Automobile intelligent chassis fault prediction self-healing method, device and equipment and storage medium

Abstract

The invention discloses an automobile intelligent chassis fault prediction self-healing method, device and equipment and a storage medium, and the method comprises the steps: synchronously collecting bus network data, controller internal variables, physical sensor data and environment perception data of a current new energy vehicle in real time, and constructing a space-time feature matrix through space-time alignment and feature level fusion; performing fault prediction and root cause positioning based on the spatial-temporal characteristic matrix, determining a fault type, a fault occurrence probability and a fault root cause position, and generating a corresponding self-healing decision instruction; according to the self-healing decision instruction, a multi-objective optimization algorithm is used for coordinating and controlling a plurality of subsystems of the chassis to execute a self-healing strategy; according to the method, the state of the chassis system can be comprehensively perceived and accurately represented, the recognition capability of early fault features is remarkably improved, the safety, reliability and user experience of the intelligent chassis system of the new energy automobile can be greatly improved, and the fault prediction and self-healing speed and efficiency of the intelligent chassis of the automobile are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent chassis control technology for new energy vehicles, and in particular to a method, device, equipment, and storage medium for predicting and self-healing faults in an intelligent chassis for automobiles. Background Technology

[0002] With the rapid development of new energy vehicle technology, the safety and reliability of intelligent chassis systems are receiving increasing attention. Currently, the mainstream vehicle fault diagnosis and handling technologies on the market mainly take the following forms: Threshold alarm systems are currently the most widely used technical solutions. Their working principle is based on triggering fault codes by setting preset fixed thresholds (such as excessive temperature, abnormal current, etc.) and relying on the on-board diagnostic (OBD) system for reporting. However, this system can only provide alarms after a fault occurs, lacks fault prediction capabilities, and cannot achieve proactive repair, resulting in the inability to eliminate potential vehicle safety hazards in advance.

[0003] Some high-end models use single-system fault-tolerant control technology, such as the braking system which uses a redundant design of dual electronic control units (ECUs). Such technology is limited to local hardware backup. When a cross-subsystem coupling failure occurs (such as steering system failure causing delay in the intervention of the Electronic Stability Program (ESP)), the lack of inter-system coordination mechanism still cannot effectively guarantee driving safety.

[0004] Furthermore, with the development of vehicle networking technology, cloud-based fault diagnosis technology has gradually emerged, which uploads vehicle data to the cloud for analysis and processing via 4G / 5G networks; however, due to the inherent latency of network transmission and cloud computing (usually greater than 500ms), this technology is difficult to meet the millisecond-level response requirements of control commands under high-speed driving conditions and cannot support real-time self-healing control.

[0005] Existing technologies generally suffer from the following drawbacks: First, there is a lack of effective identification of early fault characteristics, making it impossible to predict faults before they occur. Secondly, bus data, sensor data, and controller internal variables are often analyzed independently, ignoring the dynamic coupling relationship between various chassis subsystems; Furthermore, the fault handling strategy is limited to single-point degradation and fails to make full use of the chassis's global actuator resources for coordinated compensation. Finally, the response delay of existing technologies is relatively large, making it difficult to meet the real-time safety control requirements of modern intelligent electric vehicles. Summary of the Invention

[0006] The main objective of this invention is to provide a method, device, equipment, and storage medium for predicting and self-healing faults in an intelligent automotive chassis, aiming to solve the technical problems in the prior art, such as inaccurate fault prediction due to fragmented analysis of multi-source data, lack of collaborative self-healing capability for cross-subsystem faults, and difficulty in achieving millisecond-level real-time control.

[0007] In a first aspect, the present invention provides a method for predicting and self-healing faults in an intelligent automotive chassis, the method comprising the following steps: Real-time synchronous collection of bus network data, controller internal variables, physical sensor data and environmental perception data of current new energy vehicles, and construction of spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion; Based on the spatiotemporal feature matrix, fault prediction and root cause localization are performed to determine the fault type, fault occurrence probability and fault root cause location, and corresponding self-healing decision instructions are generated. According to the self-healing decision instruction, the self-healing strategy is executed by multiple subsystems of the chassis through a multi-objective optimization algorithm.

[0008] Optionally, the real-time synchronous acquisition of current new energy vehicle bus network data, controller internal variables, physical sensor data, and environmental perception data, and the construction of a spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion, includes: Real-time acquisition of bus network data of current new energy vehicles through vehicle communication bus protocol, synchronous acquisition of internal variables of controller through diagnostic interface or internal memory of electronic control unit (ECU), synchronous acquisition of physical sensor data through measuring device, and synchronous acquisition of environmental perception data through external sensing device. The bus network data, the controller internal variables, the physical sensor data, and the environmental perception data are spatiotemporally aligned and feature-level fused to construct a spatiotemporal feature matrix.

[0009] Optionally, the step of performing spatiotemporal alignment and feature-level fusion on the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data to construct a spatiotemporal feature matrix includes: The IEEE 1588 Precise Network Time Protocol (PTP) is used to synchronize the time deviations of the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data. Based on the vehicle kinematics model and coordinate transformation algorithm, spatial alignment of various data sources is performed, and sensor data distributed at different locations of the vehicle are uniformly mapped to a global coordinate system with the vehicle's center of mass as the origin. The system extracts periodic features and abnormal fluctuation indicators from the bus network data, performs state space reconstruction and outlier detection on the internal variables of the controller, applies wavelet packet decomposition to extract energy entropy features from the physical sensor data and calculates time-frequency domain statistics, and performs target recognition and scene semantic segmentation on the environmental perception data to extract road curvature and obstacle position features. The processed bus network data, controller internal variables, physical sensor data, and environmental perception data are organized according to a unified time series to construct a spatiotemporal feature matrix with N×M dimensions, where N represents the time series length and M represents the fused feature dimension.

[0010] Optionally, the step of performing fault prediction and root cause localization based on the spatiotemporal feature matrix, determining the fault type, fault occurrence probability, and fault root cause location, and generating corresponding self-healing decision instructions includes: When performing fault prediction and root cause localization based on the spatiotemporal feature matrix, an LSTM-Transformer hybrid neural network architecture is used to process the input spatiotemporal feature matrix to perform multi-level feature extraction of temporal features, and output the fault probability distribution and fault type prediction of each key component within a future preset time period. In the LSTM-Transformer hybrid neural network architecture, the LSTM layer captures short-term dynamic features, and the Transformer layer captures long-term dependencies through a self-attention mechanism. Simultaneously, a system dependency model based on a graph neural network is constructed, and the location of the root cause of the failure is determined based on the system dependency model; Based on the fault type, the fault occurrence probability, and the fault root cause location, a self-healing decision instruction containing the optimal actuator combination scheme, execution priority, and safety constraints is generated according to the preset fault strategy mapping rules and the preset real-time optimization algorithm.

[0011] Optionally, the step of simultaneously constructing a system dependency model based on a graph neural network and determining the root cause location of the fault based on the system dependency model includes: Simultaneously, a system dependency model based on graph neural networks is constructed, abstracting each subsystem of the current new energy vehicle chassis as graph nodes and the signal interaction between each subsystem as edges. The propagation path of abnormal patterns in the feature matrix in the system graph is analyzed through anomaly propagation algorithm, and the location of the root cause of the fault is determined by combining Bayesian inference.

