Failure-tolerant control method and device for motor of four-wheel hub-driven light commercial vehicle

Abstract

The present invention provides a method and device for fault-tolerant control of motor failure of a four-wheel hub driven light commercial vehicle, and relates to the field of motor control technology. The method is based on digital twin technology, integrates Simulink simulation model and Trucksim vehicle model, and constructs a comprehensive mapping of physical entity and virtual entity. Multi-data integration is achieved through multi-domain and multi-scale fusion modeling. The BiLSTM model is used to train the simulation data set to learn the complex characteristics of the vehicle's driving state. When the motor fails, the torque is redistributed to the remaining normal actuators. In addition, the method also involves a vehicle state monitoring algorithm, which ensures the vehicle's driving performance through a recurrent neural network to realize real-time monitoring of the vehicle's driving state and fault warning, and timely and accurately performs fault-tolerant control on the failed wheels. It also includes the design of fault-tolerant control, which can timely detect the vehicle's failure status and position and alarm, and redistribute the torque to the non-failed wheels to improve the stability and safety of the vehicle.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and in particular to a method and device for controlling a motor failure of a four-wheel hub driven light commercial vehicle. Background Art

[0002] In today's automotive technology, four-wheel hub drive light commercial vehicles have attracted attention due to their unique drive mode and potential performance advantages. However, this drive system also faces some technical challenges, especially in terms of fault-tolerant control in the event of motor failure. Traditionally, the driving safety and stability of the vehicle will be seriously affected when the motor fails, because each motor in the four-wheel hub drive system works independently. Once one or more motors fail, the dynamic balance of the vehicle may be broken, resulting in deviation, vibration, or even more serious traffic accidents.

[0003] Currently, most of the existing fault-tolerant control technologies are designed based on rules. Although they can cope with simple fault conditions to a certain extent, the limitations of these technologies gradually emerge when faced with complex working conditions and changing driving environments. For example, when a vehicle encounters a motor failure while driving at high speed or under harsh road conditions, traditional fault-tolerant control strategies may not be able to adjust the vehicle's dynamics in a timely and effective manner to ensure the stability and safety of the vehicle. In addition, there is currently a lack of effective testing and verification methods for the failure-tolerant characteristics of distributed drive systems. Carrying out real-vehicle testing in real environments and road conditions is not only costly, but also poses safety risks. Therefore, a new method is needed to simulate and verify the behavior of the vehicle in the event of a motor failure in order to develop a more reliable and safer fault-tolerant control strategy.

[0004] Although there are some technical solutions on the market that attempt to solve similar problems, they often focus on specific control algorithms or single fault response measures. For example, some solutions use sliding mode control algorithms to improve vehicle stability when a motor fails, but these algorithms may not be flexible enough when dealing with complex multi-motor failure situations. Other solutions focus on the redistribution of wheel torque, but in actual applications, these distribution strategies may not fully take into account the real-time dynamics of the vehicle and changes in the driving environment.

[0005] In summary, the current technological status quo urgently requires a comprehensive solution, which not only needs to accurately monitor the vehicle's driving status, but also be able to quickly and accurately redistribute torque when the motor fails, while also having sufficient flexibility and adaptability to cope with various complex working conditions and driving environments.

[0006] In view of this, this application is filed. Summary of the invention

[0007] The present invention provides a method and device for controlling a motor failure of a four-wheel hub driven light commercial vehicle, which can at least partially improve the above-mentioned problem.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A method for controlling a motor failure of a four-wheel hub driven light commercial vehicle, comprising:

[0010] Acquire a physical data set collected by a sensor assembly configured on a commercial vehicle, and establish a virtual entity corresponding to the commercial vehicle based on the physical data set;

[0011] Acquire a virtual data set of the virtual entity, fuse the virtual data set with the physical data set, and generate a data fusion simulation set;

[0012] The BiLSTM model is used to perform training preprocessing on the data fusion simulation set to obtain the optimal torque distribution data after the commercial vehicle drive system fails;

[0013] Based on the current state parameters of the commercial vehicle, the optimal torque distribution data is subjected to fault-tolerant control processing to achieve control of the motor of the commercial vehicle.

