Stability control method and system for electric vehicle

Abstract

The present invention relates to the field of electric vehicle control technology and discloses a stability control method and system for electric vehicles. The method comprises: collecting vehicle driving data in real time and dynamically estimating the vehicle's state parameters using an unscented Kalman filter algorithm; calculating the expected yaw moment in the short future based on the vehicle's state parameters, in combination with a kinematic model and a forward-looking trajectory, and obtaining the vehicle's longitudinal traction requirement; and decoupling the yaw moment and traction force based on the vehicle's state parameters and the vehicle's longitudinal traction requirement, distributing them to the four electrically driven wheels to achieve stable control of the electric vehicle. Compared to traditional stability control systems based solely on closed-loop feedback, the present invention offers faster control response and more precise yaw intervention, demonstrating superior control performance and vehicle safety under low-adhesion road conditions and high-load conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle control, and in particular to a stability control method and system for an electric vehicle. Background Art

[0002] With the development of new energy vehicle technology, four-wheel independent drive (FWD) architectures are becoming increasingly prevalent in electric vehicles. This allows for independent control of the drive torque at each wheel, creating a technical foundation for advanced vehicle dynamics. However, most current vehicles still employ traditional stability control strategies that rely on mechanical braking systems or passively intervene only after instability occurs, lacking proactive control logic based on road condition and state awareness. This approach results in delayed response and limited control effectiveness in high-risk driving scenarios.

[0003] Furthermore, electric drive systems are limited by the motor's thermal capacity and cooling capabilities. Prolonged loads can easily cause the temperature to rise above a threshold, triggering power protection mechanisms and limiting driving capability. In current control systems, the motor's thermal state is often overlooked. This is especially true during stability control, where torque may be forced to be distributed across the wheel motors under high load, causing some motors to overheat and derating, ultimately compromising vehicle stability control and potentially posing safety risks.

[0004] This invention fundamentally improves the electric vehicle's ability to perceive the risk of yaw instability and the control response efficiency. It integrates three dimensions: yaw torque prediction, adhesion coefficient estimation, and motor thermal state constraint optimization, and realizes a stability control method that truly combines forward-looking judgment with multi-objective coordination to meet the higher requirements of future intelligent electric drive platforms for safety and dynamic control capabilities. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is that the existing electric vehicle control technology has the problems of yaw instability and ignores the influence of the motor thermal state on torque distribution.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a stability control method for an electric vehicle, comprising: real-time acquisition of vehicle driving data, and dynamic estimation of the vehicle's state parameters using an unscented Kalman filter algorithm; based on the vehicle's state parameters, combined with a kinematic model and a forward-looking trajectory, calculating the expected yaw moment in the short future period and obtaining the vehicle's longitudinal traction requirement; and based on the vehicle's state parameters and the vehicle's longitudinal traction requirement, decoupling the yaw moment and traction force and distributing them to the four electric drive wheel ends to achieve stable control of the electric vehicle.

[0008] As a preferred solution of the stability control method of an electric vehicle described in the present invention, the vehicle driving data includes vehicle speed, steering wheel angle, yaw angular velocity, lateral acceleration and wheel speed.

[0009] As a preferred solution of the stability control method of an electric vehicle described in the present invention, wherein: the estimating the state parameters of the vehicle includes constructing an extended state vector based on vehicle driving data and calculating the instantaneous slip rate of each wheel;

[0010] Generate a set of filters point, and dynamically adjust the process noise covariance matrix according to the current maximum slip rate;

[0011] The saturation value determined by the adhesion coefficient and the vertical load is introduced, and the lateral stiffness model is enveloped by the hyperbolic tangent function to construct a lateral tire force saturation model. The lateral tire force saturation model is embedded in the state propagation function. Nonlinear prediction of points;

[0012] according to The weighted mean and covariance of the predicted values of the points are calculated to obtain the prior estimation result of the state at the current moment, and the innovation term is calculated. The a priori estimation update is completed based on the unscented Kalman gain to obtain the vehicle state parameters, which include vehicle speed, sideslip angle, yaw rate and lateral adhesion coefficient.

[0013] As a preferred solution of the stability control method of an electric vehicle described in the present invention, the forward-looking trajectory is a sequence of target path points within a certain time window forward with the current moment of the vehicle as a reference, including multiple spatial position points and curvature information corresponding to the spatial position points.