[0012] Optionally, the step of coordinating and controlling the execution of the self-healing strategy by the chassis multi-subsystems through a multi-objective optimization algorithm according to the self-healing decision instruction includes: The current fault type information contained in the self-healing decision instruction is analyzed, and the chassis system combination to be activated and the corresponding safety boundary are determined based on the current fault type information. Based on the chassis system assembly and the safety boundary, a multi-actuator cooperative control model is established, encompassing the steering system, braking system, suspension system, and drive system of the current new energy vehicle. The multi-objective optimization function is determined using the following formula:

[0013] in, For multi-objective optimization functions, , , and The weighting coefficients are dynamically adjusted based on the severity of the fault and the vehicle's operating status. To track errors, For energy consumption, For comfort; By applying safety constraints during the solution of the multi-objective optimization function, it is ensured that the current new energy vehicle always remains within the stability domain; The optimal control quantity obtained from the optimization solution is assigned to the corresponding actuator, generating a control command sequence for each subsystem; The control chassis multi-subsystem is controlled to execute a self-healing strategy according to the control command sequence.

[0014] Optionally, controlling the control chassis multi-subsystem to execute a self-healing strategy according to the control command sequence includes: When the steering torque sensor of the current new energy vehicle is detected to be faulty, the system automatically switches to the redundant sensor channel according to the steering self-healing command in the control command sequence, and generates auxiliary steering torque through differential braking within a preset steering response time. When a brake fluid pressure deficiency fault is detected in the current new energy vehicle, the drive motor provides regenerative braking force compensation within a preset motor response time according to the brake hydraulic self-healing command in the control command sequence. At the same time, the damping force of the shock absorber is adjusted in real time through the suspension continuous damping control system CDC to control the vehicle pitch angle change rate to meet the preset change rate threshold. If the communication bus of the current new energy vehicle is detected to be congested, the transmission priority of non-critical signals is dynamically downgraded according to the communication self-healing command in the control command sequence to ensure the transmission delay of critical steering or braking commands.

[0015] Secondly, to achieve the above objectives, the present invention also proposes an intelligent chassis fault prediction and self-healing device for automobiles, the intelligent chassis fault prediction and self-healing device comprising: The spatiotemporal feature matrix construction module is used to collect bus network data, controller internal variables, physical sensor data and environmental perception data of current new energy vehicles in real time, and construct spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion. The self-healing decision instruction generation module is used to perform fault prediction and root cause localization based on the spatiotemporal feature matrix, determine the fault type, fault occurrence probability and fault root cause location, and generate corresponding self-healing decision instructions. The strategy execution module is used to coordinate and control the execution of the self-healing strategy by multiple subsystems of the chassis through a multi-objective optimization algorithm according to the self-healing decision instructions.

[0016] Thirdly, to achieve the above objectives, the present invention also proposes an intelligent chassis fault prediction and self-healing device for automobiles. The intelligent chassis fault prediction and self-healing device for automobiles includes: a memory, a processor, and an intelligent chassis fault prediction and self-healing program stored in the memory and executable on the processor. The intelligent chassis fault prediction and self-healing program for automobiles is configured to implement the steps of the intelligent chassis fault prediction and self-healing method for automobiles as described above.

[0017] Fourthly, to achieve the above objectives, the present invention also proposes a storage medium storing a vehicle intelligent chassis fault prediction and self-healing program, wherein when the vehicle intelligent chassis fault prediction and self-healing program is executed by a processor, it implements the steps of the vehicle intelligent chassis fault prediction and self-healing method described above.

[0018] The proposed method for predicting and self-healing faults in an intelligent chassis for automobiles involves real-time synchronous collection of bus network data, controller internal variables, physical sensor data, and environmental perception data from current new energy vehicles. A spatiotemporal feature matrix is ​​constructed through spatiotemporal alignment and feature-level fusion. Based on this matrix, fault prediction and root cause localization are performed to determine the fault type, probability of occurrence, and location of the root cause, generating corresponding self-healing decision instructions. According to these instructions, a multi-objective optimization algorithm coordinates and controls multiple chassis subsystems to execute self-healing strategies. This method achieves comprehensive perception and accurate characterization of the chassis system's state, significantly improving the ability to identify early fault characteristics. Fault prediction and root cause localization based on the fused spatiotemporal feature matrix accurately determine the fault type, probability of occurrence, and root cause, avoiding misjudgments and omissions caused by data fragmentation in traditional methods. The multi-objective optimization algorithm coordinates and controls multiple chassis subsystems to execute self-healing strategies, fully utilizing global actuator resources for collaborative compensation. This allows the vehicle to maintain safe driving performance even under single or multiple fault conditions, significantly improving the safety, reliability, and user experience of the intelligent chassis system for new energy vehicles, and enhancing the speed and efficiency of fault prediction and self-healing in intelligent chassis automobiles. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of the present invention; Figure 2 This is a flowchart illustrating the first embodiment of the vehicle intelligent chassis fault prediction and self-healing method of the present invention. Figure 3 This is a flowchart illustrating the second embodiment of the vehicle intelligent chassis fault prediction and self-healing method of the present invention. Figure 4 This is a flowchart illustrating the third embodiment of the vehicle intelligent chassis fault prediction and self-healing method of the present invention. Figure 5 This is a schematic diagram of the system framework corresponding to the vehicle intelligent chassis fault prediction and self-healing method of the present invention. Figure 6 This is a schematic diagram of the dynamic optimization algorithm in the vehicle intelligent chassis fault prediction and self-healing method of the present invention. Figure 7 This is a functional block diagram of the first embodiment of the intelligent chassis fault prediction and self-healing device of the present invention.

[0020] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0022] The solution of this invention mainly involves: real-time synchronous collection of bus network data, controller internal variables, physical sensor data, and environmental perception data of the current new energy vehicle; constructing a spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion; performing fault prediction and root cause localization based on the spatiotemporal feature matrix to determine the fault type, fault occurrence probability, and fault root cause location, and generating corresponding self-healing decision instructions; coordinating the execution of self-healing strategies by controlling multiple subsystems of the chassis through a multi-objective optimization algorithm according to the self-healing decision instructions; achieving comprehensive perception and accurate characterization of the chassis system state, significantly improving the ability to identify early fault characteristics; and performing fault prediction and root cause localization based on the fused spatiotemporal feature matrix. Fault prediction and root cause localization can accurately determine the fault type, probability of occurrence, and root cause, avoiding misjudgments and omissions caused by data fragmentation in traditional methods. Through multi-objective optimization algorithms, the system coordinates the execution of self-healing strategies by controlling multiple subsystems of the chassis, making full use of global actuator resources for collaborative compensation. This enables the vehicle to maintain safe driving performance under single or multiple fault conditions, significantly improving the safety, reliability, and user experience of the intelligent chassis system for new energy vehicles. It also improves the speed and efficiency of fault prediction and self-healing in intelligent chassis systems, solving the technical problems of inaccurate fault prediction due to fragmented analysis of multi-source data, lack of collaborative self-healing capabilities for cross-subsystem faults, and difficulty in achieving millisecond-level real-time control in existing technologies.

[0023] Reference Figure 1 , Figure 1 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of the present invention.

[0024] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU; a communication bus 1002; a user interface 1003; a network interface 1004; and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0025] Those skilled in the art will understand that Figure 1 The device structure shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0026] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating device, a network communication module, a user interface module, and a fault prediction and self-healing program for an intelligent automotive chassis.

[0027] The device of this invention calls the automotive intelligent chassis fault prediction and self-healing program stored in the memory 1005 through the processor 1001, and performs the following operations: Real-time synchronous collection of bus network data, controller internal variables, physical sensor data and environmental perception data of current new energy vehicles, and construction of spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion; Based on the spatiotemporal feature matrix, fault prediction and root cause localization are performed to determine the fault type, fault occurrence probability and fault root cause location, and corresponding self-healing decision instructions are generated. According to the self-healing decision instruction, the self-healing strategy is executed by multiple subsystems of the chassis through a multi-objective optimization algorithm.

