Controllable domain based distributed drive electric vehicle lateral stability control method

Abstract

The application discloses a kind of based on controllable field's distributed drive electric vehicle lateral stability control method.The method obtains vehicle longitudinal speed, road adhesion coefficient, mass center side angle and its change rate, and whether vehicle state is in lateral stability controllable field based on the look-up table of offline closed loop simulation data construction is judged, and differential braking or balanced torque vector distribution mode is determined.When online control, controller distributes wheel end drive torque and / or brake torque according to look-up table output result, additional yaw moment is generated, and vehicle lateral state is restored to stable range;When state is uncontrollable, stability mitigation strategy is executed;The method can be executed by vehicle controller.The application can improve the recovery ability of vehicle lateral stability, reduce online calculation burden, and reduce longitudinal speed loss.

Description

Technical Field

[0001] This invention belongs to the field of intelligent electric vehicle chassis control technology, specifically relating to a lateral stability control method for distributed drive electric vehicles based on a controllable domain, which is particularly suitable for distributed drive electric vehicles with independent adjustment capabilities for wheel-end drive torque and braking torque. Background Technology

[0002] Distributed drive electric vehicles typically employ in-wheel motors, wheel-side motors, or multi-motor drive systems, allowing for relatively independent adjustment of the driving torque of each wheel. Simultaneously, the vehicle can coordinate and control the braking torque of each wheel through an electronic braking system. Therefore, compared to traditional centralized drive vehicles, distributed drive electric vehicles possess stronger longitudinal force distribution and yaw moment adjustment capabilities. They can generate additional yaw moment through differences in the driving torque or braking torque of the left and right wheels, thereby improving the vehicle's lateral stability under conditions such as high speed, low adhesion, emergency obstacle avoidance, and sharp turns.

[0003] Existing vehicle lateral stability control methods typically assess the risk of instability based on state variables such as yaw rate, sideslip angle, and lateral acceleration, and intervene in stability through differential braking, direct yaw moment control, or torque vector control. Differential braking, in particular, can rapidly generate a large additional yaw moment, offering advantages such as fast response, simplicity, and mature engineering applications. However, its stability recovery process is usually accompanied by significant longitudinal velocity loss and braking energy consumption. Torque vector control can generate additional yaw moment by coordinating the distribution of driving torque and braking torque between the left and right wheels, reducing longitudinal velocity loss and improving control smoothness to some extent. However, its control effectiveness is limited by motor torque capability, tire adhesion conditions, and actuator constraints.

[0004] In assessing lateral stability, existing methods are mostly based on the autonomous phase plane stability domain of vehicle dynamics, such as constructing stability boundaries based on the sideslip angle and its rate of change. However, these methods typically reflect the vehicle's natural stability under no active control conditions, neglecting the impact of active yaw moment control, torque distribution modes, and actuator constraints on the vehicle's stability recovery capability. Therefore, in distributed drive electric vehicles, traditional stability domain assessment methods may be overly conservative and fail to accurately reflect the vehicle's actual recoverability under control.

[0005] Furthermore, the selection of existing differential braking distribution modes and balanced torque vector distribution modes mostly relies on fixed rules or empirical thresholds, making it difficult to adaptively adjust according to changes in vehicle speed, road surface adhesion conditions, and lateral stability. Under complex operating conditions, fixed mode selection rules may lead to two types of problems: first, over-reliance on differential braking, causing unnecessary speed loss and braking burden; second, failure to select a control mode that is more conducive to rapid and stable recovery in a timely manner under high-risk boundary conditions, affecting the vehicle's lateral stability control effect.

[0006] Therefore, it is necessary to propose a lateral stability control method that can take into account control action, actuator constraints, and stability recovery capability under different torque distribution modes, so as to improve the stability control capability and real-time control feasibility of distributed drive electric vehicles under complex operating conditions. Summary of the Invention

[0007] I. Technical problems to be solved

[0008] This invention aims to solve the following problems existing in the prior art:

[0009] First, traditional methods for judging the lateral stability of vehicles based on autonomous phase plane stability domains do not fully consider the active control effect, resulting in conservative stability domain judgments that are difficult to accurately reflect the actual stability recovery capability of distributed drive electric vehicles under control.

[0010] Second, the selection of existing differential braking distribution mode and balanced torque vector distribution mode mostly relies on fixed rules or empirical thresholds, lacking a unified selection basis based on vehicle operating status and stability recovery capability.

[0011] Third, the controllable boundaries and mode selection boundaries under complex nonlinear vehicle states are difficult to describe accurately using simple analytical rules, and direct online calculation would increase the real-time computing burden on the vehicle controller.

[0012] To address the aforementioned issues, this invention proposes a lateral stability control method for distributed drive electric vehicles based on a controllable domain. This method constructs a controllable domain for vehicle lateral stability through offline numerical simulation, utilizes a data-driven model to learn controllable boundaries and mode selection boundaries, and converts the learning results into a multi-dimensional lookup table suitable for real-time use by the onboard controller. This enables the determination of vehicle lateral stability status and adaptive selection of torque distribution modes.

[0013] II. Technical Solution

[0014] To achieve the above objectives, this invention provides a lateral stability control method for distributed drive electric vehicles based on a controllable domain. The method includes an offline modeling stage and an online control stage. The offline modeling stage generates a controllable domain for vehicle lateral stability considering control actions and actuator constraints, as well as a torque distribution mode selection boundary. The online control stage queries a lookup table based on the real-time vehicle operating status to determine the controllability of the current vehicle state and identify a recommended torque distribution mode, thereby achieving wheel-end drive torque and / or braking torque distribution.

[0015] In one implementation, the real-time operating status information of the vehicle constitutes a state vector. :

[0016] (1)

[0017] in, For the longitudinal speed of the vehicle, The road surface adhesion coefficient, The sideslip angle is the angle at the vehicle's center of gravity. Let be the rate of change of the vehicle's sideslip angle. The sideslip angle can be determined by the vehicle's longitudinal and lateral velocities as follows:

[0018] (2)

[0019] in, This refers to the vehicle's lateral velocity. In actual vehicles, , , , and It can be obtained from wheel speed sensors, inertial measurement units, yaw rate sensors, vehicle condition observers, or road surface adhesion estimators.

[0020] In the offline phase, based on the vehicle dynamics model and tire model, numerical simulations are performed under different vehicle operating conditions, including no-control mode, differential braking distribution mode, and / or balanced torque vector distribution mode, to obtain the corresponding stability recovery time, speed loss, and control cost. The preset stability neighborhood of the vehicle's lateral stability state can be represented as:

[0021] (3)

[0022] in, The centroid sideslip angle threshold. This is the threshold for the rate of change of the center of gravity sideslip angle. When the vehicle's lateral stability state enters this stable neighborhood and remains there for a continuous time not less than a preset dwell time... When the state is stable, it is determined that the state has been restored.

[0023] For any vehicle operating state In torque distribution mode The stability recovery time under these conditions can be defined as:

[0024] (4)

[0025] in, Candidate recovery time, For the time variable within the preset residence time interval, and They represent The vehicle's sideslip angle at time t and its rate of change. If the vehicle's condition is within the maximum permissible recovery time... If the target does not enter and remain within a stable neighborhood, the stability recovery time in this mode can be denoted as infinity.

