Steering and stability coordination control method for distributed driving electric vehicle and related equipment

Abstract

The invention discloses a steering and stability coordination control method and related equipment for a distributed driving electric vehicle, and the method comprises the steps: firstly constructing a distributed motor and vehicle dynamics integrated model, and obtaining vehicle driving state data such as a side slip angle and a yaw velocity; then, on the basis of a phase plane analysis method, a correlation function is designed according to the side slip angle and the change rate of the side slip angle, the vehicle yaw stable state is quantitatively evaluated in real time, and the vehicle yaw stable state is divided into a stable state, a transition state and an unstable state; then, acquiring reference values of the yaw velocity and the side slip angle by using a preset reference model, and designing a state space and a reward function of the TD3 intelligent agent in combination with the correlation function and the driving state data; and generating a random target speed sequence through a Gaussian process, inputting the random target speed sequence into an integrated model to collect a data set, and training the TD3 intelligent agent until convergence. And finally, real-time vehicle driving state data are input into the trained TD3 intelligent agent, driving torques of four driving motors are output, differential power-assisted steering and stability coordination control is achieved, the problem of actuator conflict is solved, vehicle stability is guaranteed, and optimal steering assistance is achieved.

Description

Technical Field

[0001] This invention belongs to the field of electric vehicle control technology, specifically relating to a steering and stability coordination control method and related equipment for distributed drive electric vehicles. Background Technology

[0002] Distributed drive electric vehicles are equipped with independent drive motors at each wheel, enabling precise and independent control of the torque of each wheel. Through a well-designed torque distribution strategy, distributed drive and differential drive systems can not only generate steering assist torque but also further improve vehicle handling stability. Differential power steering (DDAS) applies differential torque between the left and right wheels to generate auxiliary steering torque, thereby improving vehicle handling ease. However, in complex conditions such as high-speed driving, low-traction surfaces, or emergency obstacle avoidance, a dynamic conflict may arise between steering comfort and vehicle stability. If this conflict is not properly coordinated, it can not only weaken the power steering effect but also jeopardize vehicle driving stability.

[0003] To address the aforementioned issues, existing technologies often employ control strategies such as LQR and MPC. These traditional control strategies frequently rely on high-precision vehicle models, and their control performance deteriorates significantly and exhibits poor adaptability when faced with environmental disturbances, model uncertainties, and system nonlinear dynamics. Therefore, a coordinated control scheme is needed that can adapt to complex and changing driving environments, requires no precise mathematical models, and intelligently resolve strategy conflicts. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a steering and stability coordination control method and related equipment for distributed drive electric vehicles. Its purpose is to adaptively and dynamically adjust the control priority to achieve the optimal steering assistance effect while ensuring vehicle stability, and to solve the actuator conflict problem between differential power steering and vehicle stability.

[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: According to a first aspect of the present invention, a method for coordinated steering and stability control of a distributed drive electric vehicle is provided, comprising: An integrated model of distributed motor and vehicle dynamics is constructed. Vehicle driving state data is obtained through the integrated model. The driving state data includes the center of gravity sideslip angle, yaw rate, lateral acceleration, steering wheel angle, steering wheel angular velocity, and the torque applied to the steering wheel by the driver. Based on phase plane analysis, a correlation function is designed according to the centroid sideslip angle and the rate of change of the centroid sideslip angle. The correlation function is used to quantitatively evaluate the current yaw stability of the vehicle in real time. The yaw stability includes stable state, transient state and unstable state. A preset reference model is used to obtain the yaw rate reference value and the center of gravity sideslip angle reference value. Combined with the correlation function and driving status data packet, the state space and reward function of the TD3 agent are designed. A random target velocity sequence is generated through a Gaussian process. The target velocity sequence is input into the ensemble model, a complete vehicle driving state dataset is collected, and the TD3 agent is trained using the dataset until the TD3 agent converges. The real-time vehicle driving status data is input into the TD3 agent after training and convergence. The TD3 agent outputs the driving torque of the four drive motors to achieve coordinated control of differential power steering and stability of the distributed drive electric vehicle.