[0014] As a preferred embodiment of the stability control method of an electric vehicle according to the present invention, the calculation of the expected yaw moment in the short future time period includes setting a fixed forward-looking time window. , calculated based on vehicle speed Within the foresight distance, perform cubic spline fitting on the foresight trajectory within the foresight distance, calculate the average curvature, and obtain the expected yaw angular velocity;

[0015] Calculate the yaw rate error and sideslip angle suppression error, use the vehicle's state parameters to linearize the bicycle model in real time, generate the state matrix and input matrix, and build an adaptive linearized vehicle model;

[0016] Set the prediction step size , additionally introduce the sideslip angle as the state estimation output, set the sideslip angle error as the target quantity, construct the objective function and constraints, and the objective function is expressed as:

[0017] ;

[0018] in, represents the controller objective function value; Indicates the prediction step length; express time; represents the weight of the yaw rate error; Indicates time The yaw rate error; Represents the weight of the sideslip angle error; Indicates time Side slip angle error; The weight representing the rate of change of torque; Indicates time The rate of change of torque; Represents the control input, time The yaw moment;

[0019] The adaptive linearized vehicle model, objective function and constraints are compiled into a QP problem. The optimal yaw moment sequence is obtained in a single cycle using a real-time QP solver. The first moment solution is taken as The desired yaw moment within.

[0020] As a preferred solution of the stability control method of an electric vehicle described in the present invention, the vehicle longitudinal traction demand is the target traction required by the vehicle according to the driver's input, which is calculated and obtained by the vehicle controller VCU.

[0021] As a preferred embodiment of the stability control method for an electric vehicle according to the present invention, the decoupling of the yaw moment and the traction force and the distribution thereof to the four electrically driven wheel ends includes establishing, based on the vehicle's state parameters and physical relationships, an equation constraint on the four-wheel torque on the longitudinal traction force and yaw moment of the entire vehicle, such that the four-wheel combined force satisfies the longitudinal traction requirement and the left-right distribution of the four wheels satisfies the desired yaw moment;

[0022] Dynamic constraints are set for each wheel's electric drive motor, including torque upper limit constraint, adhesion limit constraint, power limit constraint, and torque change rate constraint. The adhesion limit constraint estimates the maximum traction force based on the vehicle's state parameters and the current vertical load of each wheel, and is expressed as:

[0023] ;

[0024] in, Indicates the Motor torque output for each wheel; Indicates the effective radius of the wheel; Indicates time The lateral adhesion coefficient; Indicates the Vertical load on each wheel;

[0025] Under the premise of satisfying equality constraints and dynamic constraints, a quadratic programming optimization problem is constructed with yaw moment error, torque change rate and power loss as cost functions. The optimal torque distribution result is solved and calculated and sent to each drive motor and each wheel electric drive motor for execution.

[0026] A stability control system for an electric vehicle using any of the methods described in the present invention, wherein: a state estimation module collects vehicle driving data in real time and dynamically estimates vehicle state parameters using an unscented Kalman filter algorithm;

[0027] The yaw moment prediction module calculates the expected yaw moment in the short term based on the vehicle's state parameters, combined with the kinematic model and forward trajectory, and obtains the vehicle's longitudinal traction requirements;

[0028] The motor torque distribution module decouples the yaw torque and traction force and distributes them to the four electric drive wheel ends according to the vehicle's state parameters and longitudinal traction requirements, thereby achieving stable control of the electric vehicle.

[0029] Beneficial effects of the present invention: By constructing a predictive control model based on road condition perception and vehicle body state estimation, the present invention can predict and calculate the expected yaw moment before yaw instability occurs, thereby dynamically allocating motor torque in advance and achieving proactive stability control of the vehicle. Furthermore, the method introduces a motor thermal constraint model to prevent power derating caused by motor overheating while ensuring stability, thereby improving the long-term reliability of the system. Compared with traditional stability control systems based solely on closed-loop feedback, the present invention has a faster control response and more precise yaw intervention, demonstrating superior control performance and vehicle safety on low-adhesion roads and under high-load conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts.

[0031] Figure 1 This is an overall flow chart of a stability control method for an electric vehicle provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0032] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0033] Example 1, reference Figure 1 , as one embodiment of the present invention, provides a stability control method for an electric vehicle, comprising:

[0034] S1: Collect vehicle driving data in real time and use the unscented Kalman filter algorithm to dynamically estimate the vehicle's state parameters.

[0035] The sensor network collects real-time vehicle driving data closely related to vehicle dynamic behavior. Vehicle driving data includes vehicle speed, steering wheel angle, yaw rate, lateral acceleration, and wheel speed.