[0026] Based on the stability recovery time under different torque distribution modes, controllability tags can be generated:

[0027] (5)

[0028] in, This is a set of torque distribution modes, where DBDM represents differential braking distribution mode and BTVDM represents balanced torque vector distribution mode. When... Time indicates the vehicle's operating status. This falls within the controllable range of vehicle lateral stability; when Time indicates the vehicle's operating status. The actuator constraints and maximum permissible recovery time are uncontrollable.

[0029] During the mode assignment phase, it is first determined whether the vehicle's operating state falls within the controllable domain. Only if the vehicle's operating state falls within the controllable domain is the recommended torque distribution mode determined based on the stability recovery time, vehicle speed loss, and control cost under different torque distribution modes. Vehicle speed loss can be defined as:

[0030] (6)

[0031] Control costs can be expressed as the integral of the absolute value of wheel-end power:

[0032] (7)

[0033] in, To evaluate the time interval, The initial longitudinal velocity of the vehicle, for The longitudinal speed of the vehicle at any given time, Torque distribution mode Next The wheel-end torque of each wheel, Indicates the front left, front right, rear left, and rear right wheels. For the first One wheel Angular velocity at time t.

[0034] The difference in control cost between the differential braking distribution mode and the balanced torque vector distribution mode can be defined as:

[0035] (8)

[0036] When all different torque distribution modes can restore the vehicle's lateral stability to the preset stable state range, and the stability recovery time difference between different torque distribution modes is greater than the preset time difference threshold, the torque distribution mode with the shorter stability recovery time is selected; when the stability recovery time difference between different torque distribution modes is not greater than the preset time difference threshold, the recommended torque distribution mode is determined based on the vehicle speed loss and / or control cost; when only one torque distribution mode can restore the vehicle, that torque distribution mode is selected; when all torque distribution modes cannot restore the vehicle, the preset safety backoff control strategy is executed.

[0037] In one implementation, the model is trained using vehicle operating status, controllability labels, and pattern selection labels. To enhance the expressive power for nonlinear boundaries and coupling relationships, the input features can consist of the original state variables and their interaction and quadratic terms:

[0038] (9)

[0039] The data-driven model can be a gradient boosting decision tree, random forest, support vector machine, neural network, logistic regression model, or other nonlinear classification model. In a preferred embodiment, a tree model is used to learn the controllable boundary and torque distribution mode selection boundary, and the trained model is converted into a multidimensional lookup table suitable for real-time invocation by the vehicle controller. The index of the multidimensional lookup table includes at least the vehicle's longitudinal speed, road surface adhesion information, vehicle center of gravity sideslip angle, and rate of change of center of gravity sideslip angle. The output includes at least the controllability judgment result of the current vehicle state, the recommended torque distribution mode, and / or confidence level.

[0040] Subsequently, a multi-dimensional lookup table is queried based on the real-time vehicle operating status. The multi-dimensional lookup table outputs the current state controllability assessment result and a recommended torque distribution mode. Based on the lookup table output result, the controller determines whether the current vehicle state is controllable. If the current vehicle state is uncontrollable, the controller executes a preset stability mitigation strategy. This preset stability mitigation strategy may include limiting drive torque, increasing braking intervention, reducing vehicle speed, outputting risk warnings, or requesting intervention from the upper-level driver assistance system, to reduce the risk of vehicle instability. If the current vehicle state is controllable, the wheel-end torque distribution method is further determined based on the recommended torque distribution mode.

[0041] By replacing the online inference process of the data-driven model with a lookup table, the computational burden can be reduced, the determinism of execution time can be improved, and the real-time operation requirements of the vehicle-mounted embedded controller can be adapted.

[0042] During the online control phase, the upper-level yaw moment controller calculates the desired additional yaw moment based on the vehicle's yaw rate error and center of gravity sideslip angle error. In one embodiment, the yaw moment controller can employ sliding mode control, with its sliding surface... Represented as:

[0043] (10)

[0044] in, For yaw rate error, This is the error in the centroid sideslip angle. These are weighting coefficients. The expected additional yaw moment. It can be represented as:

[0045] (11)

[0046] in, For equivalent control items, To switch control terms. It should be noted that the above-described yaw moment controller is only a preferred implementation. The present invention can also employ model predictive control, robust control, proportional-integral-derivative control, or other control methods capable of generating the desired additional yaw moment.

[0047] When the recommended torque distribution mode is differential braking distribution mode, the additional yaw moment is determined according to the desired torque distribution mode. The direction of braking torque is determined by whether the vehicle's left or right wheels are subjected to braking torque. The total braking torque required is... It can be represented as:

[0048] (12)

[0049] in, The radius of the wheel's rolling motion. and These are the front track and the rear track, respectively.

[0050] For the front and rear wheels on the selected braking side, the braking torque can be distributed according to the vertical load ratio of the corresponding wheels. For example, when braking the right wheel, it can be expressed as:

[0051] (13)

[0052] (14)

[0053] in, and These represent the wheel-end torques of the right front wheel and the right rear wheel, respectively, with positive for driving and negative for braking; and These are the vertical loads of the right front wheel and the right rear wheel, respectively; when the left wheel is selected for braking, the braking torque is symmetrically distributed according to the ratio of the vertical loads of the left front and rear wheels.

[0054] The wheel end torque of each wheel satisfies the tire adhesion constraint:

[0055] (15)

[0056] in, For the first The wheel-end torque of each wheel, For the first Vertical load on each wheel.

[0057] In equation (12) This can be considered as the equivalent track width for estimating differential braking yaw moment. When the vehicle's front track width... Rear wheel track When the difference is small or a uniform equivalent wheel track is used in the controller calibration, the required total braking torque can be calculated using formula (12); when the difference between the front and rear wheel tracks is significant or a more precise distribution is required, the braking torque distribution ratio of the front and rear wheels can be corrected according to the actual contribution of the front and rear wheels to the yaw moment.

[0058] When the recommended torque distribution mode is the balanced torque vector distribution mode, the desired additional yaw moment will be distributed according to the front and rear axle distribution coefficients. Decomposed into front axle yaw moment requirements and rear axle yaw moment requirements :

[0059] (16)

[0060] (17)

[0061] Then, based on the front axle yaw moment requirements and the rear axle yaw moment requirements, determine the torque difference requirements for the left and right wheels of the front axle respectively. Torque difference requirements between the left and right rear wheels :

[0062] (18)

[0063] (19)

[0064] For each axle, driving torque and braking torque in opposite directions are distributed between the left and right wheels to create the desired torque difference between the left and right wheels. Driving torque and braking torque It can be determined in the following way:

[0065] (20)

[0066] (twenty one)

[0067] in, Indicates front axle or rear axle. The maximum permissible drive torque for a single drive motor. This represents the maximum permissible braking torque for the corresponding axle. Based on the direction of the desired additional yaw moment, the driving torque or braking torque applied to the left and right wheels of each axle is determined, thereby generating the desired additional yaw moment while satisfying the tire adhesion constraint, the maximum driving torque constraint of the motor, and the maximum braking torque constraint of the brake.

[0068] III. Beneficial Effects

[0069] Compared with the prior art, the present invention has the following beneficial effects.