[0006] In one possible implementation of the first aspect, the specific process of constructing the integrated model of the distributed motor model and the vehicle dynamics model is as follows: A basic model of a four-wheel independent drive vehicle is established in CarSim. The original transmission system is removed, and the motor torque response is approximated using a second-order transfer function. The second-order transfer function is as follows:

[0007] In the formula, It is a second-order transfer function. This is the actual electromagnetic torque. For the target electromagnetic torque, For extreme logarithms, For rotor flux linkage, For stator resistance, For mutual intuition, For self-perception; The transfer function of a voltage pulse width modulation circuit. It is a time constant. For Laplace complex frequency domain variables, Represents the characteristic parameters of the motor; A four-degree-of-freedom steering system model was built in Simulink. The four-degree-of-freedom steering system model was then integrated with the basic four-wheel independent drive vehicle model in CarSim to obtain a distributed motor model and a vehicle dynamics integrated model.

[0008] In one possible implementation of the first aspect, the process of designing the correlation function includes: (1) A phase plane is constructed based on the centroid sideslip angle and the rate of change of the centroid sideslip angle. The phase plane is divided into a stable region and a critical region by an elliptical boundary. The classical domain expression is defined as follows:

[0009] In the formula, for The weighting coefficients, for The weighting coefficients are determined by the vehicle parameters; Let be the slope of the boundary line, and let be the intercept. ; The centroid side deflection angle of the classical domain boundary; Let be the rate of change of the centroid sideslip angle at the boundary of the classical domain; , , The centroid side deflection angle of the positive domain boundary; The rate of change of the centroid sideslip angle of the positive domain boundary; (2) Introduce characteristic parameters that characterize the vehicle's driving state The two-dimensional domain is simplified to a one-dimensional domain, and the feature parameters The expression is:

[0010] In the formula, For characteristic parameters, The sideslip angle is the vehicle's center of gravity. The rate of change of the vehicle's center of gravity sideslip angle. As a weighting factor; (3) The extended distance is used to describe the positional relationship between the feature parameters and the classical domain. The extended distance from any feature parameter to the classical domain is:

[0011] In the formula, This represents the extension distance of the feature parameters relative to the classical domain interval. For classical domain intervals; Arbitrary feature parameters to the positive region The extension distance is:

[0012] In the formula, This represents the expansion distance of the feature parameters relative to the positive domain interval. The positive interval; (4) Define the correlation function as:

[0013] in, ; like If so, the vehicle is in a stable state; like If so, the vehicle is in a transitional state from stable to unstable. like If so, the vehicle is in an unstable state; In the formula, For correlation functions, The extension distance of the feature parameters relative to the classical domain interval, This represents the difference in distance between domains.

[0014] In one possible implementation of the first aspect, the preset reference model is a linear two-degree-of-freedom model, and the yaw rate reference value... Reference value of centroid side slip angle The calculation formulas are as follows:

[0015] In the formula, This is the critical centroid sideslip angle. It is the acceleration due to gravity. For vehicle quality, For longitudinal velocity, The road surface adhesion coefficient, and These are the distances from the center of mass to the front and rear axles, respectively. This refers to the vehicle's wheelbase. For rear tire lateral stiffness, The stability coefficient, This refers to the steering angle of the front wheels.

[0016] In one possible implementation of the first aspect, the state space of the TD3 agent... for:

[0017] In the formula, Representing the state space, This represents the difference in yaw rate, where , The yaw rate is angular velocity. This is the reference value for yaw rate. This represents the difference in the centroid sideslip angle, where , The sideslip angle is the vehicle's center of gravity. This is a reference value for the centroid sideslip angle. Indicates the steering wheel angle; Indicates the angular velocity of the steering wheel; This indicates the torque applied by the driver to the steering wheel; This indicates lateral acceleration.

[0018] In one possible implementation of the first aspect, the reward function of the TD3 agent... The comprehensive expression is:

[0019] in,

[0020]

[0021]

[0022]

[0023] In the formula, For the reward function, For correlation functions, For characteristic parameters, As the first reward sub-item, As the second reward sub-item, As the third reward sub-item, This is the fourth reward sub-item. The yaw rate is angular velocity. This is the reference value for yaw rate. The sideslip angle is the vehicle's center of gravity. This is a reference value for the centroid sideslip angle. For steering wheel angle, The angular velocity of the steering wheel. This is a reference value for the steering wheel angle. This indicates the torque applied to the steering wheel by the driver.

