Driving anti-slip and torque vectoring coordination control method based on complex road condition recognition

Abstract

The application belongs to the field of intelligent control method of automobile chassis, and relates to a driving anti-skid and torque vector coordination control method based on complex road condition recognition, which comprises complex road surface recognition and adhesion level switching, driving anti-skid intervention criterion and target wheel speed design, torque vector hierarchical control method, driving anti-skid and torque vector coordination control target fusion design based on stability index, and incremental PID driving and braking controller collaborative design. The application can realize hierarchical recognition of uniform road adhesion level and accurate discrimination of butt joint and split road surfaces, integrate complex road surface recognition, intervention criterion design and slip rate target decision into a human-vehicle-road comprehensive system, realize accurate control of different adhesion conditions, respectively construct driving anti-skid and torque vector hierarchical control architecture, design dynamic fusion of driving anti-skid and torque vector control targets, build incremental PID driving and braking collaborative controller, and give consideration to vehicle traction performance and handling stability.

Description

Technical Field

[0001] This invention pertains to intelligent control methods for new energy vehicle chassis, specifically a method for coordinated control of electric vehicle drive anti-slip and torque vector based on complex road condition recognition. Background Technology

[0002] The basic principle of traction control is to monitor wheel speed in real time, identify wheel slip state, and control the wheel slip ratio near the optimal slip ratio based on the road surface adhesion coefficient by adjusting the motor output torque, applying appropriate braking torque, or a combination of both, to suppress excessive wheel slip, thereby improving vehicle traction and ensuring driving stability. However, due to the current state of research on road surface adhesion estimation technology and economic limitations, real-time dynamic high-precision updating of road surface adhesion information remains extremely challenging in practical vehicle applications, making it difficult to accurately determine wheel slip and set the target slip ratio.

[0003] In actual driving, vehicle operation is often not limited to straight-line travel. When a vehicle simultaneously performs steering and acceleration operations on a low-traction surface, the driver's steering input will trigger a corresponding yaw torque demand. However, due to limitations in road surface adhesion, the wheels may experience excessive slippage, leading to nonlinear responses such as understeer or oversteer. Therefore, how to meet the driver's yaw demand while keeping the wheels within the optimal slip ratio range, achieving coordinated control of drive anti-slip and torque vector, and simultaneously satisfying the vehicle's dynamic performance and handling stability requirements, is a problem that urgently needs to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for coordinated control of drive anti-skid and torque vector based on complex road condition identification, comprising the following steps:

[0005] Step 1: Construct an online estimation of single-wheel adhesion coefficient based on unscented particle filtering, and combine it with the slip ratio to update the threshold to suppress the jitter of adhesion coefficient estimation when the slip is low; and combine it with the identification of open / connected road surfaces and the switching of uniform road surface adhesion level finite state machine to form a reliable road surface type perception and level management.

[0006] Step 2: Establish a drive anti-skid control intervention / exit criterion that takes into account both driver intent and slip threshold. Calculate the optimal slip ratio according to road surface level to generate target wheel speeds for all four wheels, and set a low-speed threshold to adaptively adjust the target wheel speeds for stability guidance.

[0007] Step 3: A torque vector hierarchical control method is proposed. The upper layer calculates the ideal yaw target based on the steady-state vehicle dynamics model. The middle layer calculates the additional yaw moment through a three-step method of steady-state compensation + feedforward + linear quadratic programming feedback with embedded linear quadratic programming. The lower layer outputs the target wheel speed for maneuver guidance through a multi-objective optimization allocation method.

[0008] Step 4: Define vehicle stability indices based on the front and rear wheel slip angle phase planes, and divide the safe zone, transition zone, and danger zone; use the vehicle stability indices as weights to integrate the target wheel speeds guided by handling and stability, and dynamically adjust the priority according to the stability margin to achieve the unified goal of drive anti-slip-torque vector (ASR-TVC) coordinated control.

[0009] Step 5: Design an incremental PID motor / brake co-controller for electric vehicles with independent front and rear axle drive. Utilize the joint control of motor torque and braking torque to update the control quantity for the deviation between the target wheel speed and the measured wheel speed, tracking a unified target to achieve the requirements of drive anti-slip and torque vector coordinated control.

[0010] Furthermore, step one includes the following steps:

[0011] First, unscented particle filtering is used to estimate the adhesion coefficient of each wheel based on the PAC2002 tire model in real time. A slip ratio threshold update method is introduced to update the adhesion estimate only when the absolute slip ratio exceeds the threshold. Second, a sliding window integral method based on slip ratio is used to identify split-road surfaces. By comparing the trends of the integral slip ratios of the left and right wheels, misjudgments caused by instantaneous fluctuations are suppressed, and a split-road surface identification state machine is formed by combining the braking torque threshold. For docked surfaces, an identification criterion based on the rate of change of wheel velocity angular acceleration and the trend of driving torque is designed to achieve timely switching of control targets. Finally, a uniform road surface adhesion level switching rule is proposed, dividing the road surface into three levels: low, medium, and high, and dynamically adjusting the target slip ratio according to the identification results.

[0012] Furthermore, the specific steps of step one are as follows:

[0013] First, a single-wheel adhesion coefficient estimation algorithm based on unscented particle filtering is used, followed by a four-wheel adhesion coefficient estimation algorithm. As state variables, longitudinal and lateral tire forces , As an observation, a state observation equation is established based on the PAC2002 tire model:

[0014] ;

[0015] in, and For tire model functions, For process noise, , , These are the longitudinal, lateral, and vertical tire forces, respectively. and These are tire slip rate and sideslip angle, respectively. These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively, with subscripts. For the update step, and The difference is one update interval; the formulas for tire slip ratio and sideslip angle are as follows:

[0016] , ;

[0017] in, and These are the longitudinal and lateral wheel speeds, respectively. The angular velocity of the wheel. The effective rolling radius of the wheel;

[0018] An update method based on a slip ratio threshold is adopted: only when the absolute value of the tire slip ratio is... Update the adhesion coefficient when the set threshold is exceeded; otherwise, retain the previous estimate.

[0019] Secondly, the implementation steps of the adhesion coefficient estimation algorithm based on unscented Kalman filtering (UKF) are as follows:

[0020] During the initialization phase, when At that time, the initial particle set is generated. And set the weight of each particle to be ;

[0021] Entering the recursive process, for Perform the following steps in sequence:

[0022] Step 1: Determine the current slip ratio Is it less than the slip ratio threshold? ,like If the condition is met, proceed directly to Step 8 for termination condition check; otherwise, continue to Step 2.

[0023] Step 2, for each particle UKF is used for prediction and updating to obtain particle state estimates. and its covariance ;

[0024] Step 3, combining observations Calculate particle weights ;

[0025] Step 4, normalize all particle weights as follows:

[0026] ;

[0027] Step 5, Calculate the number of effective particles :

[0028] ;

[0029] like If the value is below the set threshold, proceed to Step 6; otherwise, proceed to Step 7.

[0030] Step 6: Perform system resampling, weighted... Generate a new set of particles and reset the weights to ;

[0031] Step 7, Output the state estimate :

[0032] ;

[0033] Step 8, check the termination condition: if it is met, end the algorithm; otherwise, let... Then return to the slip ratio determination step and continue iterating until convergence.

[0034] Furthermore, the steps for the uniform road surface adhesion level switching rule and the non-uniform adhesion road surface identification method are as follows:

[0035] Split-road surface recognition method: Based on each axle, the road surface state of the left and right wheels of the front and rear axles is obtained separately; the sliding window integral method of slip ratio on both sides is used to determine the overall trend relationship. The core idea of ​​sliding window integral is: slide a fixed-size time window or data segment on the data sequence, calculate the integral of the data within the window, and update it continuously; when the difference between the sliding window integrals of the slip ratio of the left and right wheels is greater than the threshold value, it is considered that the split-road surface has been entered; at the same time, the braking torque on the low-adhesion side needs to be less than the standard braking torque; combined with the finite state machine method, the split-road surface recognition is realized.

[0036] Road surface identification method:

[0037] The criteria for determining whether a vehicle with a low coefficient of adhesion is entering a road surface with a high coefficient of adhesion are set as follows:

[0038] A. Low-adhesion-coefficient road surfaces are only identified when anti-skid control is engaged.

[0039] B. Within the calibrated time range, the actual torque of the motor remains unchanged;

[0040] C. Within the calibrated time range, the wheel angular acceleration continuously decreases to the preset threshold value.

[0041] Finally, a uniform road surface adhesion level switching rule is constructed, dividing the road surface into three levels: low, medium, and high, and calibrating the optimal slip ratio for each level. The adhesion level is dynamically switched based on the real-time estimated road surface adhesion coefficient and a threshold value. When a low-adhesion-coefficient connecting road surface or a split road surface with a high-adhesion-coefficient side is detected, if the road surface adhesion coefficient estimation stops updating, the adhesion level is increased to enhance the driving anti-skid control dynamics when the adhesion estimation response is slow under non-full acceleration or braking. The adhesion level switching is implemented through a finite state machine, and the switching rule is designed as follows:

[0042] Set the initial adhesion level to: High adhesion;

[0043] A. When the current state is high adhesion: If the road adhesion coefficient is dynamically updated If the road surface is not directly connected, switch to medium adhesion; otherwise, maintain high adhesion.

[0044] B. When the current state is medium adhesion: if the road adhesion coefficient is dynamically updated If the road surface is not directly connected, switch to low adhesion; if any of the following conditions are met, switch to high adhesion: Either the road surface transitions from low-adhesion to high-adhesion and the high-adhesion side is not updated, or the road surface transitions from uniform to split and the high-adhesion side is not updated; otherwise, maintain the medium-adhesion state.

[0045] C. When the current state is low adhesion: If the road adhesion coefficient is dynamically updated If the road surface transitions from low-adhesion to high-adhesion and the high-adhesion side has not been updated, switch to medium-adhesion; if the road surface transitions from uniform to split and the high-adhesion side has not been updated, switch to high-adhesion; otherwise, maintain low-adhesion.