[0070] First, this invention constructs a controllable domain for vehicle lateral stability based on numerical simulation results, no longer relying solely on the autonomous stability domain of vehicle dynamics. This allows for a more realistic reflection of the vehicle's actual stability recovery capability under active yaw moment control and actuator constraints, thereby reducing the conservatism of traditional stability domain judgments.

[0071] Second, this invention incorporates the differential braking distribution mode and the balanced torque vector distribution mode into a unified controllable domain analysis framework. It performs mode attribution based on indicators such as stability recovery time, vehicle speed loss, and control cost, and can adaptively select a more suitable torque distribution mode according to the vehicle state.

[0072] Third, this invention utilizes a data-driven model to learn the controllable boundary of vehicle lateral stability and the torque distribution mode selection boundary, which can describe the controllability and mode selection relationship under complex nonlinear vehicle conditions, and improve the problem of insufficient adaptability of fixed thresholds or empirical rules.

[0073] Fourth, this invention converts the trained data-driven model into a multi-dimensional lookup table. During the online control phase, only a table lookup is needed to obtain the controllability judgment result and the recommended torque distribution mode, which helps to reduce the real-time calculation burden of the vehicle controller and improve the feasibility of engineering deployment.

[0074] Fifth, this invention comprehensively considers stability recovery speed, speed loss and control cost during the vehicle's lateral stability recovery process, enabling the vehicle to quickly recover stability under high-risk conditions and reduce unnecessary braking intervention under normal recoverable conditions, thereby balancing lateral stability, longitudinal speed maintenance and control smoothness. Attached Figure Description

[0075] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments are briefly described below. It should be understood that the following drawings are only used to illustrate the technical concept of the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0076] Figure 1 This is a flowchart illustrating the overall process of the lateral stability control method for distributed drive electric vehicles based on a controllable domain, as described in this invention.

[0077] Figure 2 This is a flowchart of the offline controllable domain construction process of the present invention;

[0078] Figure 3 This is a flowchart illustrating the torque distribution mode attribution process of the present invention.

[0079] Figure 4 This is a flowchart of the data-driven model training and multidimensional lookup table generation process of the present invention.

[0080] Figure 5 This is a flowchart of the online lateral stability control process of the present invention;

[0081] Figure 6 This is a schematic diagram of wheel-end torque distribution under the differential braking distribution mode of the present invention;

[0082] Figure 7 This is a schematic diagram of wheel-end torque distribution under the balanced torque vector distribution mode of the present invention;

[0083] Figure 8 This is a schematic diagram of the vehicle-mounted controller structure for which the control method of the present invention is applied.

[0084] Figure 9 is a schematic diagram comparing the controllable domains of the no-control mode, differential braking distribution mode, and balanced torque vector distribution mode in an application example of the present invention.

[0085] Figure 10This is a schematic diagram comparing the vehicle center of gravity sideslip angle recovery response under different torque distribution modes in an application example of the present invention;

[0086] Figure 11 This is a schematic diagram comparing the longitudinal speed loss of the vehicle under different torque distribution modes in an application example of the present invention;

[0087] Figure 12 This is a schematic diagram of the timing of adaptive torque distribution mode switching in an application example of the present invention. Detailed Implementation

[0088] The technical solution of the present invention will be further described below with reference to the accompanying drawings. It should be understood that the following embodiments are only used to illustrate the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, equivalent substitutions or modifications made to the vehicle dynamics model, tire model, data-driven model, lookup table dimension, controller form, and torque distribution rules without departing from the concept of the present invention should all fall within the scope of protection of the present invention.

[0089] Example 1: A Lateral Stability Control Method Based on Controllable Domain

[0090] like Figure 1 As shown, the control method in this embodiment includes an offline learning stage and an online control stage. In the offline learning stage, a vehicle dynamics model and a tire model are first established, and closed-loop simulations are performed under different torque distribution modes, including differential braking distribution mode (DBDM), no-control mode (NC), and balanced torque vector distribution mode (BTVDM). Then, stability recovery results under different vehicle operating states are obtained based on the closed-loop simulation results, and the controllable domain of vehicle lateral stability is constructed accordingly. Further, mode attribution is performed based on the stability recovery effects under different torque distribution modes to generate a mode selection dataset; then, the data-driven model is trained using the mode selection dataset to obtain the controllable boundary and mode selection boundary, and the training results are converted into a multi-dimensional lookup table.

[0091] During the online control phase, the onboard controller acquires vehicle status information in real time according to a preset control cycle. It then queries the multidimensional lookup table based on the vehicle's longitudinal speed, road surface adhesion coefficient, vehicle sideslip angle, and rate of change of the vehicle sideslip angle to obtain the recommended torque distribution mode corresponding to the current vehicle state. When the recommended torque distribution mode is Differential Braking Distribution Mode (DBDM), the onboard controller determines the braking side or braking wheel based on the desired yaw moment and calculates the corresponding wheel-end braking torque. When the recommended torque distribution mode is Balanced Torque Vector Distribution Mode (BTVDM), the onboard controller coordinates the distribution of driving torque and / or braking torque between the left and right wheels to reduce longitudinal speed loss while satisfying tire adhesion constraints and motor torque constraints. Subsequently, the onboard controller outputs wheel-end torque commands, which are then implemented by the actuators to achieve corresponding torque control, thereby realizing vehicle lateral stability control.

[0092] For vehicle operating state samples determined to be controllable during the offline learning phase, the stability recovery time under Differential Braking Distribution Mode (DBDM) and Balanced Torque Vector Distribution Mode (BTVDM) is compared. When the difference in stability recovery time between the two torque distribution modes is greater than a preset time difference threshold, the torque distribution mode with the shorter stability recovery time is selected as the recommended torque distribution mode. When the difference in stability recovery time is not greater than the preset time difference threshold, vehicle speed loss and control cost are further compared, and the torque distribution mode with smaller vehicle speed loss and / or smaller control cost is selected as the recommended torque distribution mode.

[0093] 1. Acquisition of vehicle operating status information

[0094] This invention first acquires vehicle operating status information. This vehicle operating status information includes at least vehicle speed information, road surface adhesion coefficient information, and vehicle lateral stability status information.

[0095] In this embodiment, the vehicle controller acquires vehicle speed information, road surface adhesion information, and vehicle lateral stability status information according to a preset control cycle. The vehicle speed information may include the vehicle's longitudinal speed, and may further include the vehicle's lateral speed, wheel speed, etc.; the road surface adhesion information may include the road surface adhesion coefficient, road surface adhesion level, or road surface state information obtained by a road surface recognition algorithm; the vehicle lateral stability status information may include the vehicle's sideslip angle, the rate of change of the sideslip angle, yaw rate, lateral acceleration, or a combination thereof.

[0096] In this embodiment, the vehicle yaw rate can be directly obtained from the onboard sensors, while the longitudinal velocity, center of gravity sideslip angle, and rate of change of center of gravity sideslip angle can be obtained from the vehicle state estimation algorithm. The road surface adhesion information can be obtained from the road surface adhesion estimation algorithm, the tire force estimation algorithm, or the multi-sensor fusion method. After obtaining the vehicle operating state information, the vehicle operating state vector is constructed according to equation (1). The vehicle operating state vector serves as the input for subsequent controllable domain judgment, data-driven model training, and multi-dimensional lookup table query. When it is necessary to determine the vehicle center of gravity sideslip angle from the vehicle's longitudinal velocity and lateral velocity, the vehicle center of gravity sideslip angle is calculated according to equation (2).

