ith tire longitudinal force; represents the ith tire lateral force; represents the ith tire vertical force; Represents the road adhesion coefficient; step 603, satisfying the following equation constraint: Wherein F x1 、F x2 、F x3 and F x4 represent final additional yaw moment control amounts of left front wheel, right front wheel, left rear wheel and right rear wheel of the automobile, F d represents total driving force of the automobile, d represents wheel track; representing the final additional yaw moment control amount after the undisturbed switching.
9. A computer readable storage medium, wherein the computer readable storage medium stores a computer program which, when executed by a processor, implements the method for controlling stability undisturbed switching of an independently driven electric vehicle according to any one of claims 1 to 8.
10. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the independently driven electric vehicle stability undisturbed switching control method of any of claims 1 to 8.

Description

Stability undisturbed switching control method for independently-driven electric automobile Technical Field The invention belongs to the technical field of electric automobiles, and particularly relates to a stability undisturbed switching control method for an independently driven electric automobile. Background The wheels of the electric automobile can be independently controlled by independent driving, and the response is rapid, so that the electric automobile is an important direction of development of new energy automobiles. The vehicle stability control aims at guaranteeing the capability of the automobile to resist external interference and stably run. The existing stability control strategy mainly comprises direct yaw moment control, active front wheel steering and the like, and the control method is mostly based on a vehicle model or a preset rule. However, model-based approaches rely heavily on the accuracy of the model. The simplified model can reduce the control precision, and the complex high-precision model is difficult to model and large in calculated amount, so that the real-time control requirement is difficult to meet. Meanwhile, the vehicle parameters are time-varying and complex in working conditions, and the method based on the fixed model or the parameters is insufficient in robustness. In recent years, data-driven control methods such as deep learning do not rely on accurate physical models, control is performed by self-learning ability, but there are challenges in interpretability and generalized security. Disclosure of Invention Aiming at the problems, the invention aims to provide an undisturbed switching control method for stability of an independently driven electric automobile, which solves the problem of how to organically combine model-based control with data-based control, makes up for the shortages, and designs a stability control method which is suitable for complex working conditions, high in precision and stable. In order to achieve the above purpose, the technical scheme adopted by the invention comprises the following steps: a stability undisturbed switching control method for an independently driven electric automobile comprises the following steps: step one, a two-degree-of-freedom vehicle dynamics model is established, wherein the two-degree-of-freedom vehicle dynamics model comprises a vehicle transverse dynamics equation and a vehicle yaw dynamics equation; acquiring ideal state information of stability control according to the vehicle motion state and road attachment conditions, wherein the ideal state information comprises an ideal centroid slip angle and an ideal yaw rate; Step three, a model prediction stability control method based on a physical model is designed on the basis of the two-degree-of-freedom vehicle dynamics model in the step one, the model prediction stability control method comprises a prediction model, an optimization target and constraint conditions, the actual state information of the vehicle at the current moment and the ideal state information obtained in the step two are input, and the additional yaw moment control quantity calculated based on the model prediction control is output; Building an Actor-Critic deep reinforcement learning framework, taking actual state information of the vehicle and the additional yaw moment control quantity obtained in the third step as inputs, and outputting the additional yaw moment control quantity based on deep reinforcement learning; The third step of taking the additional yaw moment control quantity obtained in the third step and the additional yaw moment control quantity based on deep reinforcement learning obtained in the fourth step as the input of a model prediction controller and a reinforcement learning controller respectively, wherein the output of the model prediction controller and the reinforcement learning controller are the final additional yaw moment control quantity after undisturbed switching; And step six, distributing the final additional yaw moment control quantity obtained in the step five after undisturbed switching to each driving wheel according to the tire adhesion utilization rate. Preferably, the vehicle transverse dynamics equation in step one is as follows: the vehicle yaw dynamics equation is as follows: Wherein, the 、Representing the cornering stiffness of the front and rear axles; The mass center side deflection angle is represented by u, the longitudinal speed is represented by u, and the distance from the mass center to the front shaft and the rear shaft is represented by a and b; the yaw rate is indicated as such, Representing the yaw rate change rate; indicating the front wheel rotation angle; representing the mass of the vehicle; indicating the lateral velocity of the vehicle, Representing the rate of change of lateral velocity; indicating the winding of the car The moment of inertia of the shaft. Preferably, the ideal centroid slip angle in step twoAnd ideal yaw r