Vehicle stability control method based on DDPG self-adaptive super-spiral sliding mode

Abstract

A vehicle stability control method based on a DDPG self-adaptive super-spiral sliding mode belongs to the field of transverse stability control of electric vehicles, and comprises the following steps: step 1, establishing a linear 2-DOF reference model, and obtaining an ideal yaw velocity and a side slip angle required by stable control; 2, designing a sliding mode controller; 3, designing a super-spiral sliding mode controller on the basis of the step 2; 4, designing a DDPG self-adaptive super-spiral sliding mode controller on the basis of the step 3; and 5, distributing the calculated additional yawing moment to each wheel to realize vehicle stability control. According to the method, the buffeting problem of traditional sliding mode control is solved by utilizing a second-order sliding mode structure of the STSMC, and meanwhile, through a DDPG dynamic optimization control strategy, parameters are adjusted in a self-adaptive manner, so that the accurate yawing moment is rapidly generated under complex road conditions and sudden working conditions.

Description

Technical Field

[0001] The present invention belongs to the field of lateral stability control of electric vehicles, and in particular relates to a vehicle stability control method based on a DDPG adaptive super-helical sliding mode. Background Art

[0002] Distributed drive electric vehicles (DDEVs) have attracted considerable attention due to their ability to independently control wheel drive torque. As an overdrive system, DDEVs have redundant actuators, making motor torque distribution a core challenge in optimizing vehicle control performance. Improper DDEV torque distribution can exacerbate the risk of lateral motion, such as sideslip and tailspin. Direct yaw moment control is a vehicle stability control technology that generates yaw torque by adjusting the drive torque difference between each wheel, thereby controlling the vehicle's lateral motion and improving its handling and stability. Traditional yaw moment control technology primarily relies on fixed-structure control and adjustment methods, making it difficult to flexibly adapt to dynamic changes in vehicle operating conditions. It cannot dynamically adjust to the vehicle's actual driving state and external environment in real time, cannot guarantee system stability and robustness, and can easily cause the vehicle to become lateral instability. Summary of the Invention

[0003] The present invention provides a vehicle stability control method based on DDPG adaptive super-helical sliding mode to solve the defects in the prior art.

[0004] The present invention is achieved through the following technical solutions: The vehicle stability control method based on DDPG adaptive super-helical sliding mode includes the following steps: Step 1: Establish a linear 2-DOF reference model to obtain the ideal yaw rate and sideslip angle required for stable control; Step 2: Design a sliding mode controller; Step 3: Design a super-helical sliding mode controller based on step 2; Step 4: Design the DDPG adaptive super-helical sliding mode controller based on step 3; Step 5: Distribute the calculated additional yaw moment to each wheel to achieve vehicle stability control.

[0005] In the vehicle stability control method based on the DDPG adaptive super-helical sliding mode, the ideal yaw rate under steady-state steering conditions can be obtained by establishing the total lateral force balance equation and the yaw moment balance equation in step 1, combined with the linear assumption of the tire cornering force and the limitation of the road adhesion coefficient. and the center of mass slip angle , the specific expression formula is as follows: (1) Where, is the front axle wheelbase, is the rear axle wheelbase, is the front wheel deflection angle, is the road adhesion coefficient, is the vehicle longitudinal velocity, is the safety factor, 、 is the cornering stiffness of the front and rear tires, To describe the vehicle's steering stability coefficient , .

[0006] In the vehicle stability control method based on DDPG adaptive super-helical sliding mode, the additional yaw moment is added to the linear 2-DOF vehicle dynamics model in step 2. , the specific expression formula is as follows: (2) Where, is the vehicle mass, is the moment of inertia of the vehicle around the z axis, is the additional yaw moment, 、 are the yaw rate and its derivative, 、 are the sideslip angle of the center of mass and its derivative respectively; In step 2, define the yaw rate sliding film surface , Center of mass slip angle sliding surface The expression formulas are: (3) (4) Where, is the sliding film surface parameter of yaw rate, is the sliding surface parameter of the center of mass sideslip angle, all of which are greater than 0; 、 are the yaw rate tracking error and its derivative, 、 is the center of mass sideslip angle error and its derivative; In step 2, the yaw rate approaching rate is defined as , Center of mass sideslip angle approach rate The expression formula is: (5) (6) Where, is the sliding mode approach coefficient of yaw rate, is the sliding mode approach coefficient of the sideslip angle at the center of mass, is a symbolic function; Derivative of formula (3) and formula (5) are substituted into formula (2) to obtain the yaw rate and yaw torque based on the sliding mode controller: : (7) The additional yaw moment of the sideslip angle of the center of mass based on the sliding mode controller is obtained by derivation of formula (4) and substitution of formula (6) into formula (2): : (8) The distribution coefficient method is used to combine formula (7) and formula (8) to output the additional yaw moment. The total yaw moment is: : (9) Where, is the additional gain for the yaw rate, Adds a bonus to the slip angle.