[0014] The present invention also provides a motor failure tolerance control device for a four-wheel hub driven light commercial vehicle, which comprises:

[0015] A virtual entity establishment unit, configured to obtain a physical data set collected by a sensor assembly configured on a commercial vehicle, and establish a virtual entity corresponding to the commercial vehicle based on the physical data set;

[0016] A data fusion unit, used for acquiring a virtual data set of the virtual entity, fusing the virtual data set with the physical data set, and generating a data fusion simulation set;

[0017] A BiLSTM unit, used for training and preprocessing the data fusion simulation set using a BiLSTM model to obtain optimal torque distribution data after a commercial vehicle drive system fails;

[0018] The fault-tolerant control unit is used to perform fault-tolerant control processing on the optimal torque distribution data based on the current state parameters of the commercial vehicle, so as to realize control of the motor of the commercial vehicle.

[0019] In summary, the motor failure-tolerant control method for the four-wheel hub-driven light commercial vehicle uses advanced digital twin technology combined with deep learning algorithms to achieve precise control of the vehicle in the event of a motor failure. By building a detailed vehicle dynamics model and simulation system, various fault scenarios can be simulated in a virtual environment, so as to predict and optimize the torque distribution scheme in advance. In actual applications, this method can monitor the vehicle status in real time. Once a motor failure is detected, the output torque of the remaining motors is quickly adjusted through a pre-trained neural network model to ensure the stable driving of the vehicle. In addition, it also covers the complete process from data collection, model training to fault diagnosis and torque redistribution, providing comprehensive technical support for improving the safety and reliability of the vehicle. It aims to improve the driving stability and safety of the vehicle in the event of a motor failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of a motor failure-tolerant control method for a four-wheel hub-driven light commercial vehicle provided by the first embodiment of the present invention;

[0021] Figure 2 It is a schematic diagram of the framework structure of the motor failure-tolerant control method for a four-wheel hub-driven light commercial vehicle provided by the first embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of a seven-degree-of-freedom dynamics model of a vehicle provided by an embodiment of the present invention;

[0023] Figure 4 It is a schematic diagram of the data fusion simulation set construction process provided by an embodiment of the present invention;

[0024] Figure 5 Schematic diagram of the BiLSTM model training process provided by the embodiment of the present invention;

[0025] Figure 6 It is a schematic diagram of a fault-tolerant control flow provided by an embodiment of the present invention;

[0026] Figure 7 It is a module schematic diagram of a motor failure-tolerant control device for a four-wheel hub-driven light commercial vehicle provided by a second embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0028] refer to Figure 1 , Figure 2As shown, the first embodiment of the present invention discloses a motor failure fault-tolerant control method for a four-wheel hub driven light commercial vehicle, which can be executed by a motor failure fault-tolerant control device for a four-wheel hub driven light commercial vehicle (hereinafter referred to as a control device), and in particular, executed by one or more processors in the control device to implement the following method:

[0029] S1, obtaining a physical data set collected by a sensor component configured on a commercial vehicle, and establishing a virtual entity corresponding to the commercial vehicle based on the physical data set;

[0030] Specifically, step S1 includes: acquiring a physical data set collected by a sensor assembly configured on a commercial vehicle, wherein the physical data set includes vehicle speed, wheel speed, front wheel steering angle, longitudinal acceleration, lateral acceleration, and battery SOC;

[0031] Establishing a whole vehicle simulation dynamics model based on the physical data set, wherein the whole vehicle simulation dynamics model includes a vehicle seven-degree-of-freedom dynamics model, a magic tire model, a hub motor model and an Ackerman steering model;

[0032] The whole vehicle simulation dynamics model, Simulink simulation model and Trucksim whole vehicle model are combined to form a virtual entity corresponding to the commercial vehicle, specifically:

[0033] Combining the Simulink simulation model with the Trucksim vehicle model;

[0034] The Simulink simulation model is compared with the Trucksim vehicle model to screen out the same vehicle model parameters, and combined with the vehicle simulation dynamics model to generate a virtual entity corresponding to the commercial vehicle.