[0097] 2. Offline Controllable Domain Construction

[0098] like Figure 2 As shown, this embodiment constructs a closed-loop controllable domain for vehicle lateral stability during the offline phase.

[0099] Specifically, a vehicle dynamics model is first established. This model describes the effects of longitudinal, lateral, and yaw motions, as well as wheel-end torque, on the vehicle's motion state. The included tire model describes the relationship between road surface adhesion conditions, tire load, and the longitudinal and lateral forces of the tires. This vehicle dynamics model can be calibrated or replaced according to different vehicle platforms.

[0100] Then, controller constraints and actuator constraints are set. The actuator constraints include at least drive motor torque constraints, braking capacity constraints, and tire adhesion constraints. By introducing these constraints, the constructed closed-loop controllable domain can reflect the vehicle's lateral stability recovery capability within the actual actuator capability range.

[0101] Subsequently, a torque distribution mode was selected, and closed-loop simulations were performed for each vehicle operating state sample. The vehicle operating state samples were obtained by sampling from a pre-defined operating state space, which includes at least the vehicle's longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and rate of change of the vehicle's center of gravity sideslip angle. The torque distribution modes included at least three modes: no-control mode (NC), differential braking distribution mode (DBDM), and balanced torque vector distribution mode (BTVDM). For each torque distribution mode, closed-loop simulations were performed on each vehicle operating state sample to obtain the corresponding stability recovery results.

[0102] Furthermore, based on the stability recovery results, it is determined whether the vehicle state meets the stability recovery conditions. Specifically, if the vehicle's center of gravity sideslip angle and its rate of change can enter a preset stable neighborhood within the maximum allowable recovery time and remain within the preset stable neighborhood for a preset dwell time, then the vehicle operating state sample is determined to meet the stability recovery conditions and is marked as a controllable state; otherwise, it is marked as an uncontrollable state.

[0103] For each vehicle operating state sample, the preset stable neighborhood is first determined according to equation (3), then the stability recovery time under the corresponding torque distribution mode is calculated according to equation (4), and the controllability label of the vehicle operating state sample is generated according to equation (5). After all vehicle operating state samples have been traversed, the closed-loop controllable domain corresponding to the current torque distribution mode is constructed based on the controllable state samples and the uncontrollable state samples.

[0104] After traversing all torque distribution modes, a closed-loop controllable domain dataset is output. This closed-loop controllable domain characterizes whether the vehicle can recover lateral stability under given operating conditions, control actions, and actuator constraints. Unlike traditional autonomous stability domains based on vehicle dynamics, the closed-loop controllable domain in this embodiment considers active yaw moment control and actuator capability limitations, thus more realistically reflecting the lateral stability recovery capability of distributed drive electric vehicles under actual control conditions.

[0105] 3. Torque Distribution Pattern Attribution

[0106] like Figure 3 As shown, after obtaining the stability recovery results of the closed-loop controllable domain and each torque distribution mode, this embodiment further performs torque distribution mode attribution on the controllable state samples to generate a mode selection dataset.

[0107] Specifically, the current controllable state sample is first read, and evaluation indicators are extracted under the Differential Braking Distribution Mode (DBDM) and Balanced Torque Vector Distribution Mode (BTVDM). These evaluation indicators include at least stability recovery time, vehicle longitudinal speed loss, and control cost. Stability recovery time characterizes the time required for the vehicle to recover from its current state to a stable state; vehicle longitudinal speed loss characterizes the impact of control intervention on the vehicle's longitudinal driving performance; and control cost characterizes the intensity of wheel-end drive torque and / or braking torque usage, actuator workload, or control energy consumption.

[0108] In this embodiment, the vehicle longitudinal speed loss is calculated according to equation (6), the control cost is calculated according to equation (7), and the difference in control cost between the differential braking distribution mode (DBDM) and the balanced torque vector distribution mode (BTVDM) is calculated according to equation (8). Thus, the mode selection label not only reflects whether the vehicle can recover stability, but also reflects the differences in recovery speed, longitudinal speed maintenance capability, and actuator usage intensity of different torque distribution modes.

[0109] For a sample in a currently controllable state, first determine whether both the differential braking distribution mode (DBDM) and the balanced torque vector distribution mode (BTVDM) can satisfy the stability recovery condition. If only one torque distribution mode can restore the vehicle to stability, then select this unique controllable torque distribution mode as the recommended torque distribution mode for the current sample and record the recommended label; if neither torque distribution mode can satisfy the stability recovery condition, then mark the current sample as an unattributable sample or retain an empty label.

[0110] If both torque distribution modes can restore vehicle stability, the stability recovery time of the two modes is further compared. When the difference in stability recovery time is significant, the torque distribution mode with the shorter recovery time is selected as the recommended torque distribution mode for the current sample, and a recommendation label is recorded. When the difference in stability recovery time is not significant, the longitudinal speed loss and control cost of the vehicle are compared, and either the differential braking distribution mode (DBDM) or the balanced torque vector distribution mode (BTVDM) is selected as the recommended torque distribution mode based on the overall cost.

[0111] The process of traversing all controllable state samples in the manner described above continues until all samples have been attributed, generating a torque distribution mode selection dataset and outputting the mode attribution results. This mode selection dataset characterizes the correspondence between vehicle operating states and recommended torque distribution modes, providing a sample foundation for subsequent data-driven model training and multidimensional lookup table construction.

[0112] 4. Data-driven model training and multidimensional lookup table generation

[0113] like Figure 4 As shown, this embodiment uses vehicle operating status information, closed-loop controllable domain information, and mode selection dataset to train a data-driven model, and further generates a multi-dimensional lookup table suitable for real-time invocation by the vehicle controller.

[0114] Specifically, the sample data is first read and organized. This sample data includes vehicle operating status information, closed-loop controllable domain information, and a mode selection dataset. The vehicle operating status information includes at least the vehicle's longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and rate of change of the vehicle center of gravity sideslip angle. The closed-loop controllable domain information characterizes whether the current vehicle operating state can be restored to a stable state. The mode selection dataset characterizes the recommended torque distribution mode corresponding to the current vehicle operating state.

[0115] Then, input features and output labels are extracted. The input features include at least vehicle longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and vehicle center of gravity sideslip angle change rate; the output labels include at least controllability judgment results and recommended torque distribution modes. In this embodiment, the input features of the data-driven model are constructed according to equation (9). By introducing interaction terms and quadratic terms based on vehicle longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and vehicle center of gravity sideslip angle change rate, the data-driven model's ability to express nonlinear controllable boundaries and mode selection boundaries can be improved.

[0116] Subsequently, the sample data undergoes preprocessing. This preprocessing includes at least one of the following: data cleaning, outlier removal, normalization, standardization, category labeling, and sample balancing. After preprocessing, the sample data is divided into a training set and a validation set.

[0117] Next, a data-driven model is trained using the training set to learn the mapping relationship between vehicle operating status and controllability judgment results, as well as the mapping relationship between vehicle operating status and recommended torque distribution mode. The data-driven model can employ a decision tree model, ensemble learning model, neural network model, or other classification model.