[0007] In the vehicle stability control method based on the DDPG adaptive super-helical sliding mode described above, the expression formula for defining the super-helical sliding mode approach rate in step 3 is: (10) Where, is the control input, 、 is the gain of the super-helical sliding mode controller; Substitute the derivative of formula (3) and formula (10) into formula (2) to obtain the additional yaw moment of the yaw angular velocity based on the super-helical sliding mode controller: The expression formula is: (11) Where, 、 is the super-helical sliding mode yaw rate approach coefficient; Derivative of formula (4) and substitute formula (10) into formula (2) The center of mass slip angle moment and additional yaw moment based on super-helical sliding mode controller The expression formula is: (12) Where, 、 is the sideslip angle approach coefficient of the super-helical sliding mode center of mass.

[0008] In the vehicle stability control method based on the DDPG adaptive super-spiral sliding mode described above, in step 4, the DDPG algorithm is embedded in the STSMC framework to construct a state space-action space-reward interaction mechanism.

[0009] As described above, the vehicle stability control method based on DDPG adaptive super-helical sliding mode, the state space is defined as the real-time vehicle dynamics state quantity , its expression formula is: (13) Where, is the sideslip angle at the center of mass, is the yaw angular velocity, is the yaw rate tracking error, is the center of mass sideslip angle tracking error, and is the integral term; The action space is defined as the key parameter of STSMC , its expression formula is: (14) Where, the super-helical sliding film yaw rate torque controller parameters are: is the sliding surface parameter, is the main approach rate coefficient of the yaw angular velocity; the controller parameters of the super-helical sliding film center of mass sideslip angle torque are: is the sliding surface parameter, is the main approach rate coefficient of the sideslip angle at the center of mass; Controller allocation coefficient; Reward Function It plays a significant role in DRL, guiding the agent to pursue the optimal control effect. Its expression formula is: (15) Where, is the tracking error weight coefficient.

[0010] As described above, the vehicle stability control method based on the DDPG adaptive super-helical sliding mode adopts the Actor-Critic network structure, generates parameter adjustment strategies through the Actor network, and evaluates the strategy value through the Critic network to achieve multi-parameter optimization.

[0011] The vehicle stability control method based on DDPG adaptive super spiral sliding mode as described above, the input state in the Actor network , output action , through the deterministic policy function Mapping state to optimal action, the expression formula is: (16) Where, Represents random noise, giving the agent the ability to explore; Represents the parameters of the Actor network at the current moment; Indicates that the real-time vehicle dynamics state is input through a deterministic strategy function , output its action ; The input state of the Critic network and action , output state-action value function , evaluate the long-term benefits of the current strategy; The network parameters in the Actor-Critic network structure are and , the Actor network calculates the policy gradient through the back propagation method , and then use the policy gradient ascent algorithm to update the online actor network weights , its expression formula is: (17) Where, Strategy parameters Find the gradient, Represents the policy function the gradient with respect to its parameters; To obtain the empirical average; Represents the learning rate of the Actor function network; The Critic network uses the gradient descent method to update the parameters : (18) in, It represents the deviation between the predicted value and the target value; is the target value, by integrating the immediate reward and future value Discount estimates; is the discount factor, ; Represents the learning rate of the Critic function network; To solve the problem of single network learning instability, two sets of networks, online and target, are deployed in actual applications. The target network parameters are used to calculate the TD target and update the online critic network weights. After obtaining the online network parameters, the parameters in the target network can be updated according to the soft update principle. The expression is: (19) Where, is the target Critic network parameter, is the target Actor network parameter, represents the update rate, .