[0035] Preferably, the sensor assembly includes a lateral acceleration sensor, a longitudinal acceleration sensor, a vehicle speed sensor, and a wheel speed sensor.

[0036] In this embodiment, it is first necessary to obtain the physical data set of the vehicle. This process involves configuring a series of high-precision sensors on commercial vehicles. These sensors can collect the key operating parameters of the vehicle in real time to form a physical data set. Specifically, these sensors include lateral acceleration sensors, longitudinal acceleration sensors, vehicle speed sensors and wheel speed sensors, which are installed at appropriate positions of the vehicle to monitor the lateral acceleration, longitudinal acceleration, driving speed and wheel speed of the vehicle. In addition, a battery SOC sensor that monitors the battery status is also included to ensure that the power supply status of the vehicle is also taken into consideration. In short, the physical data set can be obtained by sensor signals, driver input and vehicle speed following control. The sensor components are shown in Table 1.

[0037] Table 1 Sensor categories and their functions

[0038]

[0039] The collection of these physical data sets is the basis for the subsequent establishment of virtual entities and fault-tolerant control. Based on these data, a vehicle simulation dynamics model can be built on the Simulink platform. This model is a highly complex mathematical model that integrates multiple sub-modules such as the vehicle's seven-degree-of-freedom dynamics model, magic tire model, hub motor model, and Ackerman steering model. These sub-modules work together to accurately simulate the dynamic behavior of the vehicle under various working conditions, providing a solid theoretical basis for the creation of virtual entities.

[0040] Next, the Simulink simulation model is combined with the Trucksim vehicle model. Trucksim is a professional vehicle dynamics simulation software that provides detailed vehicle models and rich road scenarios. By combining the two models, the control strategy in Simulink and the vehicle dynamics characteristics in Trucksim can be combined to form a more comprehensive and accurate virtual entity. In the process of combining, the consistency and accuracy of the model are ensured by screening out the same vehicle model parameters in the two models. These parameters include the mass, size, tire characteristics, etc. of the vehicle, which are crucial for accurately simulating the behavior of the vehicle. Specifically, the built-in interface of the Trucksim software and the extended interface in MATLAB / Simulink are used to jointly generate the S function in the Simulink library, and data interaction between the vehicle model and other systems is carried out based on the interface. Improve the accuracy of the established model. Realize multi-domain and multi-scale fusion modeling of the vehicle system, combine actual operation with theoretical simulation, and improve the accuracy of the digital twin model.

[0041] In this embodiment, the focus of the motor failure fault-tolerant control method for a four-wheel hub-driven light commercial vehicle is the fault-tolerant control after the failure of the distributed drive actuator. The criterion is whether the vehicle can continue to maintain stability after the actuator fails. A vehicle dynamics model that can reflect the lateral movement, longitudinal movement, and yaw movement of the vehicle is established, such as Figure 3 As shown, the vehicle's seven-degree-of-freedom model has moderate complexity and can well characterize the vehicle's motion state.

[0042] S2, obtaining a virtual data set of the virtual entity, fusing the virtual data set with the physical data set, and generating a data fusion simulation set;

[0043] Specifically, step S2 includes: obtaining a virtual data set of the virtual entity, wherein the virtual data set includes yaw angular velocity, center of mass sideslip angle, motor fault information, and torque;

[0044] Using an SMC sliding mode control algorithm to calculate the physical data set and the virtual data set under different preset working conditions;

[0045] The data calculated by the SMC sliding mode control algorithm and the physical data set are subjected to HIL hardware online testing, and a data fusion simulation set is constructed according to the test results.

[0046] See also Figure 4 In this embodiment, step S2 involves data flow and connection interaction between the physical entity (i.e., commercial vehicle) and the virtual model; that is, obtaining the virtual data set of the virtual entity and fusing it with the physical data set to generate a data fusion simulation set. This process first involves extracting key virtual data from the virtual entity, including yaw rate, center of mass sideslip angle, motor fault information and torque, etc. These virtual data are obtained by simulating various operating conditions in Simulink and Trucksim models, and they can reflect the dynamic response and potential fault status of the vehicle in different driving scenarios. In short, through the data transmission of sensors, the real-time data of the physical entity can be transmitted to the virtual model to obtain a virtual data set.