[0118] After model training, the classification performance is verified using a validation set. If the model performance does not meet the preset requirements, the model parameters are adjusted or the sample data is reprocessed, and the model is retrained. When the model performance meets the preset requirements, offline inference is performed on a discrete state grid. The discrete state grid consists of vehicle longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and vehicle center of gravity sideslip angle change rate. By traversing the state combinations in the discrete state grid, the trained data drives the model output results and stores them as a multidimensional lookup table.

[0119] The multidimensional lookup table index includes at least the vehicle's longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and rate of change of vehicle center of gravity sideslip angle; the output of the multidimensional lookup table includes at least the current state controllability judgment result and the recommended torque distribution mode. By converting the data-driven model into a multidimensional lookup table, the online control phase does not need to directly run the complex data-driven model, but only needs to perform table lookups and simple logical judgments, thereby reducing the real-time computational burden on the vehicle controller.

[0120] 5. Online lateral stability control

[0121] like Figure 5 As shown, this embodiment achieves vehicle lateral stability control during the online control phase.

[0122] Specifically, the onboard controller acquires vehicle status information in real time according to a preset control cycle. This vehicle status information includes at least the vehicle's longitudinal speed, road surface adhesion information, vehicle sideslip angle, and rate of change of the vehicle's sideslip angle. Subsequently, the onboard controller acquires or calculates the desired yaw moment, which can be calculated by the yaw moment controller based on the deviation between the vehicle's actual lateral stability state and the desired lateral stability state.

[0123] In this embodiment, the vehicle controller constructs a sliding surface according to equation (10) based on the vehicle yaw rate error and the vehicle center of gravity sideslip angle error, and calculates the desired additional yaw moment according to equation (11). It should be noted that equations (10) and (11) are only one implementation method for calculating the desired additional yaw moment. Without changing the control logic of this invention, the desired additional yaw moment can also be calculated by model predictive control, robust control, or proportional-integral-derivative control.

[0124] Then, the on-board controller queries a multidimensional lookup table and determines the controllability of the current vehicle state based on the lookup table output, while simultaneously determining the recommended torque distribution mode. The output of the multidimensional lookup table includes at least the current state controllability judgment result and the recommended torque distribution mode.

[0125] When the current vehicle state is uncontrollable, the onboard controller executes a stability mitigation strategy. This strategy includes at least one of the following: limiting drive torque, increasing braking intervention, reducing vehicle speed, outputting risk warnings, or requesting intervention from a higher-level driver assistance system, to reduce the risk of vehicle instability and implement stability protection control.

[0126] When the current vehicle state is controllable, the onboard controller determines the wheel-end torque distribution method according to the recommended torque distribution mode. If the recommended mode is Differential Braking Distribution Mode (DBDM), the braking side or braking wheel is determined according to the desired yaw moment direction, and the braking torque is determined in combination with wheel load, road adhesion conditions, and braking capacity constraints. If the recommended mode is Balanced Torque Vector Distribution Mode (BTVDM), the driving torque and braking torque of the left and right wheels are coordinated and distributed according to the desired yaw moment direction, and the longitudinal speed loss of the vehicle is reduced while satisfying tire adhesion constraints and motor torque constraints.

[0127] Subsequently, the on-board controller generates wheel-end torque commands and outputs them to the corresponding actuators; the actuators perform wheel-end torque control, causing the vehicle to generate additional yaw moment for lateral stability recovery, ultimately achieving lateral stability control of the vehicle.

[0128] 6. Differential Braking Distribution Mode

[0129] like Figure 6As shown, Differential Brake Distribution Mode (DBDM) applies braking torque to one side of the vehicle's wheels, creating a longitudinal force difference between the left and right wheels, thereby generating an additional yaw moment. In the figure, the downward hollow arrows indicate the braking torque applied to the corresponding wheels, and the arc-shaped arrows near the vehicle's center of gravity indicate the direction of the yaw moment generated by this differential braking action.

[0130] Specifically, Figure 6 The left side illustrates applying braking torque to the right wheel of the vehicle, thereby generating an additional yaw moment in the clockwise direction; Figure 6 The right side illustrates applying braking torque to the left wheel of the vehicle, thereby generating an additional counterclockwise yaw moment. In actual control, the direction of the braking force on the vehicle side is determined based on the desired additional yaw moment direction, the vehicle coordinate system definition, the convention for the positive direction of the yaw moment, and the controller calibration rules.

[0131] In this embodiment, the vehicle controller first determines the side of the vehicle that needs braking based on the direction of the desired additional yaw moment, and calculates the required total braking torque according to equation (12). Subsequently, the controller, taking into account the vertical load of the front and rear wheels on that side, the road surface adhesion conditions, and the braking capacity constraints, distributes the required total braking torque to the front and rear wheels of the selected braking side according to equations (13) and (14). At the same time, it checks whether the wheel end torque of each wheel meets the tire adhesion constraints according to equation (15).

[0132] During torque distribution, wheels with higher loads or greater adhesion margins can receive larger braking torques; when a wheel's adhesion margin is small or close to its slip limit, its braking torque should be limited to avoid excessive wheel slippage. Through this method, the differential braking distribution mode can quickly create a longitudinal force difference between the left and right wheels, generating an additional yaw moment for lateral stability recovery. This is suitable for situations where the vehicle is in a high-risk boundary state and needs to quickly restore stability.

[0133] It should be noted that, Figure 6 This is only used to illustrate the wheel-end torque distribution logic in differential braking distribution mode, and does not limit the specific braking vehicle side, braking torque magnitude, or torque distribution ratio between the front and rear wheels.

[0134] 7. Balanced Torque Vector Distribution Mode

[0135] like Figure 7 As shown, Balanced Torque Vector Distribution Mode (BTVDM) generates additional yaw moment through the coordinated distribution of driving torque and braking torque between the left and right wheels. In the figure, the upward hollow arrows indicate the application of driving torque to the corresponding wheels, the downward hollow arrows indicate the application of braking torque to the corresponding wheels, and the arc-shaped arrows near the vehicle's center of gravity indicate the direction of the yaw moment generated by the torque difference between the left and right wheel ends.

[0136] Specifically, Figure 7The left side illustrates applying driving torque to the left wheel of the vehicle and braking torque to the right wheel, thereby generating an additional yaw moment in the clockwise direction. Figure 7 The right side illustrates applying driving torque to the right wheel and braking torque to the left wheel, thereby generating an additional counter-clockwise yaw moment. In actual control, the driving and braking sides are determined based on the desired direction of the additional yaw moment, the definition of the vehicle coordinate system, the convention for the positive direction of the yaw moment, and the controller calibration rules.

[0137] In this embodiment, after determining the desired additional yaw moment direction, the vehicle controller designates one wheel as the driving side and the other wheel as the braking side. When the desired additional yaw moment direction changes, the driving side and braking side switch accordingly. The driving side wheels are allocated driving torque, and the braking side wheels are allocated braking torque. The longitudinal force difference between the two wheels together forms the additional yaw moment used for restoring the vehicle's lateral stability.