[0028] The device of the present invention, through processor 1001 calling the automotive intelligent chassis fault prediction and self-healing program stored in memory 1005, also performs the following operations: Real-time acquisition of bus network data of current new energy vehicles through vehicle communication bus protocol, synchronous acquisition of internal variables of controller through diagnostic interface or internal memory of electronic control unit (ECU), synchronous acquisition of physical sensor data through measuring device, and synchronous acquisition of environmental perception data through external sensing device. The bus network data, the controller internal variables, the physical sensor data, and the environmental perception data are spatiotemporally aligned and feature-level fused to construct a spatiotemporal feature matrix.

[0029] The device of the present invention, through processor 1001 calling the automotive intelligent chassis fault prediction and self-healing program stored in memory 1005, also performs the following operations: The IEEE 1588 Precise Network Time Protocol (PTP) is used to synchronize the time deviations of the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data. Based on the vehicle kinematics model and coordinate transformation algorithm, spatial alignment of various data sources is performed, and sensor data distributed at different locations of the vehicle are uniformly mapped to a global coordinate system with the vehicle's center of mass as the origin. The system extracts periodic features and abnormal fluctuation indicators from the bus network data, performs state space reconstruction and outlier detection on the internal variables of the controller, applies wavelet packet decomposition to extract energy entropy features from the physical sensor data and calculates time-frequency domain statistics, and performs target recognition and scene semantic segmentation on the environmental perception data to extract road curvature and obstacle position features. The processed bus network data, controller internal variables, physical sensor data, and environmental perception data are organized according to a unified time series to construct a spatiotemporal feature matrix with N×M dimensions, where N represents the time series length and M represents the fused feature dimension.

[0030] The device of the present invention, through processor 1001 calling the automotive intelligent chassis fault prediction and self-healing program stored in memory 1005, also performs the following operations: When performing fault prediction and root cause localization based on the spatiotemporal feature matrix, an LSTM-Transformer hybrid neural network architecture is used to process the input spatiotemporal feature matrix to perform multi-level feature extraction of temporal features, and output the fault probability distribution and fault type prediction of each key component within a future preset time period. In the LSTM-Transformer hybrid neural network architecture, the LSTM layer captures short-term dynamic features, and the Transformer layer captures long-term dependencies through a self-attention mechanism. Simultaneously, a system dependency model based on a graph neural network is constructed, and the location of the root cause of the failure is determined based on the system dependency model; Based on the fault type, the fault occurrence probability, and the fault root cause location, a self-healing decision instruction containing the optimal actuator combination scheme, execution priority, and safety constraints is generated according to the preset fault strategy mapping rules and the preset real-time optimization algorithm.

[0031] The device of the present invention, through processor 1001 calling the automotive intelligent chassis fault prediction and self-healing program stored in memory 1005, also performs the following operations: Simultaneously, a system dependency model based on graph neural networks is constructed, abstracting each subsystem of the current new energy vehicle chassis as graph nodes and the signal interaction between each subsystem as edges. The propagation path of abnormal patterns in the feature matrix in the system graph is analyzed through anomaly propagation algorithm, and the location of the root cause of the fault is determined by combining Bayesian inference.

[0032] The device of the present invention, through processor 1001 calling the automotive intelligent chassis fault prediction and self-healing program stored in memory 1005, also performs the following operations: The current fault type information contained in the self-healing decision instruction is analyzed, and the chassis system combination to be activated and the corresponding safety boundary are determined based on the current fault type information. Based on the chassis system assembly and the safety boundary, a multi-actuator cooperative control model is established, encompassing the steering system, braking system, suspension system, and drive system of the current new energy vehicle. The multi-objective optimization function is determined using the following formula:

[0033] in, For multi-objective optimization functions, , , and The weighting coefficients are dynamically adjusted based on the severity of the fault and the vehicle's operating status. To track errors, For energy consumption, For comfort; By applying safety constraints during the solution of the multi-objective optimization function, it is ensured that the current new energy vehicle always remains within the stability domain; The optimal control quantity obtained from the optimization solution is assigned to the corresponding actuator, generating a control command sequence for each subsystem; The control chassis multi-subsystem is controlled to execute a self-healing strategy according to the control command sequence.

[0034] The device of the present invention, through processor 1001 calling the automotive intelligent chassis fault prediction and self-healing program stored in memory 1005, also performs the following operations: When the steering torque sensor of the current new energy vehicle is detected to be faulty, the system automatically switches to the redundant sensor channel according to the steering self-healing command in the control command sequence, and generates auxiliary steering torque through differential braking within a preset steering response time. When a brake fluid pressure deficiency fault is detected in the current new energy vehicle, the drive motor provides regenerative braking force compensation within a preset motor response time according to the brake hydraulic self-healing command in the control command sequence. At the same time, the damping force of the shock absorber is adjusted in real time through the suspension continuous damping control system CDC to control the vehicle pitch angle change rate to meet the preset change rate threshold. If the communication bus of the current new energy vehicle is detected to be congested, the transmission priority of non-critical signals is dynamically downgraded according to the communication self-healing command in the control command sequence to ensure the transmission delay of critical steering or braking commands.

[0035] This embodiment, through the above-described scheme, synchronously collects current new energy vehicle bus network data, controller internal variables, physical sensor data, and environmental perception data in real time, and constructs a spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion. Based on the spatiotemporal feature matrix, fault prediction and root cause localization are performed to determine the fault type, fault occurrence probability, and fault root cause location, and corresponding self-healing decision instructions are generated. According to the self-healing decision instructions, a multi-objective optimization algorithm coordinates and controls multiple subsystems of the chassis to execute self-healing strategies. This enables comprehensive perception and accurate characterization of the chassis system state, significantly improving the ability to identify early fault characteristics. Fault prediction and root cause localization based on the fused spatiotemporal feature matrix can accurately determine the fault type, occurrence probability, and root cause, avoiding misjudgments and omissions caused by data fragmentation in traditional methods. By coordinating and controlling multiple subsystems of the chassis to execute self-healing strategies through a multi-objective optimization algorithm, the global actuator resources are fully utilized for collaborative compensation, enabling the vehicle to maintain safe driving performance under single or multiple fault conditions. This significantly improves the safety, reliability, and user experience of the intelligent chassis system of new energy vehicles, and increases the speed and efficiency of fault prediction and self-healing in automotive intelligent chassis.

[0036] Based on the above hardware structure, an embodiment of the vehicle intelligent chassis fault prediction and self-healing method of the present invention is proposed.

[0037] Reference Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the vehicle intelligent chassis fault prediction and self-healing method of the present invention.

[0038] In the first embodiment, the vehicle intelligent chassis fault prediction and self-healing method includes the following steps: Step S10: Real-time synchronous collection of bus network data, controller internal variables, physical sensor data and environmental perception data of the current new energy vehicle, and construction of spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion.

[0039] It should be noted that after real-time synchronous collection of bus network data, controller internal variables, physical sensor data and environmental perception data of current new energy vehicles, a corresponding spatiotemporal feature matrix can be constructed through spatiotemporal alignment and feature-level fusion, providing a comprehensive and unified data foundation for subsequent accurate fault prediction and root cause analysis.