[0024] In one possible implementation of the first aspect, the Gaussian process employs a radial basis function kernel as the covariance function, and the expression for the radial basis function kernel is:

[0025] in, Let the covariance function be defined at any two time points. Indicates the time input point. Indicates another time input point, For signal variance, For length scale, To control the smoothness of changes in vehicle speed.

[0026] In one possible implementation of the first aspect, the driving torques of the four drive motors output by the TD3 agent are respectively the driving torques of the left front wheel. Right front wheel drive torque Left rear wheel drive torque and right rear wheel drive torque That is, output action .

[0027] According to a second aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the aforementioned steering and stability coordination control method for a distributed drive electric vehicle.

[0028] According to a third aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned steering and stability coordination control method for a distributed drive electric vehicle.

[0029] According to a fourth aspect of the present invention, a computer program product is provided that, when executed by a processor, implements the aforementioned steering and stability coordination control method for a distributed drive electric vehicle.

[0030] Compared with the prior art, the present invention has at least the following beneficial effects: This invention provides a steering and stability coordination control method for distributed drive electric vehicles. Addressing the dynamic conflicts that may arise between differential power steering and vehicle stability control under complex conditions, it utilizes a TD3 agent for real-time decision-making and torque allocation. This intelligently coordinates the weights between steering assist and stability control without relying on high-precision vehicle mathematical models, ensuring vehicle yaw stability while providing appropriate steering assist torque and avoiding performance degradation or instability risks caused by conflicting control objectives. A correlation function is constructed based on phase plane analysis to quantitatively identify whether the vehicle is in a stable, transitional, or unstable state in real time. This allows the control system to dynamically adjust the priority of control strategies based on the current stability margin, enhancing the system's ability to perceive changes in vehicle state. This results in control responses that better match actual stability requirements under conditions such as low-adhesion surfaces and high-speed steering. End-to-end torque allocation decisions using the TD3 agent overcome the dependence on model accuracy and linearization assumptions of traditional methods such as LQR and MPC. During training, the agent learns the complex nonlinear relationship between vehicle state and optimal torque mapping, improving control performance under model uncertainty and external disturbances. This invention employs a Gaussian process to generate diverse velocity sequences during the training phase, covering dynamic scenarios across multiple operating conditions, ensuring the agent learns comprehensive control strategies. In practical applications, the converged agent can rapidly output the driving torque of the four motors based on real-time vehicle status data, while simultaneously ensuring that the yaw rate and sideslip angle efficiently track reference values, resulting in easy vehicle handling and stable driving. This method is not only suitable for conventional driving conditions but also effectively addresses scenarios such as high-speed lane changes, low-friction road steering, and emergency obstacle avoidance. By intelligently coordinating and distributing torque to each wheel, it maintains vehicle stability while reducing the driver's steering burden. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the specific embodiments of the present invention, the drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0032] Figure 1 This is a flowchart of a steering and stability coordination control method for a distributed drive electric vehicle according to the present invention.

[0033] Figure 2 This is a control strategy diagram for a steering and stability coordination control method for a distributed drive electric vehicle according to the present invention.

[0034] Figure 3 This is a phase plane diagram of the centroid side deflection angle and the rate of change of the centroid side deflection angle according to the present invention.

[0035] Figure 4 This is an architecture diagram of the TD3 intelligent agent of the present invention.

[0036] Figure 5 This is a diagram showing the training results of the TD3 agent of this invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] Combination Figure 1 and Figure 2 As shown, this invention provides a method for coordinated steering and stability control of a distributed drive electric vehicle, specifically including the following steps: Step 1: Construct an integrated model of the distributed motor and the vehicle dynamics. Obtain vehicle driving state data through the integrated model. The driving state data includes the center of gravity sideslip angle, yaw rate, lateral acceleration, steering wheel angle, steering wheel angular velocity, and the torque applied to the steering wheel by the driver.