[0138] When distributing torque, the vehicle controller decomposes the desired additional yaw moment into the front axle yaw moment requirement and the rear axle yaw moment requirement according to equations (16) and (17), and calculates the torque difference requirement between the left and right wheels of the front and rear axles according to equations (18) and (19). Then, it determines the driving torque and braking torque of each wheel according to equations (20) and (21), so that the vehicle generates an additional yaw moment for lateral stability recovery under the conditions of satisfying the tire adhesion constraint, the maximum driving torque constraint of the motor and the maximum braking torque constraint of the brake.

[0139] Compared to the differential braking distribution mode, the balanced torque vector distribution mode does not rely solely on braking torque to generate yaw moment, but rather utilizes the synergistic effect of driving torque and braking torque. Therefore, while meeting stability recovery requirements, this mode can reduce the change in net longitudinal force of the vehicle, which helps to reduce longitudinal speed loss and improve control smoothness.

[0140] During torque distribution, the on-board controller simultaneously considers tire adhesion constraints, motor torque constraints, and braking capacity constraints. When the drive motor torque capacity is insufficient, it can be compensated for by braking torque; when the tire adhesion margin is insufficient, the driving torque or braking torque of the corresponding wheel should be limited.

[0141] It should be noted that, Figure 7 This is only used to illustrate the wheel-end torque distribution relationship in the balanced torque vector distribution mode, and does not limit the specific left and right torque symbols, drive / brake side selection rules, or torque distribution ratio.

[0142] Example 2: On-board controller structure for implementing this method

[0143] like Figure 8As shown, this embodiment provides an on-board controller structure for executing the method described in Embodiment 1. The on-board controller includes a vehicle state acquisition module, a multidimensional lookup table, a desired dynamic response model, a lateral stability controller, and a torque distribution controller. The controlled object is a distributed drive electric vehicle.

[0144] The vehicle state acquisition module obtains vehicle speed information, road surface adhesion information, and vehicle lateral stability state information through state estimation and sensor detection. A multidimensional lookup table outputs the controllability judgment result of the current vehicle state and a recommended torque distribution mode based on the vehicle speed information, road surface adhesion information, and vehicle lateral stability state information. The recommended torque distribution modes include no-control mode (NC), differential braking distribution mode (DBDM), and balanced torque vector distribution mode (BTVDM).

[0145] The torque distribution controller distributes torque according to the recommended torque distribution mode output by the multidimensional lookup table. When the recommended torque distribution mode is differential braking distribution mode (DBDM), the torque distribution controller calculates the braking torque of the same side wheel according to equations (13) and (14); when the recommended torque distribution mode is balanced torque vector distribution mode (BTVDM), the torque distribution controller calculates the driving torque and / or braking torque of the left and right wheels on the same axle according to equations (20) and (21); when the recommended torque distribution mode is no-control mode (NC), the torque distribution controller does not change the wheel torque.

[0146] In this embodiment, the desired dynamic response model adopts a linear two-degree-of-freedom vehicle dynamics model, which calculates the desired lateral stability index based on the vehicle's front wheel steering angle input and vehicle speed. The deviation between the actual value and the desired value of the lateral stability index is used as the input of the lateral stability controller, which calculates the desired additional yaw moment based on the deviation. As a specific implementation, the on-board controller constructs a sliding surface according to equation (10) based on the vehicle's yaw rate error and the vehicle's center of gravity sideslip angle error, and calculates the desired additional yaw moment according to equation (11). It should be noted that equations (10) and (11) are only one implementation method for calculating the desired additional yaw moment. Without changing the control logic of this invention, model predictive control, robust control, or proportional-integral-derivative control can also be used to calculate the desired additional yaw moment.

[0147] The torque distribution controller outputs the calculated braking torque and / or drive torque to the actuators of the distributed drive electric vehicle. These actuators include hydraulic brakes and / or motor controllers to generate corresponding braking torque and / or drive torque. In this embodiment, the torque distribution controller also includes a safety mitigation module. This module generates a stability mitigation command when the multidimensional lookup table outputs that the current vehicle state is uncontrollable. The stability mitigation command includes at least one of limiting drive torque, increasing braking intervention, reducing vehicle speed, outputting a risk warning, or requesting intervention from a higher-level driver assistance system. This stability mitigation command is also output by the torque distribution controller to the actuators of the distributed drive electric vehicle.

[0148] Example 3: Real-time Implementation in Vehicle Controller

[0149] In this embodiment, the multidimensional lookup table is pre-stored in the vehicle controller. The vehicle controller can be a vehicle controller, a chassis domain controller, or a lateral stability controller.

[0150] During vehicle operation, the onboard controller reads the vehicle speed, road surface adhesion, and lateral stability status according to a preset control cycle, and queries a multidimensional lookup table. If the real-time vehicle status lies between the grid points of the lookup table, the lookup table output result can be obtained using nearest neighbor indexing, linear interpolation, or other interpolation methods.

[0151] The lookup table output includes the controllability assessment result of the current vehicle state and the recommended torque distribution mode. When the output is in a controllable state, the on-board controller distributes the wheel-end torque according to the recommended torque distribution mode; when the output is in an uncontrollable state, the on-board controller can enter a conservative stability mitigation strategy.

[0152] Since the online phase mainly involves reading state variables, calling lookup tables, and making logical judgments, avoiding online inference of complex data-driven models, this method is suitable for deployment in resource-constrained vehicle controllers.

[0153] Example 4: Application Cases and Simulation Results

[0154] To further illustrate the implementation effects of the present invention, this embodiment provides an application example under high-speed, low-adhesion conditions. It should be noted that the simulation results in this embodiment are used to illustrate the technical effects and implementation methods of the present invention, and are not intended to limit the vehicle model, controller type, parameter values, or specific application scenarios of the present invention.

[0155] In this embodiment, the vehicle is a distributed drive electric vehicle with four-wheel independent drive and independent braking capabilities. A representative high-speed, low-adhesion condition is selected, with the initial longitudinal speed of the vehicle being 105 km / h and the road surface adhesion coefficient being 0.3. In the offline stage, the vehicle's longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and center of gravity sideslip angle change rate are used as lookup table index variables. Numerical simulations are performed in the no-control mode, differential braking distribution mode, and balanced torque vector distribution mode, respectively, and the controllable domain of the vehicle's lateral stability is determined according to equations (3) to (5).

[0156] like Figure 9 As shown, under the same vehicle speed and road adhesion conditions, there are significant differences in the controllable domain of vehicle lateral stability corresponding to different torque distribution modes. In the figure, the horizontal axis represents the vehicle's sideslip angle β, and the vertical axis represents the rate of change of the vehicle's sideslip angle β̇; the dotted line represents the controllable domain boundary under the no-control mode (NC), the dashed line represents the controllable domain boundary under the differential braking distribution mode (DBDM), and the solid line represents the controllable domain boundary under the balanced torque vector distribution mode (BTVDM).

[0157] Depend on Figure 9 It is evident that the recoverable region corresponding to the no-control mode (NC) is relatively small, mainly concentrated in areas where both the sideslip angle and the rate of change of the sideslip angle are small. The differential brake distribution mode (DBDM) can significantly expand the controllable region of vehicle lateral stability. The controllable region corresponding to the balanced torque vector distribution mode (BTVDM) is further expanded, covering a wider range. These results indicate that after introducing active yaw moment control, the vehicle can recover from a wider range of lateral instability states to a stable state.