[0012] In the vehicle stability control method based on the DDPG adaptive super-helical sliding mode described above, in step 5, the vehicle state quantity and the ideal values ​​of the yaw rate and the sideslip angle are input into the top-level controller, that is, input into the DDPG adaptive super-helical sliding mode controller established in step 4, to obtain the additional yaw torque to be output. , the additional yaw moment obtained is Reasonably distribute it to the four wheels to maintain vehicle driving stability.

[0013] In the vehicle stability control method based on the DDPG adaptive super-helical sliding mode described above, the specific operations of step 5 include the following steps: Step 1): The four-wheel torque must first meet the additional yaw moment and tire longitudinal force requirements determined by the upper-level motion tracking controller. The expression is: (20) in, 、 、 、 is the longitudinal force of the left front, right front, left rear and right rear wheels, is the wheelbase, For vehicles in The net external force in the direction For vehicles to go around The total external moment of the shaft; Rewrite formula (19) into matrix form: (twenty one) in, ; ; The minimum torque distribution error is defined as one of the torque optimization distribution objectives, and the objective function expression formula is: (twenty two) Among them, the matrix represents the longitudinal force and yaw moment demand weights; Step 2): The sum of the squares of the longitudinal force and lateral force of the four wheels is divided by the square of the product of the vertical load and the friction coefficient to represent the vehicle stability margin. For four-wheel independent drive electric vehicles, the lateral force is uncontrollable under the constraint conditions. Therefore, the longitudinal force is selected as the control variable, and the expression formula of the optimization objective function is: (twenty three) in, is the wheel lateral force, is the longitudinal force of the wheel, is the vertical load of the wheel; Formula (23) is rewritten into matrix form: in, ; Step 3): The two torque optimization distribution control objective functions are combined, and the expression formula is: (twenty four) in, is the weight coefficient of the objective function; The constraint conditions are constructed by considering three factors: tire utilization, motor output torque, and road adhesion conditions. The constraint conditions are expressed as follows: (25) According to the friction ellipse constraint formula (25), the longitudinal force constraint condition can be obtained, which is expressed as follows: (26) The maximum output torque of the motor should meet the following constraints, which are expressed as follows: (27) in, is the maximum output torque of the motor, r is the rolling radius of the tire; According to formula (26) and formula (27), the longitudinal force constraint condition of each wheel is expressed as follows: (28) According to formula (24) and formula (28), the allocation problem of control variables can be transformed into an optimization problem, which is expressed as follows: (29) By solving the QP, the torque optimization distribution result of the four-wheel motor driven electric vehicle is obtained.

[0014] The advantages of the present invention are: the present invention utilizes the second-order sliding mode structure of STSMC to reduce the chattering problem of traditional sliding mode control, and at the same time, through the DDPG dynamic optimization control strategy, adaptively adjusts parameters to enable it to quickly generate accurate yaw torque under complex road conditions and sudden working conditions; compared with the fixed structure control method, the present invention significantly improves the driving stability of distributed drive vehicles and their adaptability to different working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0016] Figure 1 It is a schematic diagram of the strategy framework of the present invention; Figure 2 It is a schematic diagram of the optimization process of the state space-action space-reward interaction mechanism of the present invention; Figure 3 is the learning training curve of the DDPG based on episode rewards of the present invention; Figure 4 This is one of the yaw rate comparison diagrams of the present invention at 100 km / h and a slippery road (with a road adhesion coefficient of 0.3); Figure 5 This is the second comparison diagram of the yaw rate at 100 km / h on a slippery road (with a road adhesion coefficient of 0.3) according to the present invention; Figure 6 This is one of the comparison graphs of the center of mass slip angle at 100 km / h and a slippery road (with a road adhesion coefficient of 0.3) according to the present invention; Figure 7 This is the second comparison diagram of the center of mass sideslip angle at 100 km / h and a slippery road (with a road adhesion coefficient of 0.3) according to the present invention; Figure 8 This is a comparison diagram of the torques of each wheel at 100 km / h and on a slippery road (with a road adhesion coefficient of 0.3) according to the present invention. DETAILED DESCRIPTION

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0018] like Figure 1 As shown in FIG, the vehicle stability control method based on the DDPG adaptive super-helical sliding mode includes the following steps: Step 1: Establish a linear 2-DOF reference model to obtain the ideal yaw rate and sideslip angle required for stable control; Step 2: Design a sliding mode controller; Step 3: Design a super-helical sliding mode controller based on step 2; Step 4: Design the DDPG adaptive super-helical sliding mode controller based on step 3; Step 5: Distribute the calculated additional yaw moment to each wheel to achieve vehicle stability control.