[0036] By continuously collecting and processing this driving data, the vehicle's dynamic state data structure is constructed in real time, providing accurate, low-latency data support for subsequent yaw rate control, adhesion coefficient estimation, and motor torque distribution algorithms. All sensor acquisition modules are connected to the controller via the vehicle's CAN bus or high-speed Ethernet interface, supporting an update frequency exceeding 100Hz to ensure rapid response and anticipation of unexpected operating conditions.

[0037] Furthermore, in order to achieve real-time estimation of the key dynamic states of electric vehicles and improve the responsiveness and environmental adaptability of the yaw stability control system, the unscented Kalman filter (UKF) is used to synchronously estimate important state quantities that are difficult to measure directly, such as the vehicle's sideslip angle and lateral adhesion coefficient.

[0038] Obtain the current vehicle driving data, including longitudinal speed, steering wheel angle, yaw rate, lateral acceleration and wheel speed of each wheel, and construct an extended state vector ,in Indicates the moment, represents the sideslip angle, represents the yaw angular velocity, Represents the lateral adhesion coefficient. Based on this, the system automatically calculates the slip ratio of the four tires at the current moment and extracts the maximum slip ratio value as an indicator variable for sudden changes in operating conditions.

[0039] Generate a set of filters point and dynamically adjusts the process noise covariance matrix based on the current maximum slip rate. When the vehicle experiences sudden changes such as forced slip, low adhesion, or sudden acceleration or deceleration, the process noise is automatically amplified, thereby improving the dynamic tracking capability of state propagation and enhancing the estimation robustness.

[0040] According to the vehicle nonlinear lateral dynamics model (single track model), State propagation is performed at each point. While traditional linear tire force models are approximately effective under conditions of small slip angles and high adhesion, they tend to overestimate lateral forces when the road adhesion coefficient decreases or the slip angle increases, leading to distorted state estimation. This paper introduces a saturation value determined jointly by the adhesion coefficient and vertical load, and uses a hyperbolic tangent function to envelop the cornering stiffness model, constructing a saturated lateral tire force model with adhesion adjustability, expressed as:

[0041] ;

[0042] in, Indicates the lateral force generated by the current tire; Indicates the current vertical load on the wheel; represents the hyperbolic tangent function; Indicates the tire's cornering stiffness coefficient; Represents the current wheel slip angle; the model can scale the output amplitude in real time according to the road adhesion coefficient under different road conditions to ensure that force estimation overshoot is avoided in high slip or low adhesion scenarios.

[0043] After completing the state propagation, according to The weighted mean and covariance of the predicted values of the points are calculated to obtain the prior estimate of the state at the current moment. The state vector is mapped to the observation space through the measurement function and compared with the measured yaw rate, lateral acceleration and wheel speed data. The innovation term is calculated and the a priori estimate update is completed based on the unscented Kalman gain to obtain the vehicle state parameters, which include vehicle speed, sideslip angle, yaw rate and lateral adhesion coefficient.

[0044] It should be noted that the fixed covariance assumption causes the filter to converge too slowly in high-slip scenarios, and may even cause it to diverge due to modeling errors. Using the maximum tire slip rate as an "uncertainty indicator," the four-wheel slip rate is calculated each cycle and the maximum value is taken. This is multiplied by an empirical coefficient to form incremental noise, which is then added to the reference matrix to update the adaptive process noise covariance:

[0045] ;

[0046] in, represents the adaptive process noise covariance; represents the initial process noise covariance; represents the empirical coefficient; Indicates the Maximum tire slip at the moment; represents the identity matrix; it linearly amplifies the process noise as the slip rate increases, allowing the filter to make faster corrections to the model error.

[0047] S2: Based on the vehicle's state parameters, the kinematic model and the forward trajectory are combined to calculate the expected yaw moment in the short future and determine the vehicle's longitudinal traction requirements. Furthermore, the forward trajectory is a spatial representation of the expected driving path over a period of time or distance forward, based on driver input or data provided by the autonomous driving path planner, under the vehicle's current driving state. This is used to predict the vehicle's future yaw behavior. The forward trajectory includes the forward distance or forward time period, a sequence of trajectory points, and a target yaw behavior reference. Path points can be derived from the driver's steering input or provided by the autonomous driving system's path planning module.

[0048] Furthermore, a fixed look-ahead time window is set , calculated based on vehicle speed Within the foresight distance, a cubic spline fitting is performed on the foresight trajectory within the foresight distance, the average curvature is calculated, and the expected yaw angular velocity is obtained.