[0040] Step S20: Based on the spatiotemporal feature matrix, perform fault prediction and root cause localization, determine the fault type, fault occurrence probability and fault root cause location, and generate corresponding self-healing decision instructions.

[0041] It should be understood that fault prediction and root cause localization are performed based on the spatiotemporal feature matrix, thereby determining the fault type, fault occurrence probability, and fault root cause location. Corresponding self-healing decision instructions can be generated according to the fault type, fault occurrence probability, and fault root cause location, which significantly improves the accuracy of fault prediction and the precision of root cause localization, and provides a scientific basis for subsequent collaborative self-healing control.

[0042] Step S30: According to the self-healing decision instruction, coordinate and control the multi-subsystem of the chassis to execute the self-healing strategy through a multi-objective optimization algorithm.

[0043] It is understood that, according to the self-healing decision instructions, the self-healing strategy can be executed by coordinating the control of multiple subsystems of the chassis through a multi-objective optimization algorithm, which can effectively avoid the risk of system-level failure caused by a single component failure and significantly improve the safety and reliability of new energy vehicles.

[0044] This embodiment, through the above-described scheme, synchronously collects current new energy vehicle bus network data, controller internal variables, physical sensor data, and environmental perception data in real time, and constructs a spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion. Based on the spatiotemporal feature matrix, fault prediction and root cause localization are performed to determine the fault type, fault occurrence probability, and fault root cause location, and corresponding self-healing decision instructions are generated. According to the self-healing decision instructions, a multi-objective optimization algorithm coordinates and controls multiple subsystems of the chassis to execute self-healing strategies. This enables comprehensive perception and accurate characterization of the chassis system state, significantly improving the ability to identify early fault characteristics. Fault prediction and root cause localization based on the fused spatiotemporal feature matrix can accurately determine the fault type, occurrence probability, and root cause, avoiding misjudgments and omissions caused by data fragmentation in traditional methods. By coordinating and controlling multiple subsystems of the chassis to execute self-healing strategies through a multi-objective optimization algorithm, the global actuator resources are fully utilized for collaborative compensation, enabling the vehicle to maintain safe driving performance under single or multiple fault conditions. This significantly improves the safety, reliability, and user experience of the intelligent chassis system of new energy vehicles, and increases the speed and efficiency of fault prediction and self-healing in automotive intelligent chassis.

[0045] Furthermore, Figure 3 This is a flowchart illustrating the second embodiment of the automotive intelligent chassis fault prediction and self-healing method of the present invention, as shown below. Figure 3 As shown, based on the first embodiment, a second embodiment of the automotive intelligent chassis fault prediction and self-healing method of the present invention is proposed. In this embodiment, step S10 specifically includes the following steps: Step S11: Collect the bus network data of the current new energy vehicle in real time through the vehicle communication bus protocol, collect the internal variables of the controller synchronously through the diagnostic interface or internal memory of the electronic control unit (ECU), collect the physical sensor data synchronously through the measuring device, and collect the environmental perception data synchronously through the external sensing device.

[0046] It should be noted that the bus network data of the current new energy vehicle can be collected in real time through the vehicle communication bus protocol, the internal variables of the controller can be collected synchronously through the diagnostic interface or internal memory of the electronic control unit (ECU), the physical sensor data can be collected synchronously through the measuring device, and the environmental perception data of the current external environment of the new energy vehicle can be collected synchronously through the external sensing device.

[0047] In its implementation, key operating parameters transmitted through the vehicle's internal network, including vehicle speed, steering angle, and braking pressure, are acquired in real time via vehicle communication bus protocols such as CAN, LIN, and FlexRay. Diagnostic protocols such as XCP-on-Ethernet or KWP2000 are used to synchronously acquire implicit state variables and intermediate calculation results within the controller, such as proportional-integral-derivative (PID) control parameters and internal diagnostic counters, through the diagnostic interface of the electronic control unit (ECU) or by directly reading its internal memory. High-precision measuring devices such as accelerometers, gyroscopes, and wheel speed sensors are used to synchronously acquire sensor data reflecting the vehicle's actual physical state. Simultaneously, external sensing devices such as cameras, millimeter-wave radar, and lidar are used to synchronously acquire environmental perception data, including road curvature, obstacle positions, and traffic signs. This four-dimensional data synchronous acquisition mechanism ensures that data from different sources and of different types are strictly aligned on a time reference, avoiding data distortion caused by differences in acquisition timing.

[0048] Step S12: Perform spatiotemporal alignment and feature-level fusion on the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data to construct a spatiotemporal feature matrix.

[0049] It is understood that after performing spatiotemporal alignment and feature-level fusion on the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data, a spatiotemporal feature matrix can be constructed.

[0050] Furthermore, step S12 specifically includes the following steps: The IEEE 1588 Precise Network Time Protocol (PTP) is used to synchronize the time deviations of the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data. Based on the vehicle kinematics model and coordinate transformation algorithm, spatial alignment of various data sources is performed, and sensor data distributed at different locations of the vehicle are uniformly mapped to a global coordinate system with the vehicle's center of mass as the origin. The system extracts periodic features and abnormal fluctuation indicators from the bus network data, performs state space reconstruction and outlier detection on the internal variables of the controller, applies wavelet packet decomposition to extract energy entropy features from the physical sensor data and calculates time-frequency domain statistics, and performs target recognition and scene semantic segmentation on the environmental perception data to extract road curvature and obstacle position features. The processed bus network data, controller internal variables, physical sensor data, and environmental perception data are organized according to a unified time series to construct a spatiotemporal feature matrix with N×M dimensions, where N represents the time series length and M represents the fused feature dimension.

[0051] It should be understood that, firstly, in the time dimension, the IEEE 1588 Precision Time Protocol (PTP) is used to perform high-precision time synchronization of bus network data, controller internal variables, physical sensor data, and environmental perception data. This protocol achieves microsecond-level (<10μs) time deviation correction through master-slave clock mechanism and hardware timestamp technology, effectively solving the timing misalignment problem caused by inconsistent sampling periods of different data sources in traditional systems (such as 10ms for CAN bus, 1ms for IMU sensor, and 33ms for camera), ensuring strict alignment of multi-source data on the time axis, and laying a precise time reference for subsequent timing analysis. Secondly, in the spatial dimension, spatial alignment processing is performed on each data source based on the vehicle kinematics model and coordinate transformation algorithm. By establishing the transformation relationship between the vehicle coordinate system and the local coordinate systems of each sensor, the data collected from different locations such as the front millimeter-wave radar, the middle inertial measurement unit (IMU), the rear camera, and the wheel hub sensors are uniformly mapped to the global coordinate system with the vehicle's center of mass as the origin. This eliminates spatial information distortion caused by differences in sensor installation positions, enabling the system to accurately reflect the overall motion state of the vehicle. In the feature extraction stage, specialized processing methods are adopted for the four types of data: For bus network data, periodic features of the signal are extracted through autocorrelation analysis and spectrum analysis, and abnormal fluctuation indicators are calculated using sliding window standard deviation to identify potential communication anomalies; For controller internal variables, state space reconstruction technology is applied to expand the one-dimensional time series into a multi-dimensional phase space, and the Local Outlier Factor (LOF) algorithm is combined to detect outliers and reveal hidden faults inside the ECU; For physical sensor data, wavelet packet decomposition technology is used to decompose the signal into multi-frequency band components, and the energy entropy of each frequency band is calculated as a fault-sensitive feature, while time-domain statistics (mean, variance, kurtosis) and frequency-domain features (dominant frequency, energy concentration) are extracted; For environmental perception data, deep learning models are used for target recognition and scene semantic segmentation to extract high-level semantic features such as road curvature, lane position, obstacle distance, and relative speed.Finally, the processed multi-source features are organized according to a unified time series to construct a spatiotemporal feature matrix with N×M dimensions, where N represents the time series length (usually 100-500 time points, covering 0.5-2.5 seconds of vehicle dynamics) and M represents the fused feature dimension (usually 50-200 features). This matrix not only contains the current state information of each subsystem, but also implicitly contains the dynamic coupling relationship between systems and the fault propagation path, forming a structured data representation with clear physical meaning. This effectively overcomes the limitations of fragmented multi-source data analysis in traditional fault diagnosis and significantly improves the system's ability to identify early fault features and the accuracy of root cause localization.