[0046] Furthermore, step two designs a drive anti-slip control intervention criterion and slip ratio control target that consider the integrated human-vehicle-road system. The intervention criterion determines the intervention and withdrawal of drive anti-slip control by comprehensively considering the driver's acceleration demand and slip ratio threshold. When the accelerator pedal signal increases significantly and the equivalent wheel speed exceeds the calibrated value, motor torque adjustment is activated. When the braking or command torque deviation is too large, it is disengaged to avoid interference with drivability. The slip ratio target design guided by drive anti-slip combines the calibrated optimal single-wheel slip ratio under different road surface adhesion coefficients. It switches the adhesion level and generates the target wheel speed through a finite state machine. A minimum target wheel speed limit is applied in the low-speed zone to prevent starting difficulties. The specific steps are as follows:

[0047] First, for the motor torque and braking torque of the drive anti-slip control system, respectively, we design a drive anti-slip coordinated control intervention and withdrawal rule that combines the driver's acceleration intention and the preset threshold of the vehicle slip rate;

[0048] The motor torque intervention rule setting for the anti-slip control system includes the following two aspects:

[0049] A. The traction control system only needs to intervene when the accelerator pedal analysis indicates that the driver's acceleration intention is increasing, that is, the driver expects an increase in vehicle power output, which may cause the wheels to slip.

[0050] B. First, obtain the slip ratio intervention threshold through actual vehicle calibration. When the real-time wheel speed exceeds the threshold value, it indicates that the wheel has slipped significantly, and the drive anti-slip control needs to intervene and control the equivalent wheel speed at the target value.

[0051] The motor torque withdrawal rule setting for the anti-slip control system includes the following two aspects:

[0052] A. When the driver presses the brake pedal, it indicates that the vehicle needs to slow down or stop, and the traction control system should disengage in time.

[0053] B. When the driving anti-slip control system's adjustment torque Exceeding the driver's expected torque Achieving a safety margin At that time, that is If the power output may affect the driving experience, the driver should disengage and restore the power to the level desired by the driver.

[0054] Meanwhile, in order to achieve the coordinated control requirements of different slip ratios of the left and right wheels on the same axle under split-plane road conditions to enhance vehicle dynamics and handling stability, the intervention conditions for braking torque control are designed as follows:

[0055] A. When the road surface is identified as a split road surface, control the braking torque on the side with the low coefficient of adhesion to reduce wheel slippage on the side with the low coefficient of adhesion.

[0056] B. The anti-slip control system can only adjust the braking torque when the motor torque control is activated;

[0057] The braking torque withdrawal conditions are: the road surface is uniform, the decision braking torque is less than the calibrated value, or the driver depresses the brake pedal.

[0058] Then, the calculation formula for the target wheel speed of the four wheels was derived, and the driving anti-skid control target was determined based on a composite decision of road conditions and vehicle status. The adhesion level switching was realized through a finite state machine, and the optimal slip ratio for low, medium, and high adhesion surfaces was calibrated by combining real vehicle tests. In order to determine the optimal target speed for each wheel. subscript These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0059] Set low longitudinal speed threshold To avoid starting difficulties and low-speed oscillations due to excessively low target speeds during the low-speed phase; when the longitudinal speed... When the speed is below this threshold, the target wheel speed is calculated according to the threshold; the optimal target rotational speed... The calculation is as follows:

[0060] ;

[0061] Among them, process quantities .

[0062] Further, step three constructs a hierarchical control architecture suitable for distributed drive electric vehicles. The upper layer uses a steady-state vehicle dynamics model to calculate the yaw rate reference value based on the driver's steering angle input. The middle layer employs a three-step strategy to complete steady-state compensation, feedforward correction, and linear quadratic programming feedback optimization, minimizing the tracking error of the yaw rate reference value. The lower layer establishes a multi-objective optimization model to distribute the additional yaw moment calculated by the upper layer to the four-wheel drive motors, taking into account both longitudinal driving force requirements and tire adhesion and motor constraints. The specific steps are as follows:

[0063] First, the upper layer of the hierarchical control architecture is the target of ideal yaw rate control. The design is based on a linear two-degree-of-freedom vehicle dynamics model that includes lateral and yaw motions.

[0064] ;

[0065] in, For vehicle quality, , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. This refers to the vehicle's wheelbase. , , These are the equivalent lateral stiffness of the front and rear axles, respectively. Front wheel steering angle, understeer degree ;

[0066] ;

[0067] Among them, steady-state yaw rate gain , The road surface adhesion coefficient, It is the acceleration due to gravity;

[0068] Then, the intermediate layer control of the hierarchical control architecture adopts a three-step strategy: the first step is to solve the steady-state control quantity to compensate for the static deviation; the second step is to introduce reference feedforward control to improve the dynamic response; and the third step is to design state feedback control based on linear quadratic programming to minimize the yaw rate and centroid side slip angle error, while also taking into account the optimal control energy.

[0069] In the intermediate layer controller design, a nonlinear two-degree-of-freedom vehicle model is adopted, combined with a tire model to describe the nonlinear characteristics of the tire; the lateral dynamics equation of the vehicle with applied additional yaw moment control is expressed as:

[0070] ;

[0071] in, Let be the lateral speed of the vehicle. and These are the lateral forces on the front and rear axles, respectively. It is the moment of inertia of rotation about the z-axis. The yaw rate is angular velocity. To add yaw moment;

[0072] The first step of the three-step method, which involves solving for the steady-state control quantity to compensate for the static deviation, is as follows:

[0073] make Steady-state feedforward control quantity The following relationship must be satisfied:

[0074] ;

[0075] Determine the desired steady-state feedforward control input:

[0076] ;

[0077] The second step of the three-step method, introducing dynamic feedforward control to improve dynamic response, involves the following steps:

[0078] Based on steady-state control, consider the system's ideal yaw rate control objective. The dynamic changes are introduced to generate dynamic feedforward control. The control input is expanded to:

[0079] ;

[0080] Substituting this into the vehicle lateral dynamics equations, we get:

[0081] ;

[0082] make The dynamic feedforward control quantity is obtained as follows:

[0083] ;

[0084] The third step of the three-step method, which involves designing state feedback control based on linear quadratic programming to minimize yaw rate and centroid sideslip angle errors, follows these steps:

[0085] Feedback control technology based on linear quadratic programming is integrated into a three-step framework, using lateral dynamics without additional yaw moment as the reference model, and the ideal centroid sideslip angle is used. and ideal yaw rate control target Substitute; subtract the reference model from the controlled system, at which point the state vector... Represented as: β is the centroid sideslip angle; a linear quadratic programming algorithm is used to determine the feedback control quantity. To minimize state tracking error; and simultaneously construct an infinite time-domain cost function. Taking into account both the target state tracking error in maneuver stability and the execution energy loss in torque vector control:

[0086] ;

[0087] Where Q is a target state tracking weight matrix, and the execution energy loss weight matrix R=1; Weights in Q to control time and The tracking errors corresponding to the sideslip angle and yaw rate are designed as stability evaluation indicators. Functions:

[0088] ;

[0089] Where parameters =0.001, =0.5, =0.5, =0.8;

[0090] The final additional yaw moment is:

[0091] ;

[0092] in, and These are the steady-state and dynamic feedforward control quantities based on the three-step method, respectively. This is a feedback control variable based on linear quadratic programming.

[0093] Finally, the lower layer establishes a constraint optimization model with the goal of minimizing the additional yaw moment distribution error and reducing tire load utilization. The additional yaw moment control amount calculated by the middle layer is distributed to the four wheels, and motor saturation and tire adhesion constraints are taken into account.

[0094] The system dynamic equations are expressed as follows:

[0095] ;

[0096] in, , These are the front and rear axle track widths, respectively. , , , These represent the longitudinal tire forces of the left front, right front, left rear, and right rear wheels, respectively; the control input u is defined as:

[0097] ;

[0098] The problem of optimizing the allocation of lower-level drive torque is expressed as:

[0099] ;

[0100] in, The tire load utilization weighted matrix is ​​defined as follows: To prevent any single tire from saturating prematurely, For vertical loads; This is the error weighting matrix, representing the control priority between the longitudinal driving force and the additional yaw moment; and Representing the actual and expected generalized force vectors respectively, the total driving force is based on the moment vector. and additional yaw moment composition; These are control parameters.

[0101] Considering actuator and road adhesion constraints, the longitudinal force of each tire must satisfy:

[0102] ;

[0103] in, The maximum torque of each motor, This is the effective rolling radius of the wheel.

[0104] Furthermore, in step four, to simultaneously meet the stability requirements of drive anti-slip and the maneuverability requirements of torque vector, a method based on the front and rear wheel slip angles is proposed. Vehicle stability evaluation index in phase plane Quantitatively characterize the stability margin; calculate the vehicle state in real time. The position in the phase plane is dynamically divided into safe zones, transition zones, and danger zones, and the target wheel speed is integrated with maneuverability and stability guidance based on the region. and When the vehicle is in the safe zone, the control objective based on torque vector is dominant; when in the danger zone, the stability objective based on drive anti-slip is dominant; in the transition zone, the objective is continuously weighted and fused according to indicators, and the fused ideal wheel speed control objective expression is obtained. To achieve the adaptive superposition of two objectives, avoid abrupt control switching, and improve vehicle stability under complex operating conditions through drive anti-slip control, while simultaneously enhancing vehicle handling performance using torque vector control. The specific steps are as follows:

[0105] Based on the vehicle's condition, the front and rear wheel slip angles Position in the phase plane Design vehicle stability evaluation indicators:

[0106] ;

[0107] According to vehicle dynamics characteristics, the saturation sideslip angle of the front wheels is generally greater than that of the rear wheels.

[0108] Centered on the origin, the rear wheel saturation sideslip angle Draw a circle with a radius to obtain the green area representing the safe zone. , As the vehicle's state gradually moves away from its origin, the stability evaluation indicators... The value gradually increases if the location is within the safe zone. If the boundary is defined, then the stability evaluation index is taken as: ,when When the condition is met, the vehicle is stable under the current operating conditions.

[0109] When the vehicle's front wheels saturate the sideslip angle Draw a circle with radius to obtain the blue area representing the transition zone. , , ~ Corresponding to the transition zone from Boundary to The stability evaluation index of the boundary is determined by the continuous increase in the utilization rate of vehicle power capacity under the current working conditions.