[0049] Calculate the yaw rate error and slip angle suppression error. The yaw rate error is the difference between the desired yaw rate and the actual yaw rate, and the slip angle suppression error is expressed as the negative value of the slip angle. Use the vehicle's state parameters to linearize the bicycle model in real time, generate a state matrix and replace the fixed parameter matrix with the input matrix, and construct an adaptive linearized vehicle model:

[0050] ;

[0051] in, Indicates time The state vector of Indicates time The state vector of represents the discrete state matrix; represents the input matrix; Represents the control input, time The yaw moment; and According to real-time change.

[0052] Set the prediction step size , additionally introduce the sideslip angle as the state estimation output, set the sideslip angle error as the target quantity, construct the objective function and constraints, and the objective function is expressed as:

[0053] ;

[0054] Constraints:

[0055] ;

[0056] in, represents the controller objective function value; Indicates the prediction step length; express time; represents the weight of the yaw rate error; Indicates time The yaw rate error; Represents the weight of the sideslip angle error; Indicates time Side slip angle error; The weight representing the rate of change of torque; Indicates time The rate of change of the moment is obtained by calculating the difference between the yaw moment at the current moment and the previous moment; Represents the control input, time The yaw moment; represents the maximum yaw moment; Indicates the maximum yaw moment change rate. Calculated from the current adhesion coefficient estimate to ensure that the required torque is within the tire lateral force limit.

[0057] The adaptive linearized vehicle model, objective function and constraints are compiled into a QP problem. The optimal yaw moment sequence is obtained in a single cycle using a real-time QP solver. The first moment solution is taken as The desired yaw moment within.

[0058] Unlike traditional ESP (Electronic Stability Program) that only controls yaw rate, the present invention introduces a joint objective function for sideslip angle error and yaw rate error to achieve multi-objective coordinated control of the vehicle's posture, avoiding the side slip amplification problem caused by single yaw rate control and improving overall control performance.

[0059] Furthermore, the vehicle longitudinal traction demand is the target traction required by the vehicle based on the driver's input, which is calculated and obtained by the vehicle controller VCU.

[0060] Specifically, the vehicle's existing pedal sensors or higher-level control module detect the driver's intentions or the acceleration / deceleration requirements issued by the autonomous driving system. The VCU, which already has built-in vehicle longitudinal control strategies (such as starting power limit, regenerative braking, and drive distribution), calculates the total traction demand or total driveshaft torque at the current moment. The calculation of total traction demand by the vehicle controller (VCU) is a standard feature of all current electric vehicles. This system serves only as a lower-level execution module to read the VCU output.

[0061] S3: Based on the vehicle's state parameters and longitudinal traction requirements, the yaw torque and traction force are decoupled and distributed to the four electric drive wheel ends to achieve stable control of the electric vehicle.

[0062] Based on actual driving requirements, the motor torque distribution module needs to distribute the desired yaw moment and longitudinal traction requirements to the four wheel-end motor torques within a 1-2ms control cycle, taking into account tire adhesion limits, drive power, temperature rise constraints, and ride comfort. Specific steps include:

[0063] Receives control inputs from the upper layer in real time, including: desired yaw moment (output by the yaw prediction module), longitudinal traction demand (calculated by the vehicle controller or driver input), current motor temperature and torque output values of each wheel, lateral adhesion coefficient and vertical load (provided by the state estimation module).

[0064] Based on the vehicle's state parameters and physical relationships, an equation constraint is established for the four-wheel torque to the longitudinal traction and yaw moment of the vehicle. The combined force of the four wheels meets the longitudinal traction requirement, and the left and right distribution of the four wheels meets the expected yaw moment.

[0065] Set dynamic constraints for each wheel's electric drive motor, including torque upper limit constraints (including thermal derating):

[0066] ;

[0067] in, Indicates minimum torque; Indicates the Target output torque for each wheel; Indicates the maximum available torque in relation to temperature; when the motor winding temperature When it is higher, the upper torque limit automatically decreases with the temperature;

[0068] The adhesion limit constraint estimates the maximum traction through the vehicle state parameters and the current vertical load of each wheel, which is expressed as:

[0069]

[0070] in, Indicates the Motor torque output for each wheel; Indicates the effective radius of the wheel; Indicates time The lateral adhesion coefficient; Indicates the The vertical load of each wheel; the adhesion limit constraint is used to ensure that the torque at each wheel end does not exceed the maximum traction allowed by the road surface to avoid slipping;

[0071] Power limit constraints:

[0072] ;

[0073] in, represents the wheel angular velocity; Indicates the The maximum allowed electric power per wheel end, with power limit constraints to prevent overload of the drive, battery or inverter system;

[0074] Torque rate limit:

[0075] ;

[0076] in, Indicates the torque output of the wheel in the previous control cycle; Indicates the maximum permissible rate of change of torque; Indicates the control cycle time; the torque change rate limit is used to prevent large sudden changes in the motor torque command, thereby avoiding vehicle dynamic instability or ride discomfort.