[0039] In one feasible approach, the specific process of constructing the integrated model of the distributed motor model and the vehicle dynamics model is as follows: Step 1.1: Establish a basic model of a four-wheel independent drive vehicle in CarSim, remove the original transmission system, and use a second-order transfer function to approximate the motor torque response, balancing model accuracy and computational complexity.

[0040] The second-order transfer function is:

[0041] In the formula, It is a second-order transfer function. This is the actual electromagnetic torque. For the target electromagnetic torque, For extreme logarithms, For rotor flux linkage, For stator resistance, For mutual intuition, For self-perception; The transfer function of a voltage pulse width modulation circuit. It is a time constant. For Laplace complex frequency domain variables, These represent the characteristic parameters of the motor.

[0042] Step 1.2: Build a four-degree-of-freedom steering system model in Simulink, and integrate the four-degree-of-freedom steering system model with the basic four-wheel independent drive vehicle model in CarSim to obtain the integrated model of distributed motor and vehicle dynamics.

[0043] Specifically, the four-degree-of-freedom steering system model is as follows:

[0044]

[0045]

[0046]

[0047]

[0048]

[0049] In the formula, Steering wheel torque; and These are the steering column damping and the equivalent rotational inertia of the steering wheel and steering column, respectively. Let be the torsional stiffness of the torsion bar; and These are the rack mass and damping, respectively; and These are the moment of inertia of the wheel about the kingpin and the kingpin damping, respectively. This is the transmission ratio between the rack displacement and the steering knuckle angular displacement; Steering wheel angle; This represents the rack displacement; The radius of the pinion; and These are the forward transmission efficiency and reverse transmission efficiency of the steering gear, respectively. This refers to the return torque of the left and right front wheels; Coulomb friction is the force exerted by the gear and rack. The Coulomb friction force generated by the left and right front wheels rotating around the kingpin; The total torque transmitted to the kingpins of the left and right front wheels; This refers to the front wheel steering angle; This refers to the torsional stiffness of the front wheel kingpin.

[0050] Step 2: Based on the phase plane analysis method, design a correlation function according to the centroid sideslip angle and the rate of change of the centroid sideslip angle, and use the correlation function to quantitatively evaluate the current yaw stability of the vehicle in real time. The yaw stability includes stable state, transitional state and unstable state.

[0051] The yaw stability of a car can be analyzed by using the phase plane of vehicle sideslip dynamics.

[0052] To improve the learning efficiency and adaptive capability of intelligent agents in complex dynamic environments, this invention proposes a phase plane correlation function for dynamically adjusting the weights of the reward function. This mechanism can utilize the changing characteristics of key dynamic parameters in the phase plane to quantitatively assess the impact of the current operating condition on vehicle stability in real time: in the stable region, higher weights are assigned to differential drive power steering performance indicators such as steering tracking accuracy and driver comfort; after entering the critical region, the reward function is dominated by stability control terms such as yaw rate error and sideslip angle.

[0053] In one feasible approach, combining Figure 3 As shown, the process of designing an association function includes: Step 2.1: Construct a phase plane based on the sideslip angle and the rate of change of the sideslip angle. The phase plane is divided into a stable region and a critical region by an elliptical boundary. In other words, the inner side is the stable region, where the vehicle remains controllable; the outer side is the critical region, where the vehicle is about to lose control.

[0054] based on For a biphasic planar graph, the classical domain expression is defined as follows:

[0055] In the formula, for The weighting coefficients, for The weighting coefficients are determined by the vehicle parameters; Let be the slope of the boundary line, and let be the intercept. ; The centroid side slip angle of the classical domain boundary. Let be the rate of change of the centroid sideslip angle at the boundary of the classical domain; , ; The centroid sideslip angle of the positive domain boundary. denoted as the rate of change of the centroid side deflection angle at the boundary of the positive domain. and The value is related to the driving speed and the road surface adhesion coefficient. It is obtained through simulation data under different working conditions. and Numerical model.