[0110] With two saddle points and A red area representing the danger zone is obtained by drawing a circle with half the distance between them as the radius. , , ~ Corresponding to the danger zone from Boundary to The stability evaluation index at the boundary is set when the vehicle is unstable; when the vehicle is located in a danger zone. In addition, stability evaluation indicators Restricted to .

[0111] Then, the dynamic requirements for wheel speed control are adjusted according to the changes in the vehicle's stability zone; each stability zone represents a different vehicle stability state, which is used as a condition for dynamically switching control requirements in the fusion control objective; combined with vehicle stability evaluation indicators. The system quantitatively identifies the vehicle's position and stability, and reflects the priority of the composite control objective by dynamically adjusting the weighting coefficients of the fused wheel speed target. The relationship between the vehicle's stable state and the dynamic control requirements of the composite wheel speed target is described as follows:

[0112] When the vehicle is in the safe zone R1, the steerable wheel speed control objective is the priority control objective; when the vehicle is in the transition zone R2, the vehicle stability evaluation index... As the stability index increases, the priority of the stability wheel speed control target increases, while the priority of the maneuverability wheel speed control target decreases. At the same time, the weight of the maneuverability wheel speed control target still exceeds that of the stability wheel speed control target, and it is dominant. When the vehicle is in the danger zone R3, the vehicle stability evaluation index is larger, and the stability wheel speed control target is given priority.

[0113] The dynamic changes in vehicle stability evaluation indicators enable adaptive adjustment of control objective priorities to meet dynamic control requirements throughout the entire control process.

[0114] ;

[0115] in, The steering wheel speed control target is determined by torque vector control. The stability wheel speed control target determined by the anti-skid control mechanism. The target fusion coefficient is for wheel speed control. The goal is to achieve a fusion of wheel speed control that is suitable for drive anti-slip and torque vector coordinated control.

[0116] Furthermore, step five involves obtaining the driver's desired total driving torque based on the pedal analytical model. The front and rear axle torques are distributed, and the target driving torque is obtained through first-order filtering. The main function of the accelerator pedal analytical model is to generate the driver's desired front and rear axle motor torques based on the pedal opening. The final target torque is obtained through front and rear axle torque distribution combined with first-order filtering and motor limiting. The anti-slip control system then controls the vehicle based on this final target torque. Its expression in the S-domain is:

[0117] ;

[0118] in, The time constant of a first-order inertial element. , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively.

[0119] After ASR-TVC intervention, the incremental PID motor torque control module is activated. It calculates the motor torque increment based on the larger value of the fused wheel speed control target between the left and right wheels on the same axle and the deviation of the actual wheel speed. The torque arbitration module then adds this increment to the driver's torque to generate the current front axle motor torque request value. Subsequent control is based on the previous request value and continuously iterates and updates to achieve drive anti-slip motor torque control.

[0120] The formula for incremental PID motor torque control is:

[0121]

[0122] in, To control the increment; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; This represents the deviation between the target and the output at the current moment. This represents the deviation between the target and the output in the previous long-term step. This represents the deviation between the target and the output at the first two time steps.

[0123] On split-road surfaces, the wheels on the low-adhesion side are prone to slippage due to the limited road surface adhesion conditions, influenced by the dynamic performance of the high-adhesion side. When ASR-TVC intervenes and the operating condition is determined to be split-road surface, the braking torque control state is activated, applying braking torque to the low-adhesion side to suppress slippage and release greater driving force. An incremental PID braking torque control module is built using the same incremental PID control algorithm as the motor torque control. The incremental braking torque on the low-adhesion side is calculated based on the measured wheel speed difference and the target wheel speed difference, and then superimposed with the braking torque decided at the previous moment to achieve split-road surface drive anti-slip control in coordination with motor torque.

[0124] The beneficial effects of this invention are:

[0125] (1) This invention proposes a complex road surface identification method based on finite state machine, which can realize dynamic hierarchical identification of uniform road surface adhesion level and accurate identification of docking and split road surfaces, reduce the dependence of drive anti-skid control on high dynamic and high precision road surface adhesion coefficient estimation, and improve the robustness and real-time performance of the control system under complex road conditions.

[0126] (2) This invention proposes a layered drive anti-skid control method, which integrates complex road surface identification, intervention criterion design, and slip ratio target decision into the human-vehicle-road integrated system. By comprehensively considering driver input, road surface information and vehicle status, the timing of drive anti-skid intervention and control target are adaptively determined, so as to achieve precise control of different adhesion conditions. It has good engineering feasibility and is suitable for mass-produced distributed drive electric vehicles.

[0127] (3) This invention proposes a method for coordinated control of driving anti-slip and torque vector, constructs a hierarchical torque vector control architecture to calculate the target slip ratio of the four wheels under yaw control, and realizes the dynamic fusion of driving anti-slip target and torque vector target based on vehicle stability evaluation index. By designing an incremental PID controller, the coordinated adjustment of driving anti-slip and torque vector is realized, taking into account both vehicle traction performance and lateral handling stability. Attached Figure Description

[0128] Figure 1 This is a schematic diagram of the overall flow of the control method of the present invention;

[0129] Figure 2 This is a schematic diagram of the finite state machine for the anti-slip control intervention criteria of this invention;

[0130] Figure 3 This is a schematic diagram of the torque vector hierarchical control architecture of the present invention;

[0131] Figure 4 This is a schematic diagram illustrating the stability index design and adaptive weight adjustment of the present invention. Detailed Implementation

[0132] like Figure 1 As shown, this invention provides a method for coordinated control of drive anti-skid and torque vector based on complex road condition identification, comprising the following steps:

[0133] Step 1: Complex Road Surface Identification and Adhesion Level Switching Method

[0134] A single-wheel adhesion coefficient online estimation system based on unscented particle filtering is constructed, and the adhesion coefficient estimation jitter is suppressed when the slip ratio is updated by the threshold. Furthermore, the system is combined with the identification of open / connected road surfaces and the switching of uniform road surface adhesion level finite state machine to form a reliable road surface type perception and level management.

[0135] Furthermore, firstly, unscented particle filtering is used to estimate the adhesion coefficients of each wheel based on the PAC2002 tire model in real time. To improve the stability of the estimation under low slip conditions, a slip ratio threshold update method is introduced, updating the adhesion estimate only when the absolute slip ratio exceeds the threshold, thus avoiding the impact of low-speed fluctuations on the target wheel speed calculation. Secondly, a sliding window integral method based on slip ratio is used to identify split-road surfaces. By comparing the trends of the integral slip ratios of the left and right wheels, misjudgments caused by instantaneous fluctuations are suppressed, and a split-road surface identification state machine is formed by combining the braking torque threshold. For docked surfaces, to address the problem of lag in adhesion estimate updates when transitioning from low to high adhesion coefficients, an identification criterion based on the rate of change of wheel speed angular acceleration and the trend of driving torque changes is designed to achieve timely switching of control targets. Finally, a uniform road surface adhesion level switching rule is proposed, dividing the road surface into three levels: low, medium, and high. The target slip ratio is dynamically adjusted according to the identification results to ensure the robustness and response speed of the drive anti-slip control, and improve the traction and stability of the vehicle under complex road conditions. The specific steps are as follows:

[0136] To improve the accuracy and robustness of adhesion coefficient estimation under low-slip conditions, a single-wheel adhesion coefficient estimation algorithm based on unscented particle filtering is proposed, and a slip ratio threshold update method is introduced to suppress estimation fluctuations under low-slip conditions; four-wheel adhesion coefficient... As state variables, longitudinal and lateral tire forces , As an observation, a state observation equation is established based on the PAC2002 tire model:

[0137] ;

[0138] in, and For tire model functions, For process noise, , , These are the longitudinal, lateral, and vertical tire forces, respectively. and These are tire slip rate and sideslip angle, respectively. These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. and The difference is one update interval; the formulas for tire slip ratio and sideslip angle are as follows:

[0139] , ;

[0140] in, and These are the longitudinal and lateral wheel speeds, respectively. The angular velocity of the wheel. The effective rolling radius of the wheel;

[0141] When the slip ratio is low, the sensitivity of tire force to the coefficient of adhesion decreases, and direct updates can easily cause significant fluctuations, affecting the stability of the target wheel speed. Therefore, an update method based on a slip ratio threshold is adopted: updates are only performed when... The adhesion coefficient is updated when the set threshold is exceeded; otherwise, the previous estimate is retained, thereby improving stability under low slip conditions.

[0142] The implementation steps of the adhesion coefficient estimation algorithm based on unscented Kalman filtering (UKF) are as follows:

[0143] During the initialization phase, when At that time, the initial particle set is generated. And set the weight of each particle to be ;

[0144] Entering the recursive process, for Perform the following steps in sequence:

[0145] Step 1: Determine the current slip ratio Is it less than the threshold slip ratio? ,like If the condition is met, proceed directly to Step 8 for termination condition check; otherwise, continue to Step 2.

[0146] Step 2, for each particle UKF is used for prediction and updating to obtain particle state estimates. and its covariance ;

[0147] Step 3, combining observations Calculate particle weights ;

[0148] Step 4, normalize all particle weights as follows:

[0149] ;

[0150] Step 5, Calculate the number of effective particles :

[0151] ;

[0152] like If the value is below the set threshold, proceed to Step 6; otherwise, proceed to Step 7.

[0153] Step 6: Perform system resampling, weighted... Generate a new set of particles and reset the weights to ;

[0154] Step 7, Output the state estimate:

[0155] ;

[0156] Step 8, check the termination condition: if it is met, end the algorithm; otherwise, let... Then return to the slip ratio determination step and continue iterating until convergence.

[0157] To address the accuracy issue in road surface adhesion coefficient estimation, based on single-wheel road surface adhesion coefficient estimation, a uniform road surface adhesion level switching rule and a non-uniform adhesion road surface (including split road surfaces and connecting road surfaces) identification method were designed, which effectively improved the accuracy and robustness of drive anti-skid control under complex road conditions.