[0047] In order to ensure the compatibility and accuracy between these virtual data and the physical data of the actual vehicle, the SMC sliding mode control algorithm is used to process the physical data sets under different preset working conditions. The SMC sliding mode control algorithm is an advanced control strategy that can provide fast and accurate response in complex dynamic systems, and is particularly suitable for dealing with nonlinear and uncertain problems in vehicle dynamics. By combining the physical data set with the virtual data set and applying the SMC sliding mode control algorithm, the ideal dynamic response and torque distribution scheme of the vehicle under various working conditions can be calculated.

[0048] Subsequently, the data calculated by the SMC sliding mode control algorithm and the physical data set are processed by HIL (Hardware-in-the-Loop) hardware online testing. HIL testing is an effective verification method that evaluates the performance of control strategies in actual applications by combining actual hardware components (such as sensors and controllers) with simulation models. In this process, by comparing the simulation results with the response of the actual hardware, the control strategy can be fine-tuned to ensure its effectiveness and reliability in the actual vehicle. Based on the results of the HIL hardware online test, a data fusion simulation set is constructed. This data fusion simulation set is a database that integrates physical data and virtual data. It not only contains the operating data of the vehicle under normal conditions, but also includes the response data under simulated fault conditions. This data set provides rich data resources for subsequent deep learning model training, enabling the model to learn the dynamic behavior patterns of the vehicle under various complex situations, so as to more accurately predict and respond to fault conditions such as motor failure in actual applications.

[0049] In simple terms, the physical entity uses a four-wheel hub-driven light commercial vehicle for testing, and uses various sensors installed on the vehicle to obtain data in real time. Then, the data collected under different working conditions are associated with the virtual entity, and the established simulation model is entered. Through SMC sliding mode control, the vehicle dynamic response output is obtained. Since it is difficult to directly carry out failure testing on the physical entity, HIL hardware-in-the-loop testing is added to ensure the validity of physical and virtual data, and the obtained data is verified through HIL hardware-in-the-loop testing. If the verification effect is ideal, a driving data set is constructed; if the verification effect is not ideal, the simulation model and control method are improved, and simulation training is performed again until the verification effect is ideal and a data set is constructed. The working conditions involved are shown in Table 2.

[0050] Table 2 Driving conditions of four-wheel hub driven light commercial vehicles

[0051]

[0052] Since the failure of the drive system in the working environment of a four-wheel hub drive light commercial vehicle is relatively complex, the distribution torque required for failures caused by different factors is different. This requires that all failures caused by factors be taken into account and training be conducted so that the vehicle can be fault-tolerantly controlled in a timely manner to improve vehicle stability and safety. The factors and specific situations that need to be considered are shown in Table 3.

[0053] Table 3 Factors to be considered and specific conditions

[0054]

[0055] S3, using the BiLSTM model to perform training preprocessing on the data fusion simulation set to obtain optimal torque distribution data after the commercial vehicle drive system fails;

[0056] Specifically, step S3 includes: inputting the data fusion simulation set into the BiLSTM model, and multiple layers in the BiLSTM model sequentially calculate the data fusion simulation set to generate a prediction result, wherein the BiLSTM model includes an LSTM layer, a BILSTM layer, a fully connected layer, and a DropOut layer;

[0057] Determine whether the accuracy of the prediction result is higher than a preset value, wherein the preset value is 90%;

[0058] If so, generate optimal torque distribution data;

[0059] If not, the BiLSTM model is used again to perform training preprocessing on the data fusion simulation set.

[0060] See also Figure 5 In this embodiment, the BiLSTM (Bidirectional Long Short-Term Memory) model is used to perform deep learning training on the data fusion simulation set to obtain the optimal torque distribution data of the commercial vehicle drive system under failure conditions. This process begins by inputting the data fusion simulation set into the BiLSTM model. The BiLSTM model is a powerful deep learning architecture that is particularly suitable for processing sequence data and can capture long-term dependencies in time series, which is crucial for understanding and predicting vehicle dynamics behavior.