[0056] Step 2.2: To describe the relationship between vehicle state points and regions in the phase plane, feature parameters characterizing the vehicle's driving state are introduced. This simplifies the two-dimensional domain to a one-dimensional domain. The feature parameters... The expression is:

[0057] In the formula, For characteristic parameters, The sideslip angle is the vehicle's center of gravity. The rate of change of the vehicle's center of gravity sideslip angle. This is the weighting factor.

[0058] Step 2.3: The extended distance is used to describe the positional relationship between the feature parameters and the classical domain. The extended distance from any feature parameter to the classical domain is:

[0059] In the formula, This represents the extension distance of the feature parameters relative to the classical domain interval. For characteristic parameters, For classical domain intervals, The centroid side deflection angle of the classical domain boundary.

[0060] Similarly, any feature parameters to the positive region The extension distance is:

[0061] In the formula, This represents the expansion distance of the feature parameters relative to the positive domain interval. For characteristic parameters, For positive intervals, The centroid side deflection angle of the positive domain boundary.

[0062] Step 2.4: Define the association function as follows:

[0063] in, ; This correlation function reflects the degree of correlation between the system state and the extended phase plane, and fully considers the change process of the vehicle's yaw stability state. It divides the yaw rate stability into three states: stable state, transient state, and unstable state, as detailed below: like If the characteristic state is in the stable region, the vehicle is in a stable state. like If the characteristic state is in the transition region, the vehicle is in a transitional state from stable to unstable. like If the characteristic state is in the unstable region, the vehicle is in an unstable state.

[0064] In the formula, For correlation functions, The feature parameter is the extension distance relative to the positive domain interval. The extension distance of the feature parameters relative to the classical domain interval, For inter-domain extended distance difference, For characteristic parameters, For classical domain intervals, It is a positive interval.

[0065] Step 3: Obtain the yaw rate reference value and the center of gravity sideslip angle reference value using a preset reference model. Combine the correlation function and driving status data package to design the state space and reward function of the TD3 agent. The reward function is used to transform the differential power steering and stability coordination control objective into a cost minimization problem.

[0066] In one possible implementation, the preset reference model is a linear two-degree-of-freedom model, and the yaw rate reference value... Reference value of centroid side slip angle The calculation formulas are as follows:

[0067] In the formula, This is the reference value for yaw rate. This is a reference value for the centroid sideslip angle. This is the critical centroid sideslip angle. It is the acceleration due to gravity. For vehicle quality, For longitudinal velocity, The road surface adhesion coefficient, and These are the distances from the center of mass to the front and rear axles, respectively. This refers to the vehicle's wheelbase. For rear tire lateral stiffness, The stability coefficient, This refers to the steering angle of the front wheels.

[0068] like Figure 4 The diagram shows the architecture of the TD3 agent. The TD3 agent is an improvement on the deep reinforcement learning algorithm, which solves the problems of estimation bias and training instability that are prone to occur in the deep reinforcement learning algorithm.

[0069] In one possible implementation, the state space of the TD3 agent... for:

[0070] In the formula, Representing the state space, This represents the difference in yaw rate, where , The yaw rate is angular velocity. This is the reference value for yaw rate. This represents the difference in the centroid sideslip angle, where , The sideslip angle is the vehicle's center of gravity. This is a reference value for the centroid sideslip angle. Indicates the steering wheel angle; Indicates the angular velocity of the steering wheel; This indicates the torque applied by the driver to the steering wheel; This indicates lateral acceleration.

[0071] That is, the state space of the TD3 agent. Includes the following state variable: yaw rate , centroid side slip angle Steering wheel angle Steering wheel angular velocity Torque applied to the steering wheel by the driver and lateral acceleration .

[0072] In one possible implementation, the reward function of the TD3 agent The comprehensive expression is:

[0073] The components of the reward function are as follows:

[0074]

[0075]

[0076]

[0077] In the formula, For the reward function, For correlation functions, For characteristic parameters, As the first reward sub-item, As the second reward sub-item, As the third reward sub-item, This is the fourth reward sub-item. The yaw rate is angular velocity. This is the reference value for yaw rate. The sideslip angle is the vehicle's center of gravity. This is a reference value for the centroid sideslip angle. For steering wheel angle, The angular velocity of the steering wheel. This is a reference value for the steering wheel angle. This indicates the torque applied to the steering wheel by the driver.