[0158] First, the established method for identifying split-road surfaces is based on each axle, obtaining the road conditions of the left and right wheels on both the front and rear axles. Taking a low-friction left and high-friction right road surface as an example, the left wheel speed will be higher than the right wheel speed, creating a wheel speed difference between the left and right sides of the axle. When the vehicle's anti-slip control system is not activated, the two sides will maintain a large wheel speed difference, which disappears when entering a uniform road surface. If the anti-slip control system is active, the wheel speed difference will fluctuate, leading to misjudgment when encountering split-road surfaces. Therefore, wheel speed difference cannot be used to directly determine whether a vehicle has entered a uniform road surface. A sliding window integral method based on the slip ratios of both sides is used to determine the overall trend of the slip ratio to solve the problem of misjudging instantaneous wheel speed differences. The core idea of ​​sliding window integral is to slide a fixed-size time window or data segment across the data sequence, calculate the integral of the data within the window, and update it continuously. When the vehicle first enters a low-friction left and high-friction right road surface, the overall trend of the left slip ratio is consistently greater than that of the right slip ratio. When entering a uniform road surface, the overall relationship between the slip ratios of both sides will decrease. Therefore, when the difference between the sliding window integrals of the slip rates of the left and right wheels exceeds a threshold value, it is considered that the road surface has been entered into a split-plane condition. Simultaneously, to prevent a large braking torque from reducing the slip rate difference and causing misjudgment of road surface adhesion, the braking torque on the side with lower adhesion must be less than the standard braking torque. Using the above ideas and combining them with a finite state machine approach, the identification of split-plane conditions is achieved.

[0159] Then, the connecting road surface refers to the section where the road surface adhesion coefficient of the coaxial wheels changes abruptly. When moving from a high-adhesion-coefficient road surface to a low-adhesion-coefficient road surface, the wheels are prone to slippage and can quickly reach the update threshold. However, when moving from a low-adhesion-coefficient road surface to a high-adhesion-coefficient road surface, when the drive anti-slip control intervenes, the low-adhesion-coefficient side limits the motor torque output. After the vehicle enters the high-adhesion-coefficient road surface, it may not be able to trigger the adhesion estimation update threshold, resulting in an update lag. Therefore, it is necessary to specifically identify low-adhesion-coefficient road surface transitions to high-adhesion-coefficient road surfaces so that the drive anti-slip control can be adjusted in a timely manner to adapt to the adhesion change.

[0160] On low-friction surfaces, the anti-slip control maintains the slip ratio near its optimal value, keeping the motor torque low and stable. When entering a high-friction surface with a constant or decreasing slip ratio, the longitudinal tire force gradually increases. According to wheel longitudinal dynamics, the wheel angular acceleration decreases compared to the low-friction surface. Therefore, the judgment rule for entering a high-friction surface from a low-friction surface is set as follows:

[0161] A. Low-adhesion-coefficient road surfaces are only identified when anti-skid control is engaged.

[0162] B. Within the calibrated time range, the actual torque of the motor remains unchanged;

[0163] C. Within the calibrated time range, the wheel angular acceleration continuously decreases to the preset threshold value.

[0164] The time range and wheel angular acceleration threshold values ​​were obtained through actual vehicle calibration tests.

[0165] Finally, a uniform road surface adhesion level switching rule is constructed, dividing the road surface into three levels: low, medium, and high, and calibrating the optimal slip ratio for each level. The adhesion level is dynamically switched based on the real-time estimated road surface adhesion coefficient and a threshold value. When a low-adhesion-coefficient connecting road surface to a high-adhesion-coefficient connecting road surface or a split road surface with a high-adhesion-coefficient side is detected, if the road surface adhesion coefficient estimation stops updating, the adhesion level is increased to enhance the driving performance of anti-skid control when the adhesion estimation response is slow under non-full acceleration or braking. The adhesion level switching is implemented through a finite state machine, and the switching rule design is shown in Table 1.

[0166] Table 1

[0167]

[0168] Step 2, Driving anti-skid intervention criteria and target wheel speed design method:

[0169] Establish a drive anti-skid control intervention / exit criterion that takes into account both driver intent and slip threshold, calibrate the optimal slip ratio according to road surface level to generate target wheel speeds for four wheels, and set a low-speed threshold to adaptively adjust the target wheel speeds guided by stability.

[0170] Furthermore, such as Figure 2As shown, step two designs a drive anti-slip control intervention criterion and slip ratio control target that consider the integrated human-vehicle-road system. The intervention criterion determines the intervention and withdrawal of drive anti-slip control by comprehensively considering the driver's acceleration demand and slip ratio threshold. When the accelerator pedal signal increases significantly and the equivalent wheel speed exceeds the calibrated value, motor torque adjustment is activated. When the braking or command torque deviation is too large, it is disengaged to avoid interference with drivability. The slip ratio target design guided by drive anti-slip combines the calibrated optimal single-wheel slip ratio under different road surface adhesion coefficients. It switches the adhesion level and generates the target wheel speed through a finite state machine. A minimum target wheel speed limit is applied in the low-speed zone to prevent starting difficulties. The specific steps are as follows:

[0171] First, for the motor torque and braking torque of the drive anti-slip control system, respectively, we design a drive anti-slip coordinated control intervention and withdrawal rule that combines the driver's acceleration intention and the preset threshold of the vehicle slip rate;

[0172] The motor torque intervention rule setting for the anti-slip control system includes the following two aspects:

[0173] A. The traction control system only needs to intervene when the accelerator pedal analysis (APO) indicates that the driver's acceleration intention is increasing, that is, he expects an increase in vehicle power output, which may cause the wheels to slip.

[0174] B. First, obtain the slip ratio intervention threshold through actual vehicle calibration. When the real-time wheel speed When the threshold is exceeded, that is This indicates that the wheel has slipped significantly, and the drive anti-slip control needs to intervene and control the equivalent wheel speed to the target value.

[0175] The motor torque withdrawal rule setting for the anti-slip control system includes the following two aspects:

[0176] A. When the driver presses the brake pedal, it indicates that the vehicle needs to slow down or stop, and the traction control system should disengage in time.

[0177] B. When the driving anti-slip control system's adjustment torque Exceeding the driver's expected torque Achieving a safety margin At that time, that is If the power level may affect the driving experience, the system should be disengaged and the power level restored to the driver's desired level.

[0178] Meanwhile, to meet the requirement of differentiating the slip ratios of the left and right wheels on the same axle under split-plane road conditions to enhance vehicle dynamics and stability, the intervention conditions for braking torque control are designed as follows:

[0179] A. When the road surface is identified as a split road surface, the braking torque on the side with the low coefficient of friction is controlled to reduce wheel slippage on that side. Braking torque control intervenes at the same time;

[0180] B. The anti-slip control system can only adjust the braking torque when the motor torque control is activated.

[0181] The braking torque withdrawal conditions are: the road surface is uniform and the decision braking torque is less than the rated value. Or the driver presses the brake pedal.

[0182] Then, for the special configuration of distributed drive electric vehicles, the calculation formula for the target wheel speed of the four wheels was derived, and the driving anti-skid control target was determined based on the combined decision of road conditions and vehicle state.

[0183] Theoretically, neglecting load transfer, the optimal slip ratio is determined solely by the road surface adhesion coefficient. Drive anti-skid control aims to maintain the wheel slip ratio at its optimal value, thus balancing longitudinal and lateral adhesion utilization. Adhesion level switching is achieved through a finite state machine, and the optimal slip ratio for low, medium, and high adhesion road surfaces is calibrated using real-vehicle testing. To determine the optimal speed of each wheel. subscript These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0184] To avoid starting difficulties and low-speed oscillations due to excessively low target speeds, a low longitudinal speed threshold is set. When longitudinal speed When the speed is below this threshold, the target wheel speed will be calculated according to the threshold value.

[0185] , ;

[0186] Step 3, Design of Torque Vector Layered Control Method:

[0187] A torque vector hierarchical control method is proposed. The upper layer calculates the ideal yaw target based on a steady-state vehicle dynamics model. The middle layer calculates the additional yaw moment using a three-step method of steady-state compensation + feedforward + linear quadratic programming feedback with embedded linear quadratic programming. The lower layer outputs the target wheel speed for maneuver guidance through a multi-objective optimization allocation method, achieving high-precision tracking of the ideal yaw target. Figure 3 As shown;

[0188] Further, step three constructs a hierarchical control architecture suitable for distributed drive electric vehicles. The upper layer uses a steady-state vehicle dynamics model to calculate the yaw rate reference value based on the driver's steering angle input. The middle layer employs a three-step strategy to complete steady-state compensation, feedforward correction, and linear quadratic programming feedback optimization, minimizing the tracking error of the yaw rate reference value. The lower layer establishes a multi-objective optimization model to distribute the additional yaw torque calculated by the upper layer to the four-wheel drive motors, taking into account both motor saturation and tire adhesion constraints. The specific steps are as follows:

[0189] First, the upper layer is the target for ideal yaw rate control. The design is based on a linear two-degree-of-freedom vehicle dynamics model that includes lateral and yaw motions.

[0190] ;

[0191] in, For vehicle quality, , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. This refers to the vehicle's wheelbase. , , These are the equivalent lateral stiffness of the front and rear axles, respectively. Front wheel steering angle, understeer degree ;

[0192] ;

[0193] Among them, steady-state yaw rate gain , The road surface adhesion coefficient, This is the acceleration due to gravity.

[0194] Then, the intermediate layer control adopts a three-step strategy: the first step is to solve the steady-state control quantity to compensate for the static deviation; the second step is to introduce reference feedforward control to improve the dynamic response; and the third step is to design state feedback control based on linear quadratic programming to minimize the yaw rate and centroid sideslip angle error, while also taking into account the optimal control energy.

[0195] In the intermediate layer controller design, a nonlinear two-degree-of-freedom vehicle model is adopted, combined with a tire model to describe the nonlinear characteristics of the tire; the lateral dynamics equation of the vehicle with applied additional yaw moment control is expressed as:

[0196] ;

[0197] in, For vehicle quality, Let be the longitudinal speed of the vehicle. Let be the lateral speed of the vehicle. and These are the lateral forces on the front and rear axles, respectively. It is the moment of inertia of rotation about the z-axis. The yaw rate is angular velocity. , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. This refers to the front wheel steering angle. To add yaw moment.