[0061] After obtaining the physical data collected by the sensor and the virtual data obtained by the simulation model, the data is fused. After a large amount of operation data, no less than 100,000 groups of data are obtained to construct a digital twin simulation data set, that is, the data fusion simulation set, and the BiLSTM model algorithm is used to train the data to obtain the optimal torque distribution data after the failure of the light commercial vehicle drive system. Specifically, multiple layers in the BiLSTM model, including the LSTM layer, the BILSTM layer, the fully connected layer and the DropOut layer, calculate the data fusion simulation set in turn. The LSTM layer can process and predict the time dependency in the sequence data, and the BILSTM layer further enhances the model's ability to understand the context information of the sequence on this basis, which is particularly important for accurately predicting the behavior of the vehicle under complex working conditions. The fully connected layer is used to integrate the output of the BILSTM layer and map the learned features to the final output space, that is, the torque distribution scheme. The DropOut layer is used to prevent the model from overfitting, improve the generalization ability of the model, and ensure that the model can still maintain good performance when facing new and unseen data.

[0062] During the model training process, prediction results are generated, which are the model's predictions of the optimal torque distribution of commercial vehicles under different failure conditions. Next, the accuracy of these prediction results needs to be evaluated / precisely checked to determine whether they are higher than the preset accuracy threshold, which is set to 90% in this embodiment. The setting of this accuracy threshold is based on the high requirements for model prediction accuracy and reliability, ensuring that the torque distribution scheme output by the model can effectively guide the stable driving of the vehicle in practical applications.

[0063] If the accuracy of the prediction result is higher than 90%, the model training is considered successful, and the generated prediction result can be used as the optimal torque distribution data for subsequent vehicle control strategies. These optimal torque distribution data will guide the vehicle's torque redistribution in the event of motor failure, ensuring that the vehicle can continue to drive safely and stably, significantly improving the safety and reliability of the vehicle in the event of a fault.

[0064] On the contrary, if the accuracy of the prediction results fails to reach 90%, it means that the current training effect of the model is not enough to meet the needs of practical applications. In this case, the BiLSTM model will be reused to train and preprocess the data fusion simulation set. This iterative training process will continue until the model can generate prediction results that meet the accuracy requirements. Through this continuously optimized training method, this method ensures that the adopted deep learning model can adapt to different data characteristics and working conditions, so as to maximize its effectiveness in actual vehicle control. Among them, the obtained operation result data is shown in Table 4.

[0065] Table 4 Operation result data composition table

[0066]

[0067] S4, based on the current state parameters of the commercial vehicle, performing fault-tolerant control processing on the optimal torque distribution data to achieve control of the motor of the commercial vehicle.

[0068] Specifically, step S4 includes: obtaining the current state parameter set of the commercial vehicle, and determining whether the state parameter exceeds the preset normal operating condition parameter value, wherein the state parameter set includes the longitudinal acceleration a x , lateral acceleration a v , yaw rate , center of mass sideslip angle and wheel torque T;

[0069] If not, output the state parameter set;

[0070] If so, the state parameter set is calculated using a neural network model to obtain the failure position and failure degree of the commercial vehicle;

[0071] Outputting the failure position and failure degree of the commercial vehicle and issuing a fault alarm;

[0072] Based on the failure position and failure degree of the commercial vehicle, corresponding data are selected from the optimal torque distribution data to control the motor of the commercial vehicle.

[0073] Preferably, the failure factor Describe the failure degree of commercial vehicles, where the failure factor The larger the failure factor, the lighter the failure degree. When the commercial vehicle's drive system works normally, the failure factor When the drive system of the commercial vehicle partially fails, there is execution capability. When the failure factor When the vehicle is in a state of emergency, all the drive systems of the commercial vehicle fail.