[0078] The reward function is in the form of a negative quadratic form, transforming the control objective into a cost minimization problem. Key components of the reward function include: yaw rate error, sideslip angle error, steering angle, and actuator smoothness constraints.

[0079] Specifically, yaw rate error reflects directional stability during steering; reducing this error lowers the risk of oversteer or understeer. Side slip angle error is related to lateral dynamic stability; on low-friction surfaces, controlling the magnitude of the side slip angle error can prevent excessive lateral slip. The deviation between the actual front wheel steering angle and the driver's input reflects the accuracy of the steering intention transmission, thus achieving good human-machine collaboration in differential power steering systems. Furthermore, penalizing high-frequency changes in the steering angle prevents mechanical vibration in the steering actuator and indirectly filters high-frequency noise. The steering torque applied to the driver.

[0080] Step 4: Generate a random target velocity sequence through a Gaussian process, input the target velocity sequence into the ensemble model, collect a complete vehicle driving state dataset, and use the dataset to train the TD3 agent until the TD3 agent converges.

[0081] To elaborate, the detailed process is as follows: Step 4.1: Generate a random target velocity sequence using a Gaussian process.

[0082] As a Bayesian nonparametric model, Gaussian processes can learn the underlying function distribution directly from the data without explicitly defining the function form, and can also quantify uncertainty, making them very suitable for modeling nonlinear and random vehicle driving characteristics.

[0083] Assume the function of vehicle speed changing with time is Furthermore, the function follows a Gaussian process:

[0084] in, Let be the vehicle speed at time t. This represents the Gaussian process operator. It is a mean function, and is usually set to zero. Let be the covariance function at any two times. Indicates the time input point. This indicates another time input point.

[0085] This implementation uses a radial basis function kernel as the covariance function, and its expression is:

[0086] in, Let be the vehicle speed at time t. This represents the Gaussian process operator. It is a mean function. Let the covariance function be defined at any two time points. Indicates the time input point. Indicates another time input point, For signal variance, For length scale, Used to control the smoothness of changes in vehicle speed.

[0087] Given training data And assume the existence of observation noise. The test points are derived using the following notation. Predicted distribution at:

[0088]

[0089]

[0090] Then test point The predicted distribution at location is:

[0091] in:

[0092]

[0093] In the formula, Given training data, Indicates the first training set i A time input point, Indicates in i The vehicle speed is constantly monitored. The number of training samples, To observe the noise, This indicates that the mean is 0 and the variance is 0. Gaussian distribution, To observe the noise variance, The covariance matrix of the training samples, For the first in the training set j A time input point, Indicates in The covariance function value under the given conditions, For the test time point, This represents the covariance matrix between the test points and the training points. Indicates in The covariance function value under the given conditions, This represents the covariance matrix of the test point itself. Indicates at the test point The covariance function value under the given conditions, For test points The speed of the car, The mean is variance is Gaussian distribution, The mean of the test points. The variance of the test points, for identity matrix To train the output vehicle speed vector, This is the sample index, with a value range of 1 to... n .

[0094] By analyzing a set of time points By making predictions and sampling from the obtained Gaussian distribution, new vehicle speed-time trajectories can be generated:

[0095] In the formula, Let be the random variable representing the vehicle speed at time t. The mean is variance is Gaussian distribution, The predicted mean at time t, Let be the predicted variance at time t. This is the upper limit of the time range for generating the velocity trajectory.

[0096] Step 4.2: Input the obtained target speed sequence into the ensemble model, collect a complete vehicle driving state dataset, and train the TD3 agent based on this dataset until it converges.

[0097] In other words, a smooth and random target vehicle speed sequence is generated using a Gaussian process, and the generated vehicle speed samples are input into the Carsim vehicle dynamics model to collect a complete vehicle driving state dataset.

[0098] It should be noted that after collecting the complete vehicle driving status dataset, the dataset is converted into a time-series transition dataset. ,in Indicates state, Indicates an action, Indicates the next state. This represents a reward. The TD3 agent is trained using this transitional dataset.