[0052] This embodiment, through the above-described scheme, collects real-time bus network data of the current new energy vehicle via the vehicle communication bus protocol, synchronously collects internal variables of the controller through the diagnostic interface or internal memory of the electronic control unit (ECU), synchronously collects physical sensor data through a measuring device, and synchronously collects environmental perception data through external sensing devices. The bus network data, the controller internal variables, the physical sensor data, and the environmental perception data are spatiotemporally aligned and feature-level fused to construct a spatiotemporal feature matrix. This enables comprehensive perception and accurate characterization of the new energy vehicle chassis system status, significantly improving the accuracy and timeliness of fault prediction, greatly enhancing the safety and reliability of the new energy vehicle intelligent chassis system, and providing strong support for the stable operation of the vehicle under complex working conditions.

[0053] Furthermore, Figure 4 This is a flowchart illustrating the third embodiment of the automotive intelligent chassis fault prediction and self-healing method of the present invention, as shown below. Figure 4 As shown, based on the first embodiment, a third embodiment of the automotive intelligent chassis fault prediction and self-healing method of the present invention is proposed. In this embodiment, step S20 specifically includes the following steps: Step S21: When performing fault prediction and root cause localization based on the spatiotemporal feature matrix, an LSTM-Transformer hybrid neural network architecture is used to process the input spatiotemporal feature matrix to perform multi-level feature extraction of temporal features, and output the fault probability distribution and fault type prediction of each key component within a future preset time period. In the LSTM-Transformer hybrid neural network architecture, the LSTM layer captures short-term dynamic features, and the Transformer layer captures long-term dependencies through a self-attention mechanism.

[0054] It should be noted that the LSTM-Transformer hybrid neural network architecture is used to perform deep analysis of the spatiotemporal feature matrix, achieving high-precision prediction and accurate localization of faults in the chassis system of new energy vehicles. This hybrid architecture fully leverages the complementary advantages of the two deep learning models, Long Short Term Memory (LSTM) and Transformer: the LSTM layer effectively captures short-term dynamic features and local temporal patterns during vehicle operation through its unique gating mechanism, and can keenly identify short-term fault symptoms such as sudden changes in sensor signals and instantaneous vibration anomalies; while the Transformer layer uses the self-attention mechanism to analyze the global dependencies between time steps in the feature matrix, effectively capturing the long-term degradation trend of the vehicle system and the fault evolution law across time scales, such as progressive fault characteristics such as bearing wear and electronic component aging.

[0055] In the specific implementation, the spatiotemporal feature matrix is ​​first processed by a multi-layer LSTM network for local temporal feature extraction. Then, the extracted features are input into the Transformer encoder. The long-distance dependency between features is established through a multi-head attention mechanism. Finally, the fault probability distribution and specific fault type prediction of each key component (e.g., steering gear, brake caliper, suspension damper, etc.) are output through a fully connected layer within a future preset time period (usually 200ms, but other values ​​are also possible, and this embodiment does not limit this).

[0056] Step S22: Simultaneously construct a system dependency model based on a graph neural network, and determine the location of the root cause of the fault based on the system dependency model.

[0057] It is understandable that a system dependency model based on a graph neural network can be constructed, and the root cause of the failure can be determined based on the system dependency model.

[0058] Furthermore, step S22 includes the following steps: Simultaneously, a system dependency model based on graph neural networks is constructed, abstracting each subsystem of the current new energy vehicle chassis as graph nodes and the signal interaction between each subsystem as edges. The propagation path of abnormal patterns in the feature matrix in the system graph is analyzed through anomaly propagation algorithm, and the location of the root cause of the fault is determined by combining Bayesian inference.

[0059] In the specific implementation, a system dependency model based on a Graph Neural Network (GNN) was constructed to accurately locate the root cause of chassis failures in new energy vehicles. Specifically, the various subsystems of the chassis (e.g., steering system, braking system, suspension system, and drive system) were abstracted as nodes in a graph structure, and the actual signal interactions and physical coupling relationships between subsystems were abstracted as edges, thus constructing a directed weighted graph reflecting the true topology of the vehicle system. Based on this, an anomaly propagation algorithm was designed to simulate the propagation process of fault signals in the system graph. By analyzing how the anomaly patterns detected in the spatiotemporal feature matrix propagate along the edges of the graph structure, a mapping relationship between the anomaly propagation path and the fault source was established. Simultaneously, a Bayesian inference framework was introduced. Based on prior knowledge and real-time observation data, the posterior probability of each node as a root cause of the fault was calculated. Finally, by comprehensively analyzing the anomaly propagation path and using Bayesian probabilistic inference, the most likely location of the root cause of the fault was determined, effectively distinguishing between a chain reaction caused by a single fault source and multiple independent faults. This approach can accurately identify coupled faults across subsystems (such as steering system failure leading to electronic stability issues). The delayed intervention of the ESP program improved the accuracy of root cause localization.

[0060] Step S23: Based on the fault type, the fault occurrence probability, and the fault root cause location, generate a self-healing decision instruction containing the optimal actuator combination scheme, execution priority, and safety constraints according to the preset fault strategy mapping rules and the preset real-time optimization algorithm.

[0061] It should be understood that a comprehensive assessment of the fault state is formed by comprehensively analyzing three key dimensions: fault type (such as sensor failure, actuator jamming, communication interruption, etc.), fault occurrence probability (quantifying the degree of certainty of the fault), and fault root cause location (the specific component precisely located by a graph neural network). Subsequently, through a pre-set fault strategy mapping rule base, the assessment results are matched with historical fault cases and expert knowledge. This rule base is constructed using decision trees or fuzzy logic and can quickly select a set of applicable self-healing strategies based on different combinations of fault characteristics. On this basis, a pre-set real-time optimization algorithm (such as model predictive control or dynamic programming algorithm) is applied, considering the current vehicle... The system considers vehicle operating conditions (vehicle speed, steering angle, road adhesion coefficient, etc.) and system constraints to quantitatively evaluate and rank candidate strategies, calculating the optimal actuator combination scheme under the current operating conditions (such as a coordinated scheme that simultaneously activates differential braking, adjusts suspension stiffness, and regulates motor torque). The final self-healing decision command not only includes specific control parameters and execution sequences, but also clearly defines the execution priority of each subsystem operation (ensuring that critical safety functions are executed first) and dynamically adjusted safety constraint boundary conditions (such as adjusting the upper limit of tire slip ratio in real time according to the severity of the fault), so that the command not only meets the fault compensation requirements, but also strictly guarantees the driving stability of the vehicle during the self-healing process.