[0074] See also Figure 6 In this embodiment, the optimal torque distribution data is subjected to fault-tolerant control processing according to the current state parameters of the commercial vehicle to achieve effective control of the commercial vehicle motor. This stage first involves obtaining the current state parameter set of the commercial vehicle, which includes the longitudinal acceleration (a x ), lateral acceleration (a v ), yaw rate and wheel torque (T), etc. These state parameters are collected in real time by sensors on the vehicle and can reflect the current operating state and dynamic behavior of the vehicle. These state parameters are then evaluated to determine whether they exceed the preset normal operating condition parameter values. These normal operating condition parameter values ​​are pre-set according to the vehicle's design specifications and safety standards and are used to define the parameter range of the vehicle under normal operating conditions. If the state parameters do not exceed the normal operating condition parameter values, it means that the vehicle is currently in normal operating state, and the system will directly output these state parameter sets for reference by the driver or other vehicle control systems.

[0075] If the state parameters exceed the normal operating condition parameter values, this may mean that a part of the vehicle has a fault or anomaly. In this case, the system will use a pre-trained neural network model to calculate the state parameter set. This neural network model can accurately identify the failure location and failure degree of the commercial vehicle based on the input state parameters. The determination of the failure location helps to quickly locate the source of the fault, and the evaluation of the failure degree provides a key basis for the subsequent torque distribution.

[0076] Once the failure location and failure degree are determined, the system will output this information and trigger a fault alarm to alert the driver to the abnormal state of the vehicle. At the same time, the system will filter out the data corresponding to the current fault condition from the optimal torque distribution data previously trained by the BiLSTM model based on the failure location and failure degree. This data will be used to accurately control the motor of the commercial vehicle to adjust the output torque of the motor to ensure that the vehicle can still maintain stable driving in the event of a fault.

[0077] Simply put, the failure location of the drive system and when the failure occurs are determined by detecting the vehicle status parameters and an alarm is issued. Then, the model is trained based on the simulation data set to obtain a large amount of data on various drive system failure conditions and relative stress-torque distribution to distribute the torque, thereby preventing the vehicle from running off the track and improving the vehicle's driving stability and safety.

[0078] In this embodiment, a failure factor is used to quantify the failure degree of the commercial vehicle. The failure factor ranges from 0 to 1, where the larger the failure factor, the lighter the failure degree. Specifically, when the failure factor is equal to 1, it means that the drive system of the commercial vehicle is working completely normally; when the failure factor is between 0 and 1, it means that the drive system is partially failed, but still has a certain execution capability; and when the failure factor is equal to 0, it means that the drive system is completely failed. This quantification method provides clear guidance for torque distribution, allowing the system to flexibly adjust the torque distribution strategy according to different failure degrees.

[0079] Among them, the service applications involved in the four-wheel hub driven light commercial vehicle motor failure fault tolerance control method include:

[0080] (1) If the vehicle's yaw rate is within the normal range during driving, the current vehicle status is output normally; if the yaw rate exceeds the normal range (0.1° / s~0.3° / s), the system inputs the parameter into the classification neural network, and the classification neural network determines the failure location and degree. If the failure factor is 0, an alarm is issued to remind the driver to brake the vehicle and pull over; if the failure factor is 0< <1, according to the pre-trained torque distribution model, the torque is redistributed to the wheels that have not failed.

[0081] (2) If the vehicle's side slip angle is within the normal range (-5°~+5°) during acceleration or deceleration, the driving state is normal and the current vehicle state information is output; if the vehicle's lateral acceleration and longitudinal acceleration change significantly, causing the vehicle to deviate from its original driving trajectory, the side slip angle changes greatly and exceeds the normal range, resulting in the disappearance of wheel torque and failure. At this time, the classification neural network determines the failure position and level of the drive system. If the failure factor is 0, an alarm is issued to remind the driver to brake the vehicle and pull over; if the failure factor is 0< <1, according to the pre-trained torque distribution model, the torque is redistributed to the wheels that have not failed.