[0099] Figure 5 The results distribution of the TD3 agent over 5000 training rounds is presented. Although there are occasional sharp drops, most data points are concentrated around -50, indicating that the agent's performance is stable and reliable. As training continues, the result distribution gradually stabilizes, and the frequency of extreme values ​​decreases significantly. This trend indicates that the agent's learning process has converged, enabling it to continuously optimize the control strategy and achieve robust control over time.

[0100] Step 5: Input the real-time vehicle driving status data into the TD3 agent after training convergence. The TD3 agent outputs the driving torque of the four drive motors to realize the coordinated control of differential power steering and stability of the distributed drive electric vehicle.

[0101] Specifically, the driving torque of the four drive motors output by the TD3 intelligent agent is respectively the driving torque of the left front wheel. Right front wheel drive torque Left rear wheel drive torque and right rear wheel drive torque That is, output action ,in, These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel of an electric vehicle, respectively.

[0102] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions from the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used in the operation of a steering and stability coordination control method for a distributed drive electric vehicle.

[0103] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the operating system of the terminal. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be Random Access Memory (RAM) or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the steering and stability coordination control method for a distributed drive electric vehicle in the above embodiments.

[0104] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0105] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0106] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] This invention also provides a computer program product for executing any of the above-described steering and stability coordination control methods for distributed drive electric vehicles. Since the computer program product provided by this invention belongs to the same inventive concept as the steering and stability coordination control method for distributed drive electric vehicles described above, it possesses all the advantages of the aforementioned method. Therefore, the beneficial effects of the computer program product provided by this invention will not be elaborated upon here.

[0109] In this invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0110] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention.

Claims

1. A method for steering and stability coordination control of a distributed drive electric vehicle, characterized by, include: An integrated model of distributed motor and vehicle dynamics is constructed. Vehicle driving state data is obtained through the integrated model. The driving state data includes the center of gravity sideslip angle, yaw rate, lateral acceleration, steering wheel angle, steering wheel angular velocity, and the torque applied to the steering wheel by the driver. Based on phase plane analysis, a correlation function is designed according to the centroid sideslip angle and the rate of change of the centroid sideslip angle. The correlation function is used to quantitatively evaluate the current yaw stability of the vehicle in real time. The yaw stability includes stable state, transient state and unstable state. A preset reference model is used to obtain the yaw rate reference value and the center of gravity sideslip angle reference value. Combined with the correlation function and driving status data packet, the state space and reward function of the TD3 agent are designed. A random target velocity sequence is generated through a Gaussian process. The target velocity sequence is input into the ensemble model, a complete vehicle driving state dataset is collected, and the TD3 agent is trained using the dataset until the TD3 agent converges. The real-time vehicle driving status data is input into the TD3 agent after training and convergence. The TD3 agent outputs the driving torque of the four drive motors to achieve coordinated control of differential power steering and stability of the distributed drive electric vehicle.

2. The steering and stability coordination control method for a distributed drive electric vehicle according to claim 1, characterized by, The specific process for constructing the integrated model of the distributed motor model and the vehicle dynamics model is as follows: A basic model of a four-wheel independent drive vehicle is established in CarSim. The original transmission system is removed, and the motor torque response is approximated using a second-order transfer function. The second-order transfer function is as follows: wherein is a second order transfer function, is the actual electromagnetic torque, is the target electromagnetic torque, is the number of pole pairs, is the rotor flux, is the stator resistance, is the mutual inductance, is the self-inductance; is the transfer function of the voltage pulse width modulation circuit, is a time constant, is a Laplace complex frequency domain variable, represents a motor characteristic parameter; A four-degree-of-freedom steering system model was built in Simulink. The four-degree-of-freedom steering system model was then integrated with the basic four-wheel independent drive vehicle model in CarSim to obtain a distributed motor model and a vehicle dynamics integrated model.