[0198] The first step of the three-step method, which involves solving for the steady-state control quantity to compensate for the static deviation, is as follows:

[0199] make Steady-state feedforward control quantity The following relationship must be satisfied:

[0200] ;

[0201] Determine the desired steady-state feedforward control input:

[0202] ;

[0203] The second step of the three-step method, introducing dynamic feedforward control to improve dynamic response, involves the following steps:

[0204] Based on steady-state control, consider the system's ideal yaw rate control objective. The dynamic changes are introduced to generate dynamic feedforward control. The control input is expanded to:

[0205] ;

[0206] Substituting this into the vehicle lateral dynamics equations, we get:

[0207] ;

[0208] make The dynamic feedforward control quantity is obtained as follows:

[0209] ;

[0210] The third step of the three-step method, which involves designing state feedback control based on linear quadratic programming to minimize yaw rate and centroid sideslip angle errors, follows these steps:

[0211] Feedback control techniques based on linear quadratic programming are integrated into a three-step framework to enhance the optimal control performance and parameter robustness of dynamic systems. Lateral dynamics without additional yaw moment are used as the reference model, and an ideal centroid sideslip angle is assumed. and ideal yaw rate control target Substitute; subtract the reference model from the controlled system, at which point the state vector... Represented as: The feedback control quantity is determined using a linear quadratic programming algorithm. To minimize state tracking error; and simultaneously construct an infinite time-domain cost function. Taking into account both the target state tracking error in maneuver stability and the execution energy loss in torque vector control:

[0212] ;

[0213] Where Q is a target state tracking weight matrix, and the execution energy loss weight matrix R=1; Weights in Q to control time and The tracking errors corresponding to the sideslip angle and yaw rate are designed as stability evaluation indicators. Functions:

[0214] ;

[0215] Where parameters =0.001, =0.5, =0.5, =0.8;

[0216] The weight adjustments in the cost function align with the integrated design of the composite yaw control objective under vehicle dynamic control requirements. The fused yaw objective, by balancing maneuverability and stability, eliminates conflicts between control objectives for different vehicle states, ensuring full utilization of the torque vector. The final additional yaw torque is merged as follows:

[0217] ;

[0218] in, and These are the steady-state and dynamic feedforward control quantities based on the three-step method, respectively. This is a feedback control variable based on linear quadratic programming.

[0219] Finally, the lower layer establishes a constraint optimization model with the goal of minimizing the additional yaw moment distribution error and reducing tire load utilization. The additional yaw moment control amount calculated by the middle layer is distributed to the four wheels, and motor saturation and tire adhesion constraints are taken into account.

[0220] The system dynamic equations are expressed as follows:

[0221] ;

[0222] in, , These are the front and rear axle track widths, respectively. , , , These represent the longitudinal tire forces of the left front, right front, left rear, and right rear wheels, respectively; the control input u is defined as:

[0223] ;

[0224] The problem of optimizing the allocation of lower-level drive torque is expressed as:

[0225] ;

[0226] in, The tire load utilization weighted matrix is ​​defined as follows: To prevent any single tire from saturating prematurely, The road surface adhesion coefficient, For vertical loads; This is the error weighting matrix, representing the control priority between the longitudinal driving force and the additional yaw moment; and Representing the actual and expected generalized force vectors respectively, the total driving force is based on the moment vector. and additional yaw moment composition; These are control parameters.

[0227] Considering actuator and road adhesion constraints, the longitudinal force of each tire must satisfy:

[0228] ;

[0229] in, The maximum torque of each motor, This is the effective rolling radius of the wheel.

[0230] The longitudinal slip ratio corresponding to the target longitudinal tire force is obtained by looking up a table based on the Magic Formula tire model, and the wheel speed control target guided by torque vector control is obtained by calculating the real-time vehicle speed and the longitudinal slip ratio.

[0231] Step 4, ASR-TVC target fusion design based on stability metrics:

[0232] Vehicle stability indices are defined based on the front and rear wheel slip angle phase plane, dividing the area into safe zone, transition zone, and danger zone. The target wheel speed, which is weighted by the vehicle stability index and is guided by both handling and stability, is dynamically adjusted according to the stability margin to achieve the unified goal of drive anti-slip-torque vector control (ASR-TVC).

[0233] Furthermore, to simultaneously meet the stability requirements of drive anti-slip and the maneuverability requirements of torque vector, step four proposes a method based on the front and rear wheel slip angles. Vehicle stability evaluation index in phase plane Quantitatively characterize the stability margin; calculate the vehicle state in real time. The position in the phase plane is dynamically divided into safe zones, transition zones, and danger zones, and the target wheel speed is integrated with maneuverability and stability guidance based on the region. and When the vehicle is in the safe zone, the handling objective of torque vector is dominant; when in the danger zone, the stability objective of driving anti-slip is dominant; in the transition zone, the indicators are continuously weighted and fused, and the fusion expression is... To achieve the adaptive superposition of two objectives, avoid abrupt control switching, and improve vehicle stability under complex operating conditions through drive anti-slip control, while simultaneously enhancing vehicle handling performance using torque vector control. The specific steps are as follows:

[0234] like Figure 4 As shown, the vehicle status is in Position in the phase plane Design vehicle stability evaluation indicators:

[0235] ;

[0236] According to the vehicle dynamics characteristics of the present invention, the saturated sideslip angle of the front wheels is generally greater than that of the rear wheels.

[0237] The green area representing the safety zone is drawn with the origin as the center and the rear wheel saturation sideslip angle as the radius. , As the vehicle's state gradually moves away from its origin, the stability evaluation indicators... The value gradually increases if the location is within the safe zone. If the boundary is defined, then the stability evaluation index is taken as: ,when When this condition is met, the vehicle's dynamic capabilities are utilized at a low rate under the current operating conditions, and the vehicle is stable.

[0238] The blue area representing the transition zone is obtained by drawing a circle with the front wheel saturation sideslip angle as the radius. , , ~ Corresponding to the transition zone from Boundary to The stability evaluation index of the boundary is set such that the utilization rate of the vehicle's power capacity continuously increases under the current working conditions.

[0239] With two saddle points and A red area representing the danger zone is obtained by drawing a circle with half the distance between them as the radius. , , ~ Corresponding to the danger zone from Boundary to The stability evaluation index at the boundary is highly relevant to vehicle dynamics; even slight disturbances can lead to vehicle instability. It is worth noting that, to avoid... Excessive changes in weighting factors cause oscillations and limit controller performance, especially when the vehicle is in a danger zone. In addition, stability evaluation indicators Restricted to .

[0240] Then, the dynamic requirements for wheel speed control are adjusted according to the changes in the vehicle's stability zone; each stability zone represents a different vehicle stability state, which is used as a condition for dynamically switching control requirements in the fusion control objective; combined with vehicle stability evaluation indicators. The system quantitatively identifies the vehicle's position and stability, and dynamically adjusts the weighting coefficients of the fused wheel speed target. These coefficients reflect the priority of the composite control target, thus clearly achieving the switching of handling stability requirements shown in Table 2.

[0241] Table 2

[0242]

[0243] The relationship between vehicle stability and the dynamic control requirements of the composite wheel speed target is described as follows:

[0244] When the vehicle is in the safe zone R1, the primary goal is to improve vehicle handling; therefore, handling wheel speed control is the priority. When the vehicle is far from the safe zone, lateral stability requirements need to be considered. Therefore, when the vehicle is in the transition zone R2, vehicle stability evaluation indicators... The increasing weight of the stability wheel speed control target means that the priority of the stability index increases with the increase of the stability index, while the priority of the maneuverability wheel speed control target decreases with the increase of the stability evaluation index. At the same time, since the vehicle still has a good stability state and a certain stability margin within the transition zone R2, the weight of the maneuverability wheel speed control target still exceeds that of the stability wheel speed control target, and it is dominant. When the vehicle is in the danger zone R3, the vehicle stability evaluation index is larger, and the stability wheel speed control target is given priority.

[0245] As shown in the formula, the dynamic changes in vehicle stability evaluation indicators enable adaptive adjustment of the control target priority to meet dynamic control requirements throughout the entire control process.

[0246] ;

[0247] in, The steering wheel speed control target is determined by torque vector control. The stability wheel speed control target determined by the anti-skid control mechanism. The target fusion coefficient is for wheel speed control. The goal is to achieve a fusion of wheel speed control that is suitable for drive anti-slip and torque vector coordinated control.

[0248] Step 5: Co-design of incremental PID drive and braking controller:

[0249] Design an incremental PID motor / brake co-controller for electric vehicles with independent front and rear axle drive. The controller uses the joint control of motor torque and braking torque to update the control quantity for the deviation between the target wheel speed and the measured wheel speed, and tracks the unified target to achieve the requirements of drive anti-slip and torque vector coordinated control. It also applies braking to the low-adhesion side of the split road surface to suppress slip and releases high-adhesion traction.

[0250] Furthermore, the driver's desired total driving torque is obtained based on the pedal analytical model. The front and rear axle torques are distributed, and the target driving torque is obtained through first-order filtering. The main function of the accelerator pedal analytical model is to generate the driver's desired front and rear axle motor torques based on the pedal opening. The final target torque is obtained through front and rear axle torque distribution combined with first-order filtering and motor limiting processing. The drive anti-slip control system studied in this invention is based on this torque, and its expression in the S-domain is:

[0251] ;

[0252] in, , These are the driving forces for the front and rear axles, respectively. Let be the time constant of a first-order inertial element, and s be the complex frequency variable. , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively.

[0253] After ASR-TVC intervention, the incremental PID motor torque control module is activated. It calculates the motor torque increment based on the larger value of the fused wheel speed control target between the left and right wheels on the same axle and the deviation of the actual wheel speed. The torque arbitration module then adds this increment to the driver's torque to generate the current front axle motor torque request value. Subsequent control is based on the previous request value and continuously iterates and updates to achieve drive anti-slip motor torque control.

[0254] The formula for incremental PID motor torque control is:

[0255]

[0256] in, To control the increment; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; This represents the deviation between the target and the output at the current moment. This represents the deviation between the target and the output in the previous long-term step. This represents the deviation between the target and the output at the first two time steps.