[0082] (3) If the wheel torque or speed does not change during driving, the driving status is normal and the current vehicle status information is output; if the wheel torque or speed changes to 0, the system inputs the parameter into the classification neural network, and the classification neural network determines the failure location and degree. If the failure factor is 0, an alarm is issued to remind the driver to brake the vehicle and pull over; if the failure factor is 0< <1, according to the pre-trained torque distribution model, the torque is redistributed to the wheels that have not failed.

[0083] In summary, this method aims to significantly improve the driving stability and safety of the vehicle in the face of motor failure. The method cleverly combines digital twin technology, advanced simulation models and deep learning algorithms to build a comprehensive vehicle control and fault response system. Through the combination of digital twin technology and deep learning models, precise control of the vehicle motor failure is achieved, which improves the driving safety and stability of the vehicle. The four-wheel hub drive light commercial vehicle motor failure fault-tolerant control method can quickly respond to fault conditions, adjust torque distribution, and maintain the dynamic balance of the vehicle through real-time monitoring and data fusion. In addition, the failure factor is used to quantify the degree of failure, which provides clear guidance for torque distribution and enhances the adaptability and robustness of the system. Overall, the four-wheel hub drive light commercial vehicle motor failure fault-tolerant control method provides an efficient and reliable solution for the fault-tolerant control of four-wheel hub drive light commercial vehicles, and has broad application prospects.

[0084] See also Figure 7 The second embodiment of the present invention provides a motor failure tolerance control device for a four-wheel hub driven light commercial vehicle, which comprises:

[0085] A virtual entity establishing unit 201 is used to obtain a physical data set collected by a sensor component configured on a commercial vehicle, and establish a virtual entity corresponding to the commercial vehicle based on the physical data set;

[0086] A data fusion unit 202 is used to obtain a virtual data set of the virtual entity, fuse the virtual data set with the physical data set, and generate a data fusion simulation set;

[0087] The BiLSTM unit 203 is used to perform training preprocessing on the data fusion simulation set using a BiLSTM model to obtain optimal torque distribution data after the commercial vehicle drive system fails;

[0088] The fault-tolerant control unit 204 is used to perform fault-tolerant control processing on the optimal torque distribution data based on the current state parameters of the commercial vehicle, so as to realize control of the motor of the commercial vehicle.

[0089] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for controlling a motor failure of a four-wheel hub driven light commercial vehicle, characterized in that: include: Acquire a physical data set collected by a sensor assembly configured on a commercial vehicle, and establish a virtual entity corresponding to the commercial vehicle based on the physical data set; Acquire a virtual data set of the virtual entity, fuse the virtual data set with the physical data set, and generate a data fusion simulation set; The BiLSTM model is used to perform training preprocessing on the data fusion simulation set to obtain the optimal torque distribution data after the commercial vehicle drive system fails; Based on the current state parameters of the commercial vehicle, the optimal torque distribution data is subjected to fault-tolerant control processing to achieve control of the motor of the commercial vehicle; Acquire a virtual data set of the virtual entity, fuse the virtual data set with the physical data set, and generate a data fusion simulation set, specifically: Acquire a virtual data set of the virtual entity, wherein the virtual data set includes yaw angular velocity, center of mass sideslip angle, motor fault information, and torque; Using an SMC sliding mode control algorithm to calculate the physical data set and the virtual data set under different preset working conditions; The data calculated by the SMC sliding mode control algorithm and the physical data set are subjected to HIL hardware online testing, and a data fusion simulation set is constructed according to the test results.

2. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 1 is characterized in that: Acquire a physical data set collected by a sensor assembly configured on a commercial vehicle, and establish a virtual entity corresponding to the commercial vehicle based on the physical data set, specifically: Acquire a physical data set collected by a sensor assembly configured on a commercial vehicle, wherein the physical data set includes vehicle speed, wheel speed, front wheel steering angle, longitudinal acceleration, lateral acceleration, and battery SOC; Establishing a whole vehicle simulation dynamics model based on the physical data set, wherein the whole vehicle simulation dynamics model includes a vehicle seven-degree-of-freedom dynamics model, a magic tire model, a hub motor model and an Ackerman steering model; The whole vehicle simulation dynamics model, Simulink simulation model and Trucksim whole vehicle model are combined to form a virtual entity corresponding to the commercial vehicle.

3. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 2 is characterized in that: The whole vehicle simulation dynamics model, Simulink simulation model and Trucksim whole vehicle model are combined to form a virtual entity corresponding to the commercial vehicle, specifically: Combining the Simulink simulation model with the Trucksim vehicle model; The Simulink simulation model is compared with the Trucksim vehicle model to screen out the same vehicle model parameters, and combined with the vehicle simulation dynamics model to generate a virtual entity corresponding to the commercial vehicle.

4. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 1, characterized in that: The BiLSTM model is used to train and preprocess the data fusion simulation set to obtain the optimal torque distribution data after the commercial vehicle drive system fails, specifically: The data fusion simulation set is input into the BiLSTM model, and multiple layers in the BiLSTM model calculate the data fusion simulation set in sequence to generate prediction results, wherein the BiLSTM model includes an LSTM layer, a BILSTM layer, a fully connected layer, and a DropOut layer; Determine whether the accuracy of the prediction result is higher than a preset value, wherein the preset value is 90%; If so, generate optimal torque distribution data; If not, the BiLSTM model is used again to perform training preprocessing on the data fusion simulation set.

5. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 1, characterized in that: Based on the current state parameters of the commercial vehicle, the optimal torque distribution data is subjected to fault-tolerant control processing to achieve control of the commercial vehicle motor, specifically: Obtain the current state parameter set of the commercial vehicle and determine whether the state parameter exceeds the preset normal operating condition parameter value, wherein the state parameter set includes the longitudinal acceleration a x , lateral acceleration a v , yaw rate , center of mass sideslip angle and wheel torque T; If not, output the state parameter set; If so, the state parameter set is calculated using a neural network model to obtain the failure position and failure degree of the commercial vehicle; Outputting the failure position and failure degree of the commercial vehicle and issuing a fault alarm; Based on the failure position and failure degree of the commercial vehicle, corresponding data are selected from the optimal torque distribution data to control the motor of the commercial vehicle.

6. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 5, characterized in that: Failure Factor Describe the failure degree of commercial vehicles, where the failure factor The larger it is, the less severe the failure.

7. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 6, characterized in that: When the failure factor When the commercial vehicle's drive system works normally, the failure factor When the drive system of the commercial vehicle partially fails, there is execution capability. When the failure factor When the vehicle is in a state of emergency, all the drive systems of the commercial vehicle fail.

8. The motor failure tolerance control method for a four-wheel hub driven light commercial vehicle according to claim 1, characterized in that: The sensor assembly includes a lateral acceleration sensor, a longitudinal acceleration sensor, a vehicle speed sensor, and a wheel speed sensor.

9. A motor failure tolerance control device for a four-wheel hub driven light commercial vehicle, characterized in that: include: A virtual entity establishment unit, configured to obtain a physical data set collected by a sensor assembly configured on a commercial vehicle, and establish a virtual entity corresponding to the commercial vehicle based on the physical data set; A data fusion unit, used for acquiring a virtual data set of the virtual entity, fusing the virtual data set with the physical data set, and generating a data fusion simulation set; A BiLSTM unit, used for training and preprocessing the data fusion simulation set using a BiLSTM model to obtain optimal torque distribution data after a commercial vehicle drive system fails; A fault-tolerant control unit, used for performing fault-tolerant control processing on the optimal torque distribution data based on current state parameters of the commercial vehicle, so as to realize control of the motor of the commercial vehicle; Acquire a virtual data set of the virtual entity, fuse the virtual data set with the physical data set, and generate a data fusion simulation set, specifically: Acquire a virtual data set of the virtual entity, wherein the virtual data set includes yaw angular velocity, center of mass sideslip angle, motor fault information, and torque; Using an SMC sliding mode control algorithm to calculate the physical data set and the virtual data set under different preset working conditions; The data calculated by the SMC sliding mode control algorithm and the physical data set are subjected to HIL hardware online testing, and a data fusion simulation set is constructed according to the test results.

Images