[0077] Under the premise of satisfying the equality constraints and dynamic constraints, a quadratic programming optimization problem with yaw moment error, torque change rate and power loss as cost functions is constructed, which can be expressed as:

[0078] ;

[0079] in, 、 、 Represents the weight factor in the control objective function, which can be set adaptively according to the working conditions; Represents the torque vector of the four-wheel motor; It represents the theoretical equilibrium torque under the average distribution of traction force. It consists of three parts: the first one punishes torque mutation, the second one punishes motor power loss, and the third one punishes traction distribution deviation, ensuring that the system energy efficiency and control stability are taken into account while meeting the yaw control.

[0080] The optimal torque distribution is calculated using a sparse QP solver (such as OSQP). After filtering and limiting, the result is sent to each drive motor and each wheel's electric drive motor for execution. The output is recorded and used to constrain the rate of change for the next cycle. If any wheel-end output reaches saturation or a constraint boundary, control risk signals such as insufficient adhesion or execution force are fed back to the upper-level system.

[0081] Through this method, the present invention not only achieves dynamic coordinated control of all four electric vehicle torques, but also, by jointly constraining adhesion and thermal conditions, ensures vehicle stability, safety, and driving efficiency under varying road conditions, loads, and thermal environments. This allocation mechanism is compatible with mainstream electric vehicle platforms, demonstrating excellent engineering adaptability and widespread adoption.

[0082] Example 2: In an exemplary embodiment, a stability control system for an electric vehicle is further provided, including a state estimation module, a yaw moment prediction module, and a motor torque distribution module.

[0083] The state estimation module is used to acquire and process vehicle driving data in real time, including basic information such as vehicle speed, steering wheel angle, yaw rate, lateral acceleration, and wheel speed. Based on the unscented Kalman filter algorithm, this module dynamically estimates key state parameters that cannot be directly measured, such as road adhesion coefficient, vehicle center of mass sideslip angle, vehicle mass, and vertical load distribution. The estimated results provide an accurate system state foundation for subsequent yaw moment prediction and torque distribution.

[0084] The yaw moment prediction module receives the current vehicle state parameters from the state estimation module and, combined with forward-looking trajectory data and the vehicle's kinematic model, predicts the vehicle's dynamic trends over a short period of time and calculates the desired yaw moment. This module also determines the vehicle's longitudinal traction requirement based on the vehicle's current driving instructions or automated driving path. Together, the yaw moment and traction requirement provide target guidance for the underlying execution systems.

[0085] The motor torque distribution module intelligently distributes total torque to the four independently driven wheels, taking into account the vehicle's current adhesion, motor capabilities, and load, while maintaining desired yaw moment and longitudinal traction requirements. This precise control of torque ensures stable vehicle posture while ensuring drive efficiency and motor safety, thereby enhancing vehicle stability and responsiveness under varying operating conditions.

[0086] A top-down data and command chain forms between these modules: the state estimation module provides foundational state support, the yaw moment prediction module sets control targets, and the motor torque distribution module executes these commands. The overall system is highly integrated and dynamically adaptable, responding to changes in vehicle driving conditions in real time and ensuring stability control for high-performance electric vehicles across a wide range of road and handling conditions.

[0087] If the above functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0088] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0089] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, and then editing, interpreting, or processing in another suitable manner as necessary, and then storing it in a computer memory.