[0019] Preferably, in step 1 of this embodiment, by establishing the total lateral force balance equation and the yaw moment balance equation, combined with the linear assumption of the tire cornering force and the limitation of the road adhesion coefficient, the ideal yaw angular velocity under steady-state steering conditions can be obtained. and the center of mass slip angle , the specific expression formula is as follows: (1) Where, is the front axle wheelbase, is the rear axle wheelbase, is the front wheel deflection angle, is the road adhesion coefficient, is the vehicle longitudinal velocity, is the safety factor, 、 is the cornering stiffness of the front and rear tires, To describe the vehicle's steering stability coefficient , .

[0020] Preferably, in step 2 of this embodiment, an additional yaw moment is added to the linear 2-DOF vehicle dynamics model. , the specific expression formula is as follows: (2) Where, is the vehicle mass, is the moment of inertia of the vehicle around the z-axis, is the additional yaw moment, 、 are the yaw rate and its derivative, 、 are the sideslip angle of the center of mass and its derivative respectively; In step 2, define the yaw rate sliding film surface , Center of mass slip angle sliding surface The expression formulas are: (3) (4) Where, is the sliding film surface parameter of yaw rate, is the sliding surface parameter of the center of mass sideslip angle, all of which are greater than 0; 、 are the yaw rate tracking error and its derivative, 、 is the center of mass sideslip angle error and its derivative; In step 2, the yaw rate approaching rate is defined as , Center of mass sideslip angle approach rate The expression formula is: (5) (6) Where, is the sliding mode approach coefficient of yaw rate, is the sliding mode approach coefficient of the sideslip angle at the center of mass, is a symbolic function; Derivative of formula (3) and formula (5) are substituted into formula (2) to obtain the yaw rate and yaw torque based on the sliding mode controller: : (7) The additional yaw moment of the sideslip angle of the center of mass based on the sliding mode controller is obtained by derivation of formula (4) and substitution of formula (6) into formula (2): : (8) The distribution coefficient method is used to combine formula (7) and formula (8) to output the additional yaw moment. The total yaw moment is: : (9) Where, is the additional gain for the yaw rate, Adds a bonus to the slip angle.

[0021] Preferably, the expression formula for defining the superhelical sliding mode approach rate in step 3 of this embodiment is: (10) Where, is the control input, 、 is the gain of the super-helical sliding mode controller; Substitute the derivative of formula (3) and formula (10) into formula (2) to obtain the additional yaw moment of the yaw angular velocity based on the super-helical sliding mode controller: The expression formula is: (11) Where, 、 is the super-helical sliding mode yaw rate approach coefficient; Derivative of formula (4) and substitute formula (10) into formula (2) The center of mass slip angle moment and additional yaw moment based on super-helical sliding mode controller The expression formula is: (12) Where, 、 is the sideslip angle approach coefficient of the super-helical sliding mode center of mass.

[0022] like Figure 2 As shown, preferably, in step 4 of this embodiment, the DDPG algorithm is embedded in the STSMC framework to construct a state space-action space-reward interaction mechanism.

[0023] Preferably, the state space described in this embodiment is defined as the real-time vehicle dynamics state quantity , its expression formula is: (13) Where, is the sideslip angle at the center of mass, is the yaw angular velocity, is the yaw rate tracking error, is the center of mass sideslip angle tracking error, and is the integral term; The action space is defined as the key parameter of STSMC , its expression formula is: (14) Where, the super-helical sliding film yaw rate torque controller parameters are: is the sliding surface parameter, is the main approach rate coefficient of the yaw angular velocity; the controller parameters of the super-helical sliding film center of mass sideslip angle torque are: is the sliding surface parameter, is the main approach rate coefficient of the sideslip angle at the center of mass; Controller allocation coefficient; Reward Function It plays a significant role in DRL, guiding the agent to pursue the optimal control effect. Its expression formula is: (15) Where, is the tracking error weight coefficient.

[0024] Preferably, the DDPG algorithm described in this embodiment adopts an Actor-Critic network structure, generates parameter adjustment strategies through the Actor network, and evaluates the strategy value through the Critic network to achieve multi-parameter optimization. Its learning and training curve is as follows: Figure 3shown.