[0158] Among them, NC represents the no-control mode or the state of exiting active stability intervention. It is mainly used as a benchmark in the construction of controllable domain and simulation analysis, or as the state of exiting intervention under stable conditions. It does not belong to the active torque distribution mode. DBDM and BTVDM represent differential braking distribution mode and balanced torque vector distribution mode, respectively. Their corresponding controllable domains reflect the stability recovery capability under different actuator coordination methods.

[0159] Therefore, the controllable domain constructed in this invention can reflect the actual stability recovery capability under the combined influence of active yaw moment control and actuator constraints. Compared with judgment methods based solely on autonomous stability boundaries, the controllable domain of this invention is more suitable for active stability control decisions in distributed drive electric vehicles.

[0160] like Figure 10As shown, during the lateral stability recovery process, the fixed differential brake distribution mode (DBDM), the fixed balanced torque vector distribution mode (BTVDM), and the adaptive torque distribution mode (ADM) of this invention can all gradually converge the vehicle's center of gravity sideslip angle deviation to a preset stable range. In contrast, in the no-control mode (NC), the vehicle's center of gravity sideslip angle deviation continues to increase and deviates from the stable range for a long time, indicating that the vehicle has experienced lateral instability.

[0161] The method of this invention performs online control using a recommended torque distribution mode output by a multi-dimensional lookup table, and adaptively selects the appropriate torque distribution method based on the current state of the vehicle. Figure 10 It is evident that ADM can ensure the recovery of vehicle lateral stability while avoiding the long-term use of a single forced braking mode, thus balancing stability recovery effectiveness and control coordination.

[0162] like Figure 11 As shown, there are significant differences in the longitudinal speed changes of the vehicle under different control modes. In the no-control (NC) mode, the vehicle experiences lateral instability, and the longitudinal speed continues to drop rapidly, indicating that the vehicle's motion state has deteriorated significantly. The fixed differential brake distribution mode (DBDM) provides strong lateral stability recovery capability, but because it mainly relies on the braking torque to generate additional yaw moment, its longitudinal speed loss is relatively large. The fixed balanced torque vector distribution mode (BTVDM) generates additional yaw moment through the synergistic effect of the driving torque and braking torque, thus achieving better longitudinal speed maintenance and smaller speed loss.

[0163] The Adaptive Torque Distribution Mode (ADM) of this invention switches online based on the recommended torque distribution mode output by a multidimensional lookup table. When the vehicle is in a high-risk state and requires rapid recovery of lateral stability, ADM can select the Differential Braking Distribution Mode (DBDM) to improve the stability recovery speed; when the stability recovery time is close or the risk of vehicle instability decreases, ADM can switch to the Balanced Torque Vector Distribution Mode (BTVDM) to reduce longitudinal speed loss. Figure 11 It is evident that ADM, while ensuring the recovery of vehicle lateral stability, has better longitudinal speed maintenance capability compared to fixed DBDM.

[0164] like Figure 12 As shown, the method of this invention does not use a fixed torque distribution mode, but rather adaptively switches the torque distribution mode based on the controllability judgment result and mode selection result in the multidimensional lookup table of the vehicle's current operating state. In the figure, the vertical axis represents the torque distribution mode, the horizontal axis represents time, and the stepped curve represents the mode used at different times during the online control process.

[0165] Specifically, during the vehicle's lateral stability recovery process, the controller switches between no-control mode (NC), differential brake distribution mode (DBDM), and balanced torque vector distribution mode (BTVDM) based on the current vehicle state, the desired additional yaw moment requirement, and the multidimensional lookup table output. When the vehicle state requires active stability intervention and is suitable for differential braking, the controller selects differential brake distribution mode (DBDM) to achieve faster lateral stability recovery. When the vehicle state is suitable for a drive / brake coordination distribution mode, the controller selects balanced torque vector distribution mode (BTVDM) to reduce unnecessary braking intervention and longitudinal speed loss. When the vehicle state enters the stable region or does not require active intervention, the controller selects no-control mode (NC) to reduce control action and lower control costs.

[0166] Depend on Figure 12 As can be seen, the method of the present invention can dynamically switch between different torque distribution modes according to changes in vehicle state, thereby taking into account lateral stability recovery speed, longitudinal speed maintenance capability and control cost.

[0167] The above application examples demonstrate that this invention can achieve lateral stability recovery of vehicles under typical hazardous conditions such as high speed and low adhesion, and can adaptively select between differential braking distribution mode and balanced torque vector distribution mode through a controllable domain and mode selection lookup table. Compared with a fixed single torque distribution mode, this invention can balance stability recovery speed, vehicle longitudinal speed maintenance capability, and real-time implementation requirements of the on-board controller.

[0168] This invention can be applied to distributed drive electric vehicles, hub motor drive electric vehicles, four-wheel independent drive electric vehicles, and intelligent vehicle chassis systems with independent drive and braking capabilities. The method learns the controllable domain of vehicle lateral stability and the torque distribution mode selection boundary offline, and converts the learning results into a lookup table that can be called in real time by the onboard controller. This meets the real-time, reliability, and engineering deployment requirements of vehicle active safety control, and has significant industrial application value.

[0169] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention.

[0170] For those skilled in the art, without departing from the technical concept of this invention, equivalent substitutions or modifications can be made to the vehicle dynamics model, tire model, data-driven model type, lookup table dimension, pattern attribution index, torque distribution method, and actuator configuration, and all such substitutions or modifications should fall within the protection scope of this invention.

Claims

1. A lateral stability control method for distributed drive electric vehicles based on controllable domains, characterized in that, Includes the following steps: Obtain real-time vehicle operating status information, which constitutes a state vector. And at least the vehicle's longitudinal speed Road surface adhesion coefficient Vehicle center of gravity sideslip angle and the rate of change of the vehicle's center of gravity sideslip angle : ； Based on the real-time vehicle operating status information, a pre-built lookup table is queried to determine whether the current vehicle status belongs to the controllable domain of vehicle lateral stability, and a recommended torque distribution mode is determined. The expected additional yaw moment is calculated based on the vehicle yaw rate error and the vehicle center of gravity sideslip angle error. ; Based on the recommended torque distribution pattern, the desired additional yaw moment will be applied. The driving torque and / or braking torque are distributed to each wheel; Control the corresponding wheels to execute the driving torque and / or braking torque so that the lateral stability of the vehicle is restored to a preset stable state range; When the controllability judgment result is an uncontrollable state, a preset stability mitigation strategy is executed. The preset stability mitigation strategy includes at least one of the following: limiting drive torque, increasing braking intervention, reducing vehicle speed, outputting risk warning, or requesting intervention from the upper-level driving assistance system.

2. The lateral stability control method for distributed drive electric vehicles based on controllable domain according to claim 1, characterized in that: The lookup table is pre-built in the offline stage; in the offline stage, based on the vehicle dynamics model and the tire model, numerical simulations are performed under different torque distribution modes to obtain the stability recovery results under different vehicle operating conditions. Based on the stability recovery results, a controllable domain for vehicle lateral stability is constructed that considers control actions, actuator constraints, and finite-time convergence requirements. Based on the stability recovery effect under different torque distribution modes, mode selection labels are generated for the vehicle operating status, resulting in a mode selection dataset.