[0257] On low-speed split-plane surfaces, the side with the high coefficient of friction is prone to wheel slippage due to the limited road surface adhesion conditions, influenced by the dynamics of the side with the high coefficient of friction. When ASR-TVC intervenes and the operating condition is determined to be a split-plane surface, the braking torque control state is activated, applying braking torque to the side with the low coefficient of friction to suppress slippage and release greater driving force. An incremental PID braking torque control module is built using the same incremental PID control algorithm as the motor torque control. The incremental braking torque on the side with the low coefficient of friction is calculated based on the measured wheel speed difference and the target wheel speed difference, and then superimposed with the braking torque decided at the previous moment to achieve anti-slip control of split-plane surface drive in coordination with motor torque, thereby improving the driving capability and stability of split-plane surfaces.

[0258] Multi-level experimental verification:

[0259] Based on the V-model development process of vehicle control software, a full-link test system from model design to real vehicle verification is proposed for the coordinated control of drive anti-slip and torque vector.

[0260] To ensure the stable and reliable operation of the proposed complex road condition recognition-based drive anti-skid and torque vector coordinated control method under various complex road surfaces and working conditions, this embodiment proposes a multi-level experimental verification scheme. First, a CarSim-MATLAB / Simulink co-simulation platform is established, and typical working conditions such as low-friction straight-line start, split-road and connected road surfaces, and hill start are designed to verify the correctness of the control logic and the rationality of the target generation method. Subsequently, the same working conditions are applied to the hardware on a ring test bench, focusing on evaluating the real-time performance, computational load, and compatibility with automotive-grade vehicle control units and braking systems of the drive anti-skid and torque vector coordinated control method. Finally, a real-vehicle test scheme is designed, covering high-risk road conditions and road conditions that easily trigger the control method of this invention, such as ice surfaces, asphalt-ice split-road, and 20% slopes, to examine the robustness and stability of the control method in real-world environments. The entire process targets the coordinated control of drive anti-slip and torque vector, covering three levels: software simulation, hardware-in-the-loop (HIL) testing, and real vehicle verification. It progressively advances the verification of control methods around the control objectives of improving vehicle handling and maintaining stability, ensuring consistency and traceability across the entire chain from algorithm design and controller implementation to vehicle application.

[0261] First, a joint simulation platform and experimental scheme were designed for the collaborative control method of anti-slip and torque vector control in distributed drive electric vehicles. In the early stages of development, a CarSim-MATLAB / Simulink joint simulation platform was established, constructing a full physical system including a seven-DOF vehicle dynamics model, independent front and rear axle motor drive models, brake actuator models, and a PAC2002 tire model. The control method was implemented in a modular structure within the Simulink environment, facilitating signal interface definition and parameter tuning. A sampling step of approximately 10ms was recommended for the simulation to ensure numerical accuracy and to verify the numerical stability of the algorithm under high-frequency disturbances.

[0262] The working condition design covers typical complex road surfaces, including:

[0263] 1. Full throttle start on a low-adhesion, uniform road surface: This operating condition is achieved by setting a uniform low-adhesion coefficient ( On a road surface with low traction, an electric vehicle was driven to start at full throttle to examine its traction and wheel speed control effectiveness. The test recorded the wheel speeds, drive torque, longitudinal acceleration, and yaw rate of all four wheels. This verified whether the drive anti-slip control method could quickly suppress wheel slippage and stabilize wheel speed under low traction conditions, ensuring that the wheel speed matches the equivalent target wheel speed. This ensured that the drive motor's output capacity was maximized under low traction conditions. The test also assessed whether there were any power interruptions or excessive wheel speed fluctuations during the start-up process, thus confirming the adaptability and stability of the control method under low traction conditions.

[0264] 2. Low-adhesion-high-adhesion road surface connection full throttle start: This working condition is achieved by setting the low-adhesion ( ) and high-precision ( In a composite road section with longitudinally spliced ​​surfaces, vehicles start at full throttle in the low-adhesion zone and enter the high-adhesion zone to examine the dynamic response of adhesion level identification and control target switching. The focus is on observing the smoothness of the transition from low to high adhesion level, the target slip ratio update delay, and the continuity of the drive torque ramp-up process. This ensures that the control method can promptly release the high-adhesion traction potential without causing wheel speed fluctuations or sudden changes in driving force, guaranteeing a linear increase in vehicle longitudinal acceleration and traction efficiency while maintaining lateral stability.

[0265] 3. Starting on a road surface with the left wheel lower than the right: This condition involves setting a significant difference in the adhesion coefficient between the left and right wheels (left wheel lower than the right wheel). ,right This simulates a typical starting scenario on a split-plane road surface with uneven left and right adhesion. The vehicle starts at full throttle, focusing on verifying the coordination between low-adhesion braking torque control and high-adhesion driving force release. The experiment observes the speed difference between the left and right wheels, changes in braking torque, and the distribution of driving torque to ensure rapid control of the low-adhesion slip ratio and increased right-side driving force, thus verifying the vehicle's straight-line driving ability and directional stability under uneven adhesion conditions.

[0266] 4. Starting on a 20% slope with uneven adhesion: In this condition, the vehicle is placed on a road surface with uneven adhesion on both sides at a 20% slope, with the left side having higher adhesion ( ), right-side low-grade ( The test examines the vehicle's driving anti-slip capability and traction coordination under incline conditions. The experiment focuses on the dynamic process of suppressing low-friction wheel slip and increasing high-friction driving force, ensuring coordinated output of motor torque and braking torque to avoid unilateral wheel spin or vehicle start-up failure. By recording longitudinal acceleration, wheel speed, and steering correction, the stability, climbing ability, and torque response efficiency of the method under complex incline conditions are evaluated.

[0267] 5. Low-adhesion Curve Acceleration and Lane Change: This test involves starting and accelerating the vehicle on a low-adhesion circular track with a radius of 50 m, and performing one or two lane changes during acceleration to examine the vehicle's yaw response and stability maintenance capabilities. Changes in yaw rate, sideslip angle, and steering angle are recorded to evaluate whether the control method can improve the vehicle's yaw response characteristics while maintaining longitudinal traction, keeping the driving trajectory close to the reference trajectory, preventing vehicle instability or understeer, and ensuring good handling stability under combined maneuvers.

[0268] Multiple parameter combinations are set for each operating condition, including different gradients, adhesion coefficients, initial speeds, tracking radii, lane change frequency, and accelerator pedal opening, to ensure coverage of extreme driving conditions and boundary states. Simulation outputs include key indicators such as single-wheel speed, longitudinal acceleration, yaw rate, center of gravity sideslip angle, and drive torque changes, used to evaluate the wheel slip suppression effect and vehicle handling stability control effect. The control method optimizes the slip ratio threshold, target wheel speed threshold, torque vector hierarchical architecture control weights, and PID gain parameters through iterative simulation to ensure timely control intervention and smooth exit.

[0269] After completing the joint simulation verification, due to the limited computing power of the electronic control unit in mass-produced vehicles, this embodiment designs a bench test scheme based on a hardware-in-the-loop (HIL) platform for the proposed electric vehicle drive anti-slip and torque vector cooperative control method based on complex road condition recognition to further evaluate the real-time performance, robustness, feasibility, and fault tolerance. The control method is downloaded to the automotive-grade controller prototype with the same modular architecture as in the simulation stage and communicates with the real-time simulation host via the CAN bus. The test first verifies the functional correctness under ideal signal conditions, and then introduces interference such as sensor noise, communication delay, and signal packet loss to evaluate the controller's fault tolerance to abnormal inputs and system recovery characteristics.

[0270] The operating condition design and simulation are consistent, with a focus on the following aspects:

[0271] 1. Real-time performance: Measures the sampling period jitter and task scheduling delay of the measurement controller to ensure that the calculation is completed within a 10 ms control cycle;

[0272] 2. Robustness: Simulate sudden changes in the adhesion coefficient with different amplitudes and frequencies to check whether the controller can converge stably and avoid oscillations;

[0273] 3. Safety: Verify whether the controller can promptly disengage torque regulation to prevent vehicle loss of control when sensors fail or signals are abnormal.

[0274] The HIL test output signal was compared with the simulation benchmark. Further key parameter tuning and control logic optimization were conducted to confirm that the drive anti-slip control method maintains consistent performance in the embedded environment.

[0275] Following the completion of the HIL test, a real-vehicle test scheme was designed and implemented for the electric vehicle drive anti-skid and torque vector coordinated control method based on complex road condition recognition, to verify the effectiveness, stability, and mass production feasibility of the control method under real road conditions. The test vehicle was a distributed drive electric vehicle prototype, equipped with an automotive-grade vehicle control unit (VCU), motor control unit (MCU), and integrated brake control unit (IBC), and configured with high-precision wheel speed sensors, inertial measurement units (IMU), brake pressure sensors, and a CAN data acquisition system to achieve real-time acquisition of all key dynamic signals.

[0276] The test conditions are designed to cover typical complex road surfaces and extreme driving conditions that are prone to wheel slippage, including:

[0277] 1. Full throttle start on a straight ice surface: This test is used to verify the slip suppression capability and longitudinal acceleration recovery effect under low adhesion conditions, ensuring that the control method can maximize traction output.

[0278] 2. Starting on an asphalt-ice surface: The focus is on evaluating the coordination between low-side braking torque adjustment and high-side torque release, verifying the stability of straight-line driving and the full utilization of traction potential.

[0279] 3. Starting on a 20% slope with uneven road surface: This test examines the driving anti-slip capability and the coordinated control of motor torque and braking torque under conditions of extremely uneven road surface adhesion, to ensure reliable vehicle start-up.

[0280] 4. Low-adhesion cornering full-throttle acceleration: Test the effectiveness of drive anti-slip and torque vector coordinated control under high lateral loads to ensure the accuracy, stability and controllability of trajectory tracking.

[0281] During the testing, key indicators such as average longitudinal acceleration, peak wheel speed, peak yaw rate, and steering wheel correction were collected to comprehensively evaluate the longitudinal traction capability and lateral handling stability of the traction control method combined with torque vector control. To ensure the comparability and statistical significance of the test data, all tests were conducted under uniform initial vehicle speed, accelerator pedal opening, and vehicle load conditions, and were repeated at least three times to obtain the average value. This scheme, through a systematic design covering multiple operating conditions, multiple indicators, and multiple repetitions, ensures the functional correctness, dynamic response, and robustness of the control method on real vehicles, providing sufficient technical basis for the mass production implementation of the method.