3. The method for coordination control of steering and stability of a distributed drive electric vehicle according to claim 1, wherein The process of designing the correlation function includes: (1) A phase plane is constructed based on the centroid sideslip angle and the rate of change of the centroid sideslip angle. The phase plane is divided into a stable region and a critical region by an elliptical boundary. The classical domain expression is defined as follows: wherein is a weight coefficient, is a weight coefficient, is a weight coefficient, is determined by vehicle parameters; is a slope of the boundary line, and an intercept is ; is a centroid lateral angle of the classical domain boundary; is a centroid lateral angle change rate of the classical domain boundary; , , is a centroid lateral angle of the positive domain boundary; is a centroid lateral angle change rate of the positive domain boundary; (2) introducing a characteristic parameter representing a driving state of the vehicle simplifying the two-dimensional domain into a one-dimensional domain, and the expression of the characteristic parameter is: wherein is a characteristic parameter, is a vehicle's mass center side slip angle, is a vehicle's mass center side slip angle rate of change, is a weight factor; (3) The extended distance is used to describe the positional relationship between the feature parameters and the classical domain. The extended distance from any feature parameter to the classical domain is: In the formula, denotes the extension distance of the characteristic parameter relative to the classical domain interval, is the classical domain interval; Any feature parameter to positive domain The extended distance is: wherein represents the extension distance of the characteristic parameter relative to the positive domain interval, is the positive domain interval; (4) Define the correlation function as: wherein ; If then the vehicle is in a stable state; If then the vehicle is in a transition state from stable to unstable; If then the vehicle is in an unstable state; wherein is a correlation function, is an extended distance of the characteristic parameter relative to the classical domain interval, is a difference of the inter-domain extended distance.

4. The steering and stability coordination control method for a distributed drive electric vehicle according to claim 1, characterized by, The preset reference model is a linear two-degree-of-freedom model, and the calculation formula of the yaw angular velocity reference value and the reference value of the side slip angle of the center of mass are respectively wherein is the critical centric side slip angle, is the acceleration of gravity, is the vehicle mass, is the longitudinal velocity, is the road adhesion coefficient, and are the distances from the center of mass to the front and rear axles, respectively, is the vehicle wheelbase, is the rear tire side slip stiffness, is the stability coefficient, is the front wheel steering angle.

5. The method for coordination control of steering and stability of a distributed drive electric vehicle according to claim 1, wherein The state space of the TD3 agent Is: wherein denotes the state space, denotes the yaw rate difference, wherein , is the yaw rate, is the yaw rate reference value, denotes the center of mass side slip angle difference, wherein , is the center of mass side slip angle of the vehicle, is the center of mass side slip angle reference value, denotes the steering wheel angle; denotes the steering wheel angular velocity; denotes the torque exerted by the driver on the steering wheel; denotes the lateral acceleration.

6. The steering and stability coordination control method for a distributed drive electric vehicle according to claim 1, characterized by, The reward function of the TD3 agent The comprehensive expression is: in, wherein is a reward function, is a correlation function, is a characteristic parameter, is a first reward subterm, is a second reward subterm, is a third reward subterm, is a fourth reward subterm, is a yaw rate, is a yaw rate reference value, is a vehicle's center of mass side slip angle, is a center of mass side slip angle reference value, is a steering wheel angle, is a steering wheel angular velocity, is a steering wheel angle reference value, denotes a torque applied by the driver to the steering wheel.

7. The method for coordination control of steering and stability of a distributed drive electric vehicle according to claim 1, wherein The Gaussian process uses a radial basis function kernel as the covariance function, and the expression for the radial basis function kernel is: where, denotes the covariance function at any two time instances, denotes a time input point, denotes another time input point, is the signal variance, is the length scale, is the smoothness for controlling the speed variation.

8. The method for coordination control of steering and stability of a distributed drive electric vehicle according to claim 1, wherein The driving torques of the four driving motors output by the TD3 intelligent agent are respectively left front wheel driving torque , right front wheel driving torque , left rear wheel driving torque , and right rear wheel driving torque , that is, output action .

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a steering and stability coordination control method for a distributed drive electric vehicle as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1-9. When the computer program is executed by the processor, it implements a steering and stability coordination control method for a distributed drive electric vehicle as described in any one of claims 1 to 8.

Images