[0025] Preferably, the input state in the Actor network described in this embodiment , output action , through the deterministic policy function Mapping state to optimal action, the expression formula is: (16) Where, Represents random noise, giving the agent the ability to explore; Represents the parameters of the Actor network at the current moment; Indicates that the real-time vehicle dynamics state is input through a deterministic strategy function , output its action ; The input state of the Critic network and action , output state-action value function , evaluate the long-term benefits of the current strategy; The network parameters in the Actor-Critic network structure are and , the Actor network calculates the policy gradient through the back propagation method , and then use the policy gradient ascent algorithm to update the online actor network weights , its expression formula is: (17) Where, Strategy parameters Find the gradient, Represents the policy function the gradient with respect to its parameters; To obtain the empirical average; Represents the learning rate of the Actor function network; The Critic network uses the gradient descent method to update the parameters : (18) in, It represents the deviation between the predicted value and the target value; is the target value, by integrating the immediate reward and future value Discount estimates; is the discount factor, ; Represents the learning rate of the Critic function network; To solve the problem of single network learning instability, two sets of networks, online and target, are deployed in actual applications. The target network parameters are used to calculate the TD target and update the online critic network weights. After obtaining the online network parameters, the parameters in the target network can be updated according to the soft update principle. The expression is: (19) Where, is the target Critic network parameter, is the target Actor network parameter, represents the update rate, .

[0026] Preferably, in step 5 of this embodiment, the vehicle state and the ideal values ​​of the yaw rate and the sideslip angle are input into the top-level controller, that is, input into the DDPG adaptive super-helical sliding mode controller established in step 4, to obtain the additional yaw moment to be output. , the additional yaw moment obtained is Reasonably distribute it to the four wheels to maintain vehicle driving stability.

[0027] Preferably, the specific operations of step 5 described in this embodiment include the following steps: Step 1): The four-wheel torque must first meet the additional yaw moment and tire longitudinal force requirements determined by the upper-level motion tracking controller. The expression is: (20) in, 、 、 、 is the longitudinal force of the left front, right front, left rear and right rear wheels, is the wheelbase, For vehicles in The net external force in the direction For vehicles to go around The total external moment of the shaft; Rewrite formula (19) into matrix form: (twenty one) in, ; ; The minimum torque distribution error is defined as one of the torque optimization distribution objectives, and the objective function expression formula is: (twenty two) Among them, the matrix represents the longitudinal force and yaw moment demand weights; Step 2): The sum of the squares of the longitudinal force and lateral force of the four wheels is divided by the square of the product of the vertical load and the friction coefficient to represent the vehicle stability margin. For four-wheel independent drive electric vehicles, the lateral force is uncontrollable under the constraint conditions. Therefore, the longitudinal force is selected as the control variable, and the expression formula of the optimization objective function is: (twenty three) in, is the wheel lateral force, is the longitudinal force of the wheel, is the vertical load of the wheel; Formula (23) is rewritten into matrix form: in, ; Step 3): The two torque optimization distribution control objective functions are combined, and the expression formula is: (twenty four) in, is the weight coefficient of the objective function; The constraint conditions are constructed by considering three factors: tire utilization, motor output torque, and road adhesion conditions. The constraint conditions are expressed as follows: (25) According to the friction ellipse constraint formula (25), the longitudinal force constraint condition can be obtained, which is expressed as follows: (26) The maximum output torque of the motor should meet the following constraints, which are expressed as follows: (27) in, is the maximum output torque of the motor, r is the rolling radius of the tire; According to formula (26) and formula (27), the longitudinal force constraint condition of each wheel is expressed as follows: (28) According to formula (24) and formula (28), the allocation problem of control variables can be transformed into an optimization problem, which is expressed as follows: (29) By solving the QP, the torque optimization distribution result of the four-wheel motor driven electric vehicle is obtained.

[0028] In order to verify the feasibility and effectiveness of this method, the vehicle stability control simulation was carried out using the method of the present invention in the CarSim / Simulink platform, and the simulation results were compared with the traditional sliding mode control (SMC) method. Table 1 shows the vehicle parameters. The double lane change condition was used, the vehicle speed was 100 km / h, and the road adhesion coefficient was 0.3. The simulation test results are as follows: Figure 4-Figure 8 shown.