[0062] Accordingly, step S30 specifically includes the following steps: The current fault type information contained in the self-healing decision instruction is analyzed, and the chassis system combination to be activated and the corresponding safety boundary are determined based on the current fault type information. Based on the chassis system assembly and the safety boundary, a multi-actuator cooperative control model is established, encompassing the steering system, braking system, suspension system, and drive system of the current new energy vehicle. The multi-objective optimization function is determined using the following formula:

[0063] in, For multi-objective optimization functions, , , and The weighting coefficients are dynamically adjusted based on the severity of the fault and the vehicle's operating status. To track errors, For energy consumption, For comfort; By applying safety constraints during the solution of the multi-objective optimization function, it is ensured that the current new energy vehicle always remains within the stability domain; The optimal control quantity obtained from the optimization solution is assigned to the corresponding actuator, generating a control command sequence for each subsystem; The control chassis multi-subsystem is controlled to execute a self-healing strategy according to the control command sequence.

[0064] It should be noted that the self-healing decision instructions contain fault type information (e.g., steering sensor failure, insufficient braking pressure, etc.), and the chassis system combinations that need to work together (e.g., steering and braking system linkage) and corresponding safety boundary parameters are dynamically determined based on the fault characteristics. Subsequently, a multi-actuator cooperative control model including steering, braking, suspension, and drive systems is established based on the determined subsystem combinations, and a multi-objective optimization function is constructed. The optimal balance between multiple objectives is achieved by optimizing this function. When solving the optimization problem, the system strictly applies safety constraints such as tire slip ratio and yaw rate to ensure that the vehicle is always within the stable control domain. After optimization, the calculated optimal control quantity is distributed to each actuator in real time to generate a coordinated subsystem control instruction sequence. Finally, the chassis multi-subsystems execute the self-healing strategy synchronously according to the instruction sequence. For example, in the event of a steering system failure, differential braking and suspension adjustment are used to compensate for the steering torque, so that the vehicle maintains a safe driving state.

[0065] Furthermore, the step of controlling the control chassis multi-subsystem to execute a self-healing strategy according to the control command sequence specifically includes the following steps: When the steering torque sensor of the current new energy vehicle is detected to be faulty, the system automatically switches to the redundant sensor channel according to the steering self-healing command in the control command sequence, and generates auxiliary steering torque through differential braking within a preset steering response time. When a brake fluid pressure deficiency fault is detected in the current new energy vehicle, the drive motor provides regenerative braking force compensation within a preset motor response time according to the brake hydraulic self-healing command in the control command sequence. At the same time, the damping force of the shock absorber is adjusted in real time through the suspension continuous damping control system CDC to control the vehicle pitch angle change rate to meet the preset change rate threshold. If the communication bus of the current new energy vehicle is detected to be congested, the transmission priority of non-critical signals is dynamically downgraded according to the communication self-healing command in the control command sequence to ensure the transmission delay of critical steering or braking commands.

[0066] It should be noted that when a steering torque sensor failure is detected, the system immediately and automatically switches to the redundant sensor channel according to the control command sequence to ensure the continuity of the steering signal. Simultaneously, within a preset very short time (typically <100ms, but other values ​​can be set, and this embodiment does not impose any limitations), it generates 0-200Nm of auxiliary steering torque through differential braking of the left and right wheels to compensate for the steering assist loss caused by the sensor failure and maintain the vehicle's controllable steering capability. When a brake fluid pressure deficiency is detected, the system, according to the brake hydraulic pressure self-healing command, drives the motor to provide 0-30% regenerative braking force compensation within a preset response time (typically <50ms, but other values ​​can be set, and this embodiment does not impose any limitations) to fill the braking force gap in the hydraulic braking system. Simultaneously, through continuous damping control of the suspension... The DampingControl (CDC) system adjusts the damping force of the shock absorbers in real time to suppress the forward pitching phenomenon during braking, ensuring that the rate of change of the vehicle pitch angle is strictly controlled within a safe threshold, thus guaranteeing braking stability and ride comfort. When communication bus congestion is detected, the system intelligently identifies signal priorities based on communication self-healing instructions, dynamically downgrading the transmission priority of non-critical signals (such as air conditioning control and entertainment systems) to ensure that the transmission delay of critical safety signals such as steering and braking commands is controlled within 10ms, maintaining the core functions of the chassis control system. These three self-healing strategies together constitute a complete fault response system, realizing the transformation from a single fault point to global system coordination, effectively avoiding the risk of system-level failure caused by local faults, and significantly improving the safety boundary and user experience of new energy vehicles.

[0067] In the specific implementation, see Figure 5 , Figure 5 This is a schematic diagram of the system framework corresponding to the intelligent chassis fault prediction and self-healing method of the present invention, as shown below. Figure 5 As shown, the overall system comprises three core subsystems: a multi-source signal fusion layer, a fault prediction engine, and a collaborative self-healing controller. in, 1. The functions of the Multi-source Fusion Layer are as follows: 1) Data collection: Real-time synchronous acquisition of four-dimensional data: Bus network: CAN FD / Ethernet TSN messages (timestamp accuracy ≤1μs); Internally, the controller obtains intermediate variables of the control algorithm (such as PID output and state machine) through XCP-on-Ethernet. Physical sensors: high-precision strain gauges (steering column torque), MEMS accelerometers (vehicle attitude), pressure sensors (brake wheel cylinders); Environmental perception: camera lane departure, radar target distance.

[0068] 2) Fusion method: Employing spatiotemporal alignment + feature-level fusion: Clock synchronization: PTP protocol unifies the time base of each source (deviation <10μs); Feature extraction: Wavelet packet decomposition is performed on the vibration signal (energy entropy is extracted), and sliding window statistics (mean / variance) are performed on the control variables. Fusion matrix: Construct a spatiotemporal feature matrix to input into the prediction engine.

[0069] 2. Fault Prognostics Engine 1) Predictive Model: Functions are as follows: Short-term prediction: Based on the LSTM-Transformer hybrid network, the input is a temporal feature matrix, and the output is the failure probability in the next 200ms (e.g., sensor drift probability > 90%). Long-term prediction: Combining reinforcement learning (PPO algorithm) with physical degradation models (such as the thermal aging equation of motor windings) to predict the remaining life of components.

[0070] 2) Fault location: Graph neural networks (GNNs) are used to model subsystem dependencies, and the root cause is located by reverse propagation of node anomalies (such as communication packet loss → steering command delay → ESP activation timeout). 3. Collaborative Self-healing Controller As shown in Table 1 below: Table 1. Example table of self-healing strategy library:

[0071] 4. Dynamic weight allocation Design a multi-objective optimization function to solve for the optimal actuator combination (such as steering angle + motor torque + braking force distribution) in real time. minJ = α·tracking error + β·energy consumption + γ·comfort Constraints: Tire slip ratio < safety threshold, yaw rate < instability boundary In the specific implementation, see Figure 6 , Figure 6 This is a schematic diagram of the dynamic optimization algorithm in the vehicle intelligent chassis fault prediction and self-healing method of the present invention, as shown below. Figure 6As shown, firstly, multi-source signal fusion technology is used to achieve precise fault location. After determining the specific location and cause of the fault, the fault level is assessed according to the ISO 26262 standard, namely the Automotive Safety Integrity Level (ASIL). For high-risk faults at ASIL-C / D level, a hardware reconfiguration mechanism is immediately initiated, utilizing nanoscale metal-oxide-semiconductor field-effect transistors (MOTEs). The MOSFET (Transistor for MOSFETs) completes switching within nanoseconds to achieve power transfer, dynamically redirecting the load of the faulty power module to the healthy module. Simultaneously, redundant sensors are activated (switching time <50ns), and a healthy phase current shunt technique is used to redistribute the three-phase winding current, ensuring stable total output torque. For lower-risk faults at ASIL-A / B levels, software fault-tolerance mechanisms are primarily used. In the control optimization phase, the system estimates torque based on the motor_current→vehicle_speed relationship and constructs a multi-objective optimization function minJ=α·tracking error+β·energy consumption, where α and β are weighting coefficients dynamically adjusted according to the severity of the fault and the vehicle's operating status. Kalman filtering compensation technology is applied to the signal processing phase during power transfer, correcting current and voltage in real time and eliminating transient interference during switching. The entire optimization process is solved in real time under safety constraints, generating the optimal control command sequence and achieving closed-loop control from fault detection to self-healing execution. This ensures the vehicle maintains safe driving performance even under fault conditions, significantly improving the safety and reliability of the intelligent chassis system for new energy vehicles.