[0282] During the V-shaped software development process, the cooperative control method provided by this invention enhances control robustness by utilizing complex road identification and incorporates expert knowledge from real-vehicle calibration into the design of the drive anti-slip control method, improving feasibility. It dynamically coordinates drive anti-slip and torque vector control objectives based on real-time vehicle stability status, thereby improving vehicle handling stability under complex conditions. Therefore, it can achieve considerable control performance with relatively low computational cost, ensuring real-time feasibility for mass-produced vehicle applications and significantly improving vehicle traction and handling stability under complex road conditions.

Claims

1. A method for coordinated control of drive anti-skid and torque vector based on complex road condition identification, characterized in that: Includes the following steps: Step 1: Construct an online estimation of single-wheel adhesion coefficient based on unscented particle filtering, and combine it with the slip ratio to update the threshold to suppress the jitter of adhesion coefficient estimation when the slip is low; and combine it with the identification of open / connected road surfaces and the switching of uniform road surface adhesion level finite state machine to form a reliable road surface type perception and level management. Step 2: Establish a drive anti-skid control intervention / exit criterion that takes into account both driver intent and slip threshold. Calculate the optimal slip ratio according to road surface level to generate target wheel speeds for all four wheels, and set a low-speed threshold to adaptively adjust the target wheel speeds for stability guidance. Step 3: A torque vector hierarchical control architecture is proposed. The upper layer calculates the ideal yaw target based on the steady-state vehicle dynamics model. The middle layer calculates the additional yaw moment through a three-step method of steady-state compensation + feedforward + linear quadratic programming feedback with embedded linear quadratic programming. The lower layer outputs the target wheel speed for maneuver guidance through a multi-objective optimization allocation method. Step 4: Define vehicle stability indicators based on the front and rear wheel slip angle phase plane, and divide the safe zone, transition zone and danger zone; By integrating target wheel speeds guided by both handling and stability with vehicle stability indices as weights, and dynamically adjusting the priority according to stability margin, a unified goal of drive anti-slip-torque vector coordinated control is achieved. Step 5: Design an incremental PID motor / brake co-controller for electric vehicles with independent front and rear axle drive. Utilize the joint control of motor torque and braking torque to update the control quantity for the deviation between the target wheel speed and the measured wheel speed, track the unified target to achieve the requirements of drive anti-slip and torque vector coordinated control, and apply braking to the low-adhesion side of the split road surface to suppress slip and release high-adhesion traction.

2. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: Step one includes the following steps: First, an unscented particle filter (UPF) is used to estimate the adhesion coefficient of each wheel based on the PAC2002 tire model in real time. A slip ratio threshold update method is introduced, updating the adhesion estimate only when the absolute slip ratio exceeds the threshold. Second, a sliding window integral method based on slip ratio is used to identify split-road surfaces. By comparing the trends of the integral slip ratios of the left and right wheels, misjudgments caused by instantaneous fluctuations are suppressed, and a split-road surface identification state machine is formed by combining the braking torque threshold. For docked surfaces, an identification criterion based on the rate of change of wheel velocity angular acceleration and the trend of driving torque is designed to achieve timely switching of control targets. Finally, a uniform road surface adhesion level switching rule is proposed, dividing the road surface into three levels: low, medium, and high, and dynamically adjusting the target slip ratio according to the identification results.

3. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: The specific steps for Step One are as follows: First, a single-wheel adhesion coefficient estimation algorithm based on unscented particle filtering is used, followed by a four-wheel adhesion coefficient estimation algorithm. As state variables, longitudinal and lateral tire forces , As an observation, a state observation equation is established based on the PAC2002 tire model: ； in, and For tire model functions, For process noise, , , These are the longitudinal, lateral, and vertical tire forces, respectively. and These are tire slip rate and sideslip angle, respectively. These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively, with subscripts. For the update step, and The difference is one update interval; the formulas for tire slip ratio and sideslip angle are as follows: , ； in, and These are the longitudinal and lateral wheel speeds, respectively. The angular velocity of the wheel. The effective rolling radius of the wheel; An update method based on a slip ratio threshold is adopted: only when the absolute value of the tire slip ratio is... Update the adhesion coefficient when the set threshold is exceeded; otherwise, retain the previous estimate. Secondly, the implementation steps of the adhesion coefficient estimation algorithm based on unscented Kalman filter (UKF) are as follows: During the initialization phase, when At that time, the initial particle set is generated. And set the weight of each particle to be ; Entering the recursive process, for Perform the following steps in sequence: Step 1: Determine the current slip ratio Is it less than the slip ratio threshold? ,like If the condition is met, proceed directly to Step 8 for termination condition check; otherwise, continue to Step 2. Step 2, for each particle UKF is used for prediction and updating to obtain particle state estimates. and its covariance ; Step 3, combining observations Calculate particle weights ; Step 4, normalize all particle weights as follows: ； Step 5, Calculate the number of effective particles : ； like If the value is below the set threshold, proceed to Step 6; otherwise, proceed to Step 7. Step 6: Perform system resampling, weighted... Generate a new set of particles and reset the weights to ; Step 7, Output the state estimate : ； Step 8, check the termination condition: if it is met, end the algorithm; otherwise, let... Then return to the slip ratio determination step and continue iterating until convergence.

4. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 3, characterized in that: Step three also includes rules for switching uniform road surface adhesion levels and methods for identifying non-uniform road surfaces, the steps of which are as follows: Split-road surface recognition method: Based on each axle, the road surface state of the left and right wheels of the front and rear axles is obtained separately; the sliding window integral method of slip ratio on both sides is used to determine the overall trend relationship. The core idea of ​​sliding window integral is: slide a fixed-size time window or data segment on the data sequence, calculate the integral of the data within the window, and update it continuously; when the difference between the sliding window integrals of the slip ratio of the left and right wheels is greater than a threshold value, it is considered that the split-road surface has been entered; at the same time, the braking torque on the low-adhesion side needs to be less than the standard braking torque; combined with the finite state machine method, the split-road surface recognition is realized. Road surface identification method: The criteria for determining whether a vehicle with a low coefficient of adhesion is entering a road surface with a high coefficient of adhesion are set as follows: A. Low-adhesion-coefficient road surfaces are only identified when anti-skid control is engaged. B. Within the calibrated time range, the actual torque of the motor remains unchanged; C. Within the calibrated time range, the wheel angular acceleration continuously decreases to the preset threshold value; Finally, a uniform road surface adhesion level switching rule is constructed, dividing the road surface into three levels: low, medium, and high, and calibrating the optimal slip ratio for each level. The adhesion level is dynamically switched based on the real-time estimated road surface adhesion coefficient and a threshold value. When a low-adhesion-coefficient connecting road surface or a split road surface with a high-adhesion-coefficient side is detected, if the road surface adhesion coefficient estimation stops updating, the adhesion level is increased to enhance the driving anti-skid control dynamics when the adhesion estimation response is slow under non-full acceleration or braking. The adhesion level switching is implemented through a finite state machine, and the switching rule is designed as follows: Set the initial adhesion level to: High adhesion; A. When the current state is high adhesion: If the road adhesion coefficient is dynamically updated If the road surface is not directly connected, switch to medium adhesion; otherwise, maintain high adhesion. B. When the current state is medium adhesion: If the road adhesion coefficient is dynamically updated... Furthermore, since it is a non-connected road surface, it is switched to low adhesion. Switch to high adhesion if any of the following conditions are met: Or, the road surface is connected from low-adhesion to high-adhesion and the high-adhesion side has not been updated, or the road surface is converted from uniform to split and the high-adhesion side has not been updated. Otherwise, maintain the intermediate adhesion state; C. When the current state is low adhesion: If the road adhesion coefficient is dynamically updated Or, if the road surface where low-adhesion road meets high-adhesion road and the high-adhesion side has not been updated, it will switch to medium-adhesion road. If the road surface changes from uniform to split and the high-adhesion side is not updated, switch to high-adhesion; otherwise, maintain low-adhesion.

5. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: Step two designed a drive anti-slip control intervention criterion and a slip ratio control target that take into account the human-vehicle-road integrated system. The intervention criterion determines the intervention and withdrawal of drive anti-slip control by combining the driver's acceleration demand and the slip ratio threshold. When the accelerator pedal signal increases significantly and the equivalent wheel speed exceeds the calibrated value, the motor torque adjustment is activated. When the braking or command torque deviation is too large, it is withdrawn to avoid interference with drivability. The anti-skid guide design combines the optimal slip ratio of a single wheel under different road surface adhesion coefficients. It switches the adhesion level and generates the target wheel speed through a finite state machine. The minimum target wheel speed limit is applied in the low speed area to prevent starting difficulties.

6. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: The specific steps for step two are as follows: First, for the motor torque and braking torque of the drive anti-slip control system, respectively, we design a drive anti-slip coordinated control intervention and withdrawal rule that combines the driver's acceleration intention and the preset threshold of the vehicle slip rate; The motor torque intervention rule setting for the anti-slip control system includes the following two aspects: A. When the accelerator pedal analysis indicates that the driver's acceleration intention is increasing, that is, an increase in the vehicle's power output is desired, the anti-slip control system needs to intervene. B. First, obtain the slip ratio intervention threshold through actual vehicle calibration. When the real-time wheel speed When the threshold is exceeded, that is The anti-slip control system needs to intervene and control the equivalent wheel speed to the target value. The motor torque withdrawal rule setting for the anti-slip control system includes the following two aspects: A. When the driver presses the brake pedal, it indicates that the vehicle needs to slow down or stop, and the traction control system should disengage in time. B. When the driving anti-slip control system's adjustment torque Exceeding the driver's expected torque Achieving a safety margin At that time, that is The system should be deactivated and the power level restored to the driver's desired level. Meanwhile, in order to achieve the coordinated control requirements of different slip ratios of the left and right wheels on the same axle under split-plane road conditions to enhance vehicle dynamics and handling stability, the intervention conditions for braking torque control are designed as follows: A. When the road surface is identified as a split road surface, the braking torque on the side with the low coefficient of friction is controlled to reduce wheel slippage on that side. Braking torque control intervenes at the same time; B. The anti-slip control system can only adjust the braking torque when the motor torque control is activated; The braking torque withdrawal condition is: the road surface is uniform, and the decision braking torque is less than the calibrated value. Or the driver presses the brake pedal; Then, the calculation formula for the target wheel speed of the four wheels was derived, and the driving anti-skid control target was determined based on the road conditions and vehicle status. The adhesion level switching is achieved through a finite state machine, and the optimal slip ratio for low, medium, and high adhesion surfaces is calibrated by combining real vehicle tests. In order to determine the optimal target speed for each wheel. subscript These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. Set low longitudinal speed threshold To avoid starting difficulties and low-speed oscillations due to excessively low target speeds at low speeds; when the longitudinal speed... When the speed is below this threshold, the target wheel speed is calculated according to the threshold; the optimal target rotational speed. The calculation is as follows: , 。 7. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: Step three includes the following steps: First, the upper layer of the hierarchical control architecture is the target of ideal yaw rate control. The design is based on a linear two-degree-of-freedom vehicle dynamics model that includes lateral and yaw motions. ； in, For vehicle quality, , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. This refers to the vehicle's wheelbase. , , These are the equivalent lateral stiffness of the front and rear axles, respectively. Front wheel steering angle, understeer degree ; ； Among them, steady-state yaw rate gain , The road surface adhesion coefficient, It is the acceleration due to gravity; Then, the intermediate layer control of the hierarchical control architecture adopts a three-step strategy: the first step is to solve the steady-state control quantity to compensate for the static deviation; the second step is to introduce reference feedforward control to improve the dynamic response; and the third step is to design state feedback control based on linear quadratic programming to minimize the yaw rate and centroid side slip angle error, while also taking into account the optimal control energy. In the intermediate layer controller design, a nonlinear two-degree-of-freedom vehicle model is adopted, combined with a tire model to describe the nonlinear characteristics of the tire; the lateral dynamics equation of the vehicle with applied additional yaw moment control is expressed as: ； in, For vehicle quality, Let be the longitudinal speed of the vehicle. Let be the lateral speed of the vehicle. and These are the lateral forces on the front and rear axles, respectively. It is the moment of inertia of rotation about the z-axis. The yaw rate is angular velocity. , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. This refers to the front wheel steering angle. To add yaw moment; The first step of the three-step method, which involves solving for the steady-state control quantity to compensate for the static deviation, is as follows: make Steady-state feedforward control quantity The following relationship must be satisfied: ； Determine the desired steady-state feedforward control input: ； The second step of the three-step method, introducing dynamic feedforward control to improve dynamic response, involves the following steps: Based on steady-state control, consider the system's ideal yaw rate control objective. The dynamic changes are introduced to generate dynamic feedforward control. The control input is expanded to: ； Substituting this into the vehicle lateral dynamics equations, we get: ； make The dynamic feedforward control quantity is obtained as follows: ； The third step of the three-step method, which involves designing state feedback control based on linear quadratic programming to minimize yaw rate and centroid sideslip angle errors, follows these steps: Feedback control technology based on linear quadratic programming is integrated into a three-step framework, using lateral dynamics without additional yaw moment as the reference model, and the ideal centroid sideslip angle is used. and ideal yaw rate control target Substitute; subtract the reference model from the controlled system, at which point the state vector... Represented as: β is the centroid sideslip angle; a linear quadratic programming algorithm is used to determine the feedback control quantity. To minimize state tracking error; and simultaneously construct an infinite time-domain cost function. Taking into account both the target state tracking error in maneuver stability and the execution energy loss in torque vector control: ； Where Q is a target state tracking weight matrix, and the execution energy loss weight matrix R=1; Weights in Q to control time and The tracking errors corresponding to the sideslip angle and yaw rate are designed as stability evaluation indicators. Functions: ； Where parameters =0.001, =0.5, =0.5, =0.8; The final additional yaw moment is: ； in, and These are the steady-state and dynamic feedforward control quantities based on the three-step method, respectively. This is a feedback control quantity based on linear quadratic programming; Finally, the lower layer establishes a constraint optimization model with the goal of minimizing the additional yaw moment distribution error and reducing tire load utilization. The additional yaw moment control amount calculated by the middle layer is distributed to the four wheels, and motor saturation and tire adhesion constraints are taken into account. The system dynamic equations are expressed as follows: ； in, , These are the front and rear wheelbases, respectively. , , , These represent the longitudinal tire forces of the left front, right front, left rear, and right rear wheels, respectively; the control input u is defined as: ； The problem of optimizing the allocation of lower-level drive torque is expressed as: ； in, The tire load utilization weighted matrix is ​​defined as follows: To prevent any single tire from saturating prematurely, The road surface adhesion coefficient, For vertical loads; This is the error weighting matrix, representing the control priority between the longitudinal driving force and the additional yaw moment; and Representing the actual and expected generalized force vectors respectively, the total driving force is based on the moment vector. and additional yaw moment composition; For control parameters; Considering actuator and road adhesion constraints, the longitudinal force of each tire must satisfy: ； in, This represents the maximum torque of each motor. This is the effective rolling radius of the wheel.

8. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: Step four proposes a method based on the front and rear wheel slip angles. Vehicle stability evaluation index in phase plane Quantitatively characterize the stability margin; calculate the vehicle state in real time. The position in the phase plane is dynamically divided into safe zones, transition zones, and danger zones, and the target wheel speed is integrated with maneuverability and stability guidance based on the region. and ; When the vehicle is in a safe zone, the driving objectives based on the torque vector are the primary considerations. When in a danger zone, the primary objective should be stability, which drives anti-slip measures. In the transition zone, the indicators are continuously weighted and fused, and the fused ideal wheel speed control target expression is obtained. To achieve the adaptive superposition of two objectives, For the purpose of maneuvering wheel speed control, For the purpose of stable wheel speed control, The target fusion coefficient is used for wheel speed control.

9. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: Step four involves the following steps: Based on the vehicle's condition, the front and rear wheel slip angles Position in the phase plane Design vehicle stability evaluation indicators: ； According to vehicle dynamics, the saturated sideslip angle of the front wheels is greater than that of the rear wheels. Centered on the origin, the rear wheel saturation sideslip angle Draw a circle with a radius to obtain the green area representing the safe zone. , As the vehicle's state gradually moves away from its origin, the stability evaluation indicators... The value gradually increases if the location is within the safe zone. If the boundary is defined, then the stability evaluation index is taken as: ,when When this condition is met, the vehicle is stable under the current operating conditions; When the vehicle's front wheel saturation sideslip angle Draw a circle with radius to obtain the blue area representing the transition zone. , , ~ Corresponding to the transition zone from Boundary to The stability evaluation index of the boundary is determined by the fact that the utilization rate of the vehicle's power capacity is constantly increasing under the current working conditions. With two saddle points and A red area representing the danger zone is obtained by drawing a circle with half the distance between them as the radius. , , ~ Corresponding to the danger zone from Boundary to The stability evaluation index at the boundary is set when the vehicle is unstable; when the vehicle is located in a danger zone. In addition, stability evaluation indicators Restricted to ; Then, the dynamic requirements for wheel speed control are adjusted according to the changes in the vehicle's stability zone; each stability zone represents a different vehicle stability state, which is used as a condition for dynamically switching control requirements in the fusion control objective; combined with vehicle stability evaluation indicators. The system quantitatively identifies the vehicle's position and stability, and reflects the priority of the composite control objective by dynamically adjusting the weighting coefficients of the fused wheel speed target. The relationship between the vehicle's stable state and the dynamic control requirements of the composite wheel speed target is described as follows: When the vehicle is in the safe zone R1, the steerable wheel speed control objective is the priority control objective; when the vehicle is in the transition zone R2, the vehicle stability evaluation index... As the stability index increases, the priority of the stability wheel speed control target increases, while the priority of the maneuvering wheel speed control target decreases. At the same time, the weight of the maneuvering wheel speed control target still exceeds that of the stability wheel speed control target, and it is dominant. When the vehicle is in danger zone R3, the stability wheel speed control target should be given priority. The dynamic changes in vehicle stability evaluation indicators enable adaptive adjustment of control objective priorities to meet dynamic control requirements throughout the entire control process. ； in, The steering wheel speed control target is determined by torque vector control. The stability wheel speed control target determined by the anti-skid control mechanism. The target fusion coefficient is for wheel speed control. The goal is to achieve a fusion of wheel speed control that is suitable for drive anti-slip and torque vector coordinated control.

10. The method for coordinated control of drive anti-skid and torque vector based on complex road condition recognition according to claim 1, characterized in that: Step 5: Obtain the driver's desired total driving torque based on the pedal analytical model. The front and rear axle torques are distributed, and the target driving torque is obtained through first-order filtering. The main function of the accelerator pedal analytical model is to generate the driver's desired front and rear axle motor torques based on the pedal opening. The final target torque is obtained through front and rear axle torque distribution combined with first-order filtering and motor limiting. The anti-slip control system then controls the vehicle based on this final target torque. Its expression in the S-domain is: ； in, , These are the driving forces for the front and rear axles, respectively. Let be the time constant of a first-order inertial element, and s be the complex frequency variable. , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. After ASR-TVC intervention, the incremental PID motor torque control module is activated. It calculates the motor torque increment based on the deviation between the value of the fused wheel speed control target of the left and right wheels on the same axle and the actual wheel speed. The torque arbitration module then adds this increment to the driver's torque to generate the current front axle motor torque request value. Subsequent control is based on the previous request value and continuously iterates and updates to achieve drive anti-slip motor torque control. The formula for incremental PID motor torque control is: ； in, To control the increment; This is the proportionality coefficient; The integral coefficient; These are the differential coefficients; This represents the deviation between the target and the output at the current moment. This represents the deviation between the target and the output in the previous long-term step. This represents the deviation between the target and the output at the first two time steps. On split-road surfaces, the wheels on the low-adhesion side are prone to slippage due to the limited road surface adhesion conditions, influenced by the dynamic performance of the high-adhesion side. When ASR-TVC intervenes and the operating condition is determined to be split-road surface, the braking torque control state is activated, applying braking torque to the low-adhesion side to suppress slippage and release greater driving force. An incremental PID braking torque control module is built using the same incremental PID control algorithm as the motor torque control. The incremental braking torque on the low-adhesion side is calculated based on the measured wheel speed difference and the target wheel speed difference, and then superimposed with the braking torque decided at the previous moment to achieve split-road surface drive anti-slip control in coordination with motor torque.

Images