3. The lateral stability control method for distributed drive electric vehicles based on controllable domain according to claim 2, characterized in that: The controllable domain for vehicle lateral stability is determined based on whether the vehicle's lateral stability state can enter and remain within a preset stable neighborhood within the maximum allowable recovery time; the preset stable neighborhood... Represented as: ； in, The threshold value for the vehicle's center of gravity sideslip angle. The threshold for the rate of change of the vehicle's center of gravity sideslip angle; For any torque distribution mode Stability recovery time Defined as the first time the vehicle's lateral stability state enters the preset stability neighborhood. and during the preset stay time It remains within the preset stable neighborhood. The earliest moment: ； in, Candidate recovery time, For the time variable within the preset residence time interval, and They represent The vehicle's sideslip angle at time t and its rate of change; if the vehicle's condition is within the maximum permissible recovery time If the target does not enter and remain within a stable neighborhood, the stability recovery time in this mode can be denoted as infinity.

4. The lateral stability control method for distributed drive electric vehicles based on controllable domain according to claim 3, characterized in that: The controllability tag of the vehicle's operating status The stability recovery time is determined based on different torque distribution modes and is expressed as: ； in, This is a set of torque distribution modes, where DBDM represents differential braking distribution mode and BTVDM represents balanced torque vector distribution mode. when When the vehicle's operating status is classified as controllable, then... At that time, the corresponding vehicle's operating status will be classified as uncontrollable.

5. The lateral stability control method for distributed drive electric vehicles based on controllable domain according to claim 4, characterized in that: The mode selection label is generated through a two-stage discrimination method. The first stage is used to determine whether the vehicle's operating state is controllable under at least one torque distribution mode. The second stage is used to determine the recommended torque distribution mode based on the stability recovery time, vehicle speed loss, and control cost under different torque distribution modes in the vehicle's operating state that is determined to be controllable in the first stage. The vehicle speed loss and control costs m They are represented as follows: ； ； in, To evaluate the time interval, The initial longitudinal velocity of the vehicle, for The longitudinal speed of the vehicle at any given time, Torque distribution mode Next The wheel-end torque of each wheel, Indicates the front left, front right, rear left, and rear right wheels. For the first One wheel angular velocity at time t; When different torque distribution modes can all restore the vehicle's lateral stability to a preset stable state range, and the stability recovery time difference between different torque distribution modes is greater than a preset time difference threshold. When choosing a torque distribution mode with a shorter stability recovery time, select the one that is most suitable for the situation. When the stability recovery time difference between different torque distribution modes is not greater than the preset time difference threshold At that time, the loss is based on the vehicle speed. and / or control costs Determine the recommended torque distribution mode.

6. The lateral stability control method for a distributed drive electric vehicle based on a controllable domain according to any one of claims 2 to 5, characterized in that: The model is driven by training data using vehicle operating status, controllability labels, and mode selection labels to obtain the controllable boundary of vehicle lateral stability and the torque distribution mode selection boundary. The input features of the data-driven model include the vehicle's longitudinal velocity. Road surface adhesion coefficient Vehicle center of gravity sideslip angle and the rate of change of the vehicle's center of gravity sideslip angle It also includes interaction terms and quadratic terms constructed from the input features; the input features Represented as: ； The trained data-driven model is converted into a multidimensional lookup table suitable for real-time invocation by the vehicle controller. The index of the multidimensional lookup table includes at least the vehicle's longitudinal speed. Road surface adhesion coefficient Vehicle center of gravity sideslip angle and the rate of change of the vehicle's center of gravity sideslip angle The output of the multidimensional lookup table includes at least the controllability judgment result of the current vehicle state and the recommended torque distribution mode.

7. The lateral stability control method for distributed drive electric vehicles based on controllable domain according to claim 1, characterized in that: The torque distribution modes include at least a differential braking distribution mode and a balanced torque vector distribution mode; When the recommended torque distribution mode is the differential braking distribution mode, the desired additional yaw moment is applied. The direction of braking torque is selected by applying braking torque to at least one wheel on the left or right side of the vehicle, and the required total braking torque is determined according to the following relationship. : ； in, The radius of the wheel's rolling motion. This refers to the front track width. This refers to the rear track width; For the front and rear wheels on the selected braking side, the braking torque is distributed according to the corresponding wheel vertical load ratio; when the right wheel is selected for braking, the braking torque satisfies: ； ； in, and These represent the wheel-end torques of the right front wheel and the right rear wheel, respectively, with positive for driving and negative for braking; and The vertical loads are for the right front wheel and the right rear wheel, respectively; when the left wheel is selected for braking, the braking torque is symmetrically distributed according to the ratio of the vertical loads of the left front and rear wheels. The wheel end torque of each wheel satisfies the tire adhesion constraint: ； in, For the first The wheel-end torque of each wheel, For the first Vertical load on each wheel.

8. The lateral stability control method for distributed drive electric vehicles based on controllable domain according to claim 1, characterized in that: When the recommended torque distribution mode is the balanced torque vector distribution mode, the desired additional yaw moment will be... According to the front and rear axle distribution coefficients Decomposed into front axle yaw moment requirements and rear axle yaw moment requirements : ； ； Based on the yaw moment requirements of the front axle and the rear axle, determine the torque difference requirements for the left and right wheels of the front axle respectively. Torque difference requirements between the left and right rear wheels : ； ； For each axle, driving torque and braking torque in opposite directions are distributed between the left and right wheels to create the desired torque difference between the left and right wheels; Drive torque and braking torque It can be determined in the following way: ； ； in, Indicates front axle or rear axle. The maximum permissible drive torque for a single drive motor. The maximum permissible braking torque for the corresponding axle is determined; based on the direction of the desired additional yaw moment, the driving torque or braking torque is applied to the left and right wheels of each axle respectively, thereby generating the desired additional yaw moment under the conditions of satisfying the tire adhesion constraint, the maximum driving torque constraint of the motor and the maximum braking torque constraint of the brake.

9. The lateral stability control method for a distributed drive electric vehicle based on a controllable domain according to any one of claims 1 to 8, characterized in that: The method is executed by an on-board controller, which is configured with a status acquisition module, a lookup table calling module, a yaw moment calculation module, a torque distribution module, a safety mitigation module, and an execution control module, wherein: The status acquisition module is used to acquire vehicle longitudinal speed, road adhesion coefficient, vehicle center of gravity sideslip angle, and vehicle center of gravity sideslip angle change rate; The lookup table calling module is used to query a pre-built lookup table based on the vehicle's longitudinal speed, road surface adhesion coefficient, vehicle center of gravity sideslip angle, and vehicle center of gravity sideslip angle change rate, and output the controllability judgment result of the current vehicle state and the recommended torque distribution mode. The yaw moment calculation module is used to calculate the expected additional yaw moment based on the vehicle yaw rate error and the vehicle center of gravity sideslip angle error. A torque distribution module is used to distribute the desired additional yaw moment into the driving torque and / or braking torque of each wheel according to the recommended torque distribution mode. The safety mitigation module is used to execute a preset stability mitigation strategy when the controllability judgment result is an uncontrollable state; The execution control module is used to control the vehicle motor and / or brakes to execute the drive torque and / or braking torque, and to cooperate with the safety mitigation module to execute the preset stability mitigation strategy.

Images