[0072] In practice, the following is an example: Scenario: Predicting brake pressure sensor drift during high-speed cruising; Signal fusion layer: A sudden drop in the variance of the brake master cylinder pressure signal was detected (abnormal feature), and false judgments were ruled out by combining it with camera data (lane keeping is normal); Prediction engine: The LSTM model outputs a 92% probability of sensor failure within the next 150ms; Self-healing controller: Enable redundant pressure sensors; Activate the regenerative braking compensation force of the drive motor; Adjusting suspension stiffness to suppress brake dive; Results: Vehicle deceleration maintained at the target value ±0.2g, vehicle pitch angle ≤1.5°, and no driver intervention required throughout the process.

[0073] This embodiment has the following advantages over the prior art: 1) Improved prediction accuracy: Fault detection rate ≥95% (compared to the current 80%), false alarm rate ≤3%; 2) Collaborative self-healing timeliness: From prediction to execution ≤50ms (10 times faster than cloud solutions); 3) Safety boundary guarantee: Under the condition of superimposed dual failures (such as steering failure + communication delay), maintain 80% of the original performance indicators; 4) Cost reduction: Reduce redundant hardware requirements by 30% and reduce warranty costs through predictive maintenance.

[0074] This embodiment can achieve early prediction and accurate location of multimodal faults (hardware / software / communication); construct a collaborative self-healing control architecture across steering, braking, suspension, and drive subsystems; complete the "prediction-decision-execution" closed loop within ≤50ms, and maintain the vehicle stability domain under failure conditions.

[0075] In this embodiment, when performing fault prediction and root cause localization using the spatiotemporal feature matrix, an LSTM-Transformer hybrid neural network architecture is employed to process the input spatiotemporal feature matrix through multi-level feature extraction of temporal features. This outputs the fault probability distribution and fault type prediction for each key component within a preset future time period. Specifically, in the LSTM-Transformer hybrid neural network architecture, the LSTM layer captures short-term dynamic features, while the Transformer layer captures long-term dependencies through a self-attention mechanism. Simultaneously, a system dependency model based on a graph neural network is constructed, and the location of the fault root cause is determined according to this model. Based on the fault type, the fault occurrence probability, and the fault root cause location, a self-healing decision instruction containing the optimal actuator combination scheme, execution priority, and safety constraints is generated according to the preset fault strategy mapping rules and preset real-time optimization algorithm. This significantly improves the fault prediction accuracy and self-healing capability of the new energy vehicle chassis system, accurately predicts potential faults of key components, and precisely locates the fault root cause location. It greatly shortens the response time from fault identification to self-healing execution, realizing a fundamental shift in fault handling from passive alarm to active repair. This significantly improves the safety performance and reliability of new energy vehicles under complex working conditions, providing drivers and passengers with a safer and more comfortable driving experience.

[0076] Accordingly, the present invention further provides a self-healing device for predicting and resolving faults in an intelligent automotive chassis.

[0077] Reference Figure 7 , Figure 7 This is a functional block diagram of the first embodiment of the intelligent chassis fault prediction and self-healing device of the present invention.

[0078] In a first embodiment of the intelligent chassis fault prediction and self-healing device of the present invention, the intelligent chassis fault prediction and self-healing device includes: The spatiotemporal feature matrix construction module 10 is used to collect bus network data, controller internal variables, physical sensor data and environmental perception data of the current new energy vehicle in real time, and construct the spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion.

[0079] The self-healing decision instruction generation module 20 is used to perform fault prediction and root cause localization based on the spatiotemporal feature matrix, determine the fault type, fault occurrence probability and fault root cause location, and generate corresponding self-healing decision instructions.

[0080] The strategy execution module 30 is used to coordinate and control the execution of the self-healing strategy by the chassis multi-subsystems through a multi-objective optimization algorithm according to the self-healing decision instruction.

[0081] The spatiotemporal feature matrix construction module 10 is also used to collect bus network data of the current new energy vehicle in real time through the vehicle communication bus protocol, synchronously collect internal variables of the controller through the diagnostic interface or internal memory of the electronic control unit (ECU), synchronously collect physical sensor data through the measuring device, and synchronously collect environmental perception data through external sensing devices; and perform spatiotemporal alignment and feature-level fusion on the bus network data, the internal variables of the controller, the physical sensor data, and the environmental perception data to construct and generate a spatiotemporal feature matrix.

[0082] The spatiotemporal feature matrix construction module 10 is further used to synchronize the time deviations of the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data using the IEEE 1588 Precise Network Time Protocol (PTP); to spatially align the data sources based on the vehicle kinematics model and coordinate transformation algorithm, and to uniformly map the sensor data distributed at different locations on the vehicle to a global coordinate system with the vehicle's center of mass as the origin; to extract signal periodicity features and abnormal fluctuation indicators from the bus network data; to reconstruct the state space and detect outliers from the controller internal variables; to apply wavelet packet decomposition to extract energy entropy features and calculate time-frequency domain statistics from the physical sensor data; and to perform target recognition and scene semantic segmentation on the environmental perception data to extract road curvature and obstacle position features; and to organize the processed bus network data, the controller internal variables, the physical sensor data, and the environmental perception data according to a unified time series to construct a spatiotemporal feature matrix with N×M dimensions, where N represents the time series length and M represents the fused feature dimension.

[0083] The self-healing decision instruction generation module 20 is further configured to, when performing fault prediction and root cause localization based on the spatiotemporal feature matrix, use an LSTM-Transformer hybrid neural network architecture to process the input spatiotemporal feature matrix, perform multi-level feature extraction of temporal features, and output the fault probability distribution and fault type prediction of each key component within a future preset time period. In the LSTM-Transformer hybrid neural network architecture, the LSTM layer captures short-term dynamic features, and the Transformer layer captures long-term dependencies through a self-attention mechanism. Simultaneously, a system dependency model based on a graph neural network is constructed, and the location of the fault root cause is determined according to the system dependency model. Based on the fault type, the fault occurrence probability, and the location of the fault root cause, a self-healing decision instruction containing the optimal actuator combination scheme, execution priority, and safety constraints is generated according to preset fault strategy mapping rules and preset real-time optimization algorithms.

[0084] The self-healing decision instruction generation module 20 is also used to simultaneously construct a system dependency model based on a graph neural network, abstract each subsystem of the current new energy vehicle chassis as a graph node, abstract the signal interaction between each subsystem as an edge, analyze the propagation path of the abnormal pattern in the feature matrix in the system graph through an anomaly propagation algorithm, and determine the location of the root cause of the fault by combining Bayesian inference.