[0029] Table 1 In summary, the specific implementation steps and content of the DDPG adaptive super-helical sliding mode vehicle stability control method are summarized. The second-order sliding mode structure of STSMC is utilized to mitigate the chattering problem of traditional sliding mode control. Furthermore, the DDPG dynamic optimization control strategy and adaptive parameter adjustment enable rapid and precise generation of yaw torque under complex road conditions and unexpected operating conditions.

[0030] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A vehicle stability control method based on DDPG adaptive super-helical sliding mode, characterized by: The steps include: Step 1: Establish a linear 2-DOF reference model to obtain the ideal yaw rate and sideslip angle required for stable control; Step 2: Design a sliding mode controller; Step 3: Design a super-helical sliding mode controller based on step 2; Step 4: Design the DDPG adaptive super-helical sliding mode controller based on step 3; Step 5: Distribute the calculated additional yaw moment to each wheel to achieve vehicle stability control.

2. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 1, characterized in that: The step 1 establishes the total lateral force balance equation and the yaw moment balance equation, and combines the linear assumption of tire cornering force and the limitation of road adhesion coefficient to obtain the ideal yaw angular velocity under steady-state steering conditions. and the center of mass slip angle , the specific expression formula is as follows: （1） Where, is the front axle wheelbase, is the rear axle wheelbase, is the front wheel deflection angle, is the road adhesion coefficient, is the vehicle longitudinal velocity, is the safety factor, 、 is the cornering stiffness of the front and rear tires, To describe the vehicle's steering stability coefficient , .

3. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 1, characterized in that: In step 2, an additional yaw moment is added to the linear 2-DOF vehicle dynamics model. , the specific expression formula is as follows: （2） Where, is the vehicle mass, is the moment of inertia of the vehicle around the z-axis, is the additional yaw moment, 、 are the yaw rate and its derivative, 、 are the sideslip angle of the center of mass and its derivative respectively; In step 2, define the yaw rate sliding film surface , Center of mass slip angle sliding surface The expression formulas are: （3） （4） Where, is the sliding film surface parameter of yaw rate, is the sliding surface parameter of the center of mass sideslip angle, all of which are greater than 0; 、 are the yaw rate tracking error and its derivative, 、 is the center of mass sideslip angle error and its derivative; In step 2, the yaw rate approaching rate is defined as , Center of mass sideslip angle approach rate The expression formula is: （5） （6） Where, is the sliding mode approach coefficient of yaw rate, is the sliding mode approach coefficient of the sideslip angle at the center of mass, is a symbolic function; Derivative of formula (3) and formula (5) are substituted into formula (2) to obtain the yaw rate and yaw torque based on the sliding mode controller: : （7） The additional yaw moment of the sideslip angle of the center of mass based on the sliding mode controller is obtained by derivation of formula (4) and substitution of formula (6) into formula (2): : （8） The distribution coefficient method is used to combine formula (7) and formula (8) to output the additional yaw moment. The total yaw moment is: : （9） Where, is the additional gain for the yaw rate, Adds a bonus to the slip angle.

4. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 3, characterized in that: The expression formula for defining the super-helical sliding mode approach rate in step 3 is: （10） Where, is the control input, 、 is the gain of the super-helical sliding mode controller; Substitute the derivative of formula (3) and formula (10) into formula (2) to obtain the additional yaw moment of the yaw angular velocity based on the super-helical sliding mode controller: The expression formula is: （11） Where, 、 is the super-helical sliding mode yaw rate approach coefficient; Derivative of formula (4) and substitute formula (10) into formula (2) The center of mass slip angle moment and additional yaw moment based on super-helical sliding mode controller The expression formula is: （12） Where, 、 is the sideslip angle approach coefficient of the super-helical sliding mode center of mass.

5. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 4 is characterized in that: In step 4, the DDPG algorithm is embedded in the STSMC framework to construct a state space-action space-reward interaction mechanism.

6. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 5, characterized in that: The state space is defined as the real-time vehicle dynamics state quantity , its expression formula is: (13) Where, is the sideslip angle at the center of mass, is the yaw angular velocity, is the yaw rate tracking error, is the center of mass sideslip angle tracking error, and is the integral term; The action space is defined as the key parameter of STSMC , its expression formula is: (14) Where, the super-helical sliding film yaw rate torque controller parameters are: is the sliding surface parameter, is the main approach rate coefficient of the yaw angular velocity; the controller parameters of the super-helical sliding film center of mass sideslip angle torque are: is the sliding surface parameter, is the main approach rate coefficient of the sideslip angle at the center of mass; Controller allocation coefficient; Reward Function It plays a significant role in DRL, guiding the agent to pursue the optimal control effect. Its expression formula is: (15) Where, is the tracking error weight coefficient.

7. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 5, characterized in that: The DDPG algorithm adopts an Actor-Critic network structure, generates parameter adjustment strategies through the Actor network, and evaluates the strategy value through the Critic network to achieve multi-parameter optimization.

8. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 7, characterized in that: Input state in the Actor network , output action , through the deterministic policy function Mapping state to optimal action, the expression formula is: (16) Where, Represents random noise, giving the agent the ability to explore; Represents the parameters of the Actor network at the current moment; Indicates that the real-time vehicle dynamics state is input through a deterministic strategy function , output its action ; The input state of the Critic network and action , output state-action value function , evaluate the long-term benefits of the current strategy; The network parameters in the Actor-Critic network structure are and , the Actor network calculates the policy gradient through the back propagation method , and then use the policy gradient ascent algorithm to update the online actor network weights , its expression formula is: (17) Where, Strategy parameters Find the gradient, Represents the policy function the gradient with respect to its parameters; To obtain the empirical average; Represents the learning rate of the Actor function network; The Critic network uses the gradient descent method to update the parameters : (18) in, It represents the deviation between the predicted value and the target value; is the target value, by integrating the immediate reward and future value Discount estimates; is the discount factor, ; Represents the learning rate of the Critic function network; To solve the problem of single network learning instability, two sets of networks, online and target, are deployed in actual applications. The target network parameters are used to calculate the TD target and update the online critic network weights. After obtaining the online network parameters, the parameters in the target network can be updated according to the soft update principle. The expression is: (19) Where, is the target Critic network parameter, is the target Actor network parameter, represents the update rate, .

9. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 8, characterized in that: In step 5, the vehicle state and the ideal values ​​of the yaw rate and the sideslip angle are input into the top controller, that is, input into the DDPG adaptive super-helical sliding mode controller established in step 4 to obtain the additional yaw moment to be output. , the additional yaw moment obtained is Reasonably distribute it to the four wheels to maintain vehicle driving stability.

10. The vehicle stability control method based on DDPG adaptive super-helical sliding mode according to claim 8, characterized in that: The specific operations of step 5 include the following steps: Step 1): The four-wheel torque must first meet the additional yaw moment and tire longitudinal force requirements determined by the upper-level motion tracking controller. The expression is: (20) in, 、 、 、 is the longitudinal force of the left front, right front, left rear and right rear wheels, is the wheelbase, For vehicles in The net external force in the direction For vehicles to go around The total external moment of the shaft; Rewrite formula (19) into matrix form: (21) in, ； ； The minimum torque distribution error is defined as one of the torque optimization distribution objectives, and the objective function expression formula is: (22) Among them, the matrix represents the longitudinal force and yaw moment demand weights; Step 2): The sum of the squares of the longitudinal force and lateral force of the four wheels is divided by the square of the product of the vertical load and the friction coefficient to represent the vehicle stability margin. For four-wheel independent drive electric vehicles, the lateral force is uncontrollable under the constraint conditions. Therefore, the longitudinal force is selected as the control variable, and the expression formula of the optimization objective function is: (23) in, is the wheel lateral force, is the longitudinal force of the wheel, is the vertical load of the wheel; Formula (23) is rewritten into matrix form: in, ； Step 3): The two torque optimization distribution control objective functions are combined, and the expression formula is: (24) in, is the weight coefficient of the objective function; The constraint conditions are constructed by considering three factors: tire utilization, motor output torque, and road adhesion conditions. The constraint conditions are expressed as follows: (25) According to the friction ellipse constraint formula (25), the longitudinal force constraint condition can be obtained, which is expressed as follows: (26) The maximum output torque of the motor should meet the following constraints, which are expressed as follows: (27) in, is the maximum output torque of the motor, r is the rolling radius of the tire; According to formula (26) and formula (27), the longitudinal force constraint condition of each wheel is expressed as follows: (28) According to formula (24) and formula (28), the allocation problem of control variables can be transformed into an optimization problem, which is expressed as follows: (29) By solving the QP, the torque optimization distribution result of the four-wheel motor driven electric vehicle is obtained.

Images