[0090] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0091] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A stability control method for an electric vehicle, characterized in that: include: Collect vehicle driving data in real time and use the unscented Kalman filter algorithm to dynamically estimate the vehicle's state parameters; Based on the vehicle's state parameters, combined with the kinematic model and forward trajectory, the expected yaw moment in the short term is calculated, and the vehicle's longitudinal traction requirements are obtained; According to the vehicle's state parameters and longitudinal traction requirements, the yaw torque and traction force are decoupled and distributed to the four electric drive wheels to achieve stable control of the electric vehicle; The forward trajectory is a sequence of target path points within a certain time window forward with the vehicle's current moment as a reference, including multiple spatial position points and curvature information corresponding to the spatial position points; The calculation of the expected yaw moment in the short future time includes setting a fixed look-ahead time window. , calculated based on vehicle speed Within the foresight distance, perform cubic spline fitting on the foresight trajectory within the foresight distance, calculate the average curvature, and obtain the expected yaw angular velocity; Calculate the yaw rate error and sideslip angle suppression error, use the vehicle's state parameters to linearize the bicycle model in real time, generate the state matrix and input matrix, and build an adaptive linearized vehicle model; Set the prediction step size , additionally introduce the sideslip angle as the state estimation output, set the sideslip angle error as the target quantity, construct the objective function and constraints, and the objective function is expressed as: ； in, represents the controller objective function value; Indicates the prediction step length; express time; represents the weight of the yaw rate error; Indicates time The yaw rate error; Represents the weight of the sideslip angle error; Indicates time Side slip angle error; The weight representing the rate of change of torque; Indicates time The rate of change of torque; Represents the control input, time The yaw moment; The adaptive linearized vehicle model, objective function and constraints are compiled into a QP problem. The optimal yaw moment sequence is obtained in a single cycle using a real-time QP solver. The first moment solution is taken as The desired yaw moment within.

2. The stability control method of an electric vehicle according to claim 1, wherein: The vehicle driving data includes vehicle speed, steering wheel angle, yaw rate, lateral acceleration and wheel speed.

3. The stability control method of an electric vehicle according to claim 2, wherein: The estimating of the vehicle state parameters includes constructing an extended state vector based on vehicle driving data and calculating the instantaneous slip rate of each wheel; Generate a set of filters point, and dynamically adjust the process noise covariance matrix according to the current maximum slip rate; The saturation value determined by the adhesion coefficient and the vertical load is introduced, and the lateral stiffness model is enveloped by the hyperbolic tangent function to construct a lateral tire force saturation model. The lateral tire force saturation model is embedded in the state propagation function. Nonlinear prediction of points; according to The weighted mean and covariance of the predicted values of the points are calculated to obtain the prior estimation result of the state at the current moment, and the innovation term is calculated. The a priori estimation update is completed based on the unscented Kalman gain to obtain the vehicle state parameters, which include vehicle speed, sideslip angle, yaw rate and lateral adhesion coefficient.

4. The stability control method of an electric vehicle according to claim 3, wherein: The vehicle longitudinal traction demand is the target traction required by the vehicle based on driver input, and is calculated by the vehicle controller VCU.

5. The stability control method of an electric vehicle according to claim 4, wherein: Decoupling the yaw moment and traction and distributing them to the four electrically driven wheels includes establishing, based on the vehicle's state parameters and physical relationships, an equation constraint on the four-wheel torque's effect on the vehicle's longitudinal traction and yaw moment, so that the four-wheel combined force meets the longitudinal traction requirement and the four-wheel left-right distribution meets the desired yaw moment. Dynamic constraints are set for each wheel's electric drive motor, including torque upper limit constraint, adhesion limit constraint, power limit constraint, and torque change rate constraint. The adhesion limit constraint estimates the maximum traction force based on the vehicle's state parameters and the current vertical load of each wheel, and is expressed as: ； in, Indicates the Motor torque output for each wheel; Indicates the effective radius of the wheel; Indicates time The lateral adhesion coefficient; Indicates the Vertical load on each wheel; Under the premise of satisfying equality constraints and dynamic constraints, a quadratic programming optimization problem is constructed with yaw moment error, torque change rate and power loss as cost functions. The optimal torque distribution result is solved and calculated and sent to each drive motor and each wheel electric drive motor for execution.

6. A stability control system for an electric vehicle, applied to the stability control method for an electric vehicle according to any one of claims 1 to 5, characterized in that: include, The state estimation module collects vehicle driving data in real time and uses the unscented Kalman filter algorithm to dynamically estimate the vehicle's state parameters; The yaw moment prediction module calculates the expected yaw moment in the short term based on the vehicle's state parameters, combined with the kinematic model and forward trajectory, and obtains the vehicle's longitudinal traction requirements; The motor torque distribution module decouples the yaw torque and traction force and distributes them to the four electric drive wheel ends according to the vehicle's state parameters and longitudinal traction requirements, thereby achieving stable control of the electric vehicle.

Images