[0085] The strategy execution module 30 is also used to parse the current fault type information contained in the self-healing decision instruction, and determine the chassis system combination to be activated and the corresponding security boundary based on the current fault type information. Based on the chassis system assembly and the safety boundary, a multi-actuator cooperative control model is established, encompassing the steering system, braking system, suspension system, and drive system of the current new energy vehicle. The multi-objective optimization function is determined using the following formula:

[0086] in, For multi-objective optimization functions, , , and The weighting coefficients are dynamically adjusted based on the severity of the fault and the vehicle's operating status. To track errors, For energy consumption, For comfort; By applying safety constraints during the solution of the multi-objective optimization function, it is ensured that the current new energy vehicle always remains within the stability domain; The optimal control quantity obtained from the optimization solution is assigned to the corresponding actuator, generating a control command sequence for each subsystem; The control chassis multi-subsystem is controlled to execute a self-healing strategy according to the control command sequence.

[0087] The strategy execution module 30 is further configured to automatically switch to a redundant sensor channel according to the steering self-healing command in the control command sequence when the steering torque sensor of the current new energy vehicle is detected to be faulty, and generate auxiliary steering torque through differential braking within a preset steering response time; when the brake fluid pressure of the current new energy vehicle is detected to be insufficient, drive the motor to provide regenerative braking force compensation within a preset motor response time according to the brake hydraulic self-healing command in the control command sequence, and simultaneously adjust the damping force of the shock absorber in real time through the suspension continuous damping control system CDC to control the vehicle pitch angle change rate to meet a preset change rate threshold; and when the communication bus of the current new energy vehicle is detected to be congested, dynamically downgrade the transmission priority of non-critical signals according to the communication self-healing command in the control command sequence to ensure the transmission delay of steering or braking critical commands.

[0088] The steps for implementing each functional module of the intelligent chassis fault prediction and self-healing device can be referred to in the various embodiments of the intelligent chassis fault prediction and self-healing method of the present invention, and will not be repeated here.

[0089] Furthermore, this embodiment of the invention also proposes a storage medium storing a vehicle intelligent chassis fault prediction and self-healing program. When the vehicle intelligent chassis fault prediction and self-healing program is executed by a processor, it implements the operations described in the above-described vehicle intelligent chassis fault prediction and self-healing method embodiment.

[0090] Those skilled in the art will understand that all or part of the steps in the methods described above can be implemented by a program instructing related hardware. The program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium is a computer-readable storage medium, including: USB flash drive, mobile hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and other media that can store program code.

[0091] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0092] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0093] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for predicting and self-healing faults in an intelligent automotive chassis, characterized in that, The automotive intelligent chassis fault prediction and self-healing method includes: Real-time synchronous collection of bus network data, controller internal variables, physical sensor data and environmental perception data of current new energy vehicles, and construction of spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion; Based on the spatiotemporal feature matrix, fault prediction and root cause localization are performed to determine the fault type, fault occurrence probability and fault root cause location, and corresponding self-healing decision instructions are generated. According to the self-healing decision instruction, the self-healing strategy is executed by multiple subsystems of the chassis through a multi-objective optimization algorithm.

2. The method for predicting and self-healing faults in an intelligent automotive chassis as described in claim 1, characterized in that, The real-time synchronous acquisition of current new energy vehicle bus network data, controller internal variables, physical sensor data, and environmental perception data, and the construction of a spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion, includes: Real-time acquisition of bus network data of current new energy vehicles through vehicle communication bus protocol, synchronous acquisition of internal variables of controller through diagnostic interface or internal memory of electronic control unit (ECU), synchronous acquisition of physical sensor data through measuring device, and synchronous acquisition of environmental perception data through external sensing device. The bus network data, the controller internal variables, the physical sensor data, and the environmental perception data are spatiotemporally aligned and feature-level fused to construct a spatiotemporal feature matrix.

3. The method for predicting and self-healing automotive intelligent chassis faults as described in claim 2, characterized in that, The step of performing spatiotemporal alignment and feature-level fusion on the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data to construct a spatiotemporal feature matrix includes: The IEEE 1588 Precise Network Time Protocol (PTP) is used to synchronize the time deviations of the bus network data, the controller internal variables, the physical sensor data, and the environmental perception data. Based on the vehicle kinematics model and coordinate transformation algorithm, spatial alignment of various data sources is performed, and sensor data distributed at different locations of the vehicle are uniformly mapped to a global coordinate system with the vehicle's center of mass as the origin. The system extracts periodic features and abnormal fluctuation indicators from the bus network data, performs state space reconstruction and outlier detection on the internal variables of the controller, applies wavelet packet decomposition to extract energy entropy features from the physical sensor data and calculates time-frequency domain statistics, and performs target recognition and scene semantic segmentation on the environmental perception data to extract road curvature and obstacle position features. The processed bus network data, controller internal variables, physical sensor data, and environmental perception data are organized according to a unified time series to construct a spatiotemporal feature matrix with N×M dimensions, where N represents the time series length and M represents the fused feature dimension.

4. The method for predicting and self-healing automotive intelligent chassis faults as described in claim 1, characterized in that, The process of fault prediction and root cause localization based on the spatiotemporal feature matrix, determining the fault type, fault occurrence probability, and fault root cause location, and generating corresponding self-healing decision instructions includes: When performing fault prediction and root cause localization based on the spatiotemporal feature matrix, an LSTM-Transformer hybrid neural network architecture is used to process the input spatiotemporal feature matrix to perform multi-level feature extraction of temporal features, and output the fault probability distribution and fault type prediction of each key component within a future preset time period. In the LSTM-Transformer hybrid neural network architecture, the LSTM layer captures short-term dynamic features, and the Transformer layer captures long-term dependencies through a self-attention mechanism. Simultaneously, a system dependency model based on a graph neural network is constructed, and the location of the root cause of the failure is determined based on the system dependency model; Based on the fault type, the fault occurrence probability, and the fault root cause location, a self-healing decision instruction containing the optimal actuator combination scheme, execution priority, and safety constraints is generated according to the preset fault strategy mapping rules and the preset real-time optimization algorithm.

5. The method for predicting and self-healing faults in an intelligent automotive chassis as described in claim 4, characterized in that, The simultaneous construction of a system dependency model based on a graph neural network, and the determination of the root cause location of the fault based on the system dependency model, includes: Simultaneously, a system dependency model based on graph neural networks is constructed, abstracting each subsystem of the current new energy vehicle chassis as graph nodes and the signal interaction between each subsystem as edges. The propagation path of abnormal patterns in the feature matrix in the system graph is analyzed through anomaly propagation algorithm, and the location of the root cause of the fault is determined by combining Bayesian inference.

6. The method for predicting and self-healing faults in an intelligent automotive chassis as described in claim 1, characterized in that, The step of coordinating and controlling the execution of the self-healing strategy by multiple subsystems of the chassis through a multi-objective optimization algorithm according to the self-healing decision instruction includes: The current fault type information contained in the self-healing decision instruction is analyzed, and the chassis system combination to be activated and the corresponding safety boundary are determined based on the current fault type information. Based on the chassis system assembly and the safety boundary, a multi-actuator cooperative control model is established, encompassing the steering system, braking system, suspension system, and drive system of the current new energy vehicle. The multi-objective optimization function is determined using the following formula: in, For multi-objective optimization functions, , , and The weighting coefficients are dynamically adjusted based on the severity of the fault and the vehicle's operating status. To track errors, For energy consumption, For comfort; By applying safety constraints during the solution of the multi-objective optimization function, it is ensured that the current new energy vehicle always remains within the stability domain; The optimal control quantity obtained from the optimization solution is assigned to the corresponding actuator, generating a control command sequence for each subsystem; The control chassis multi-subsystem is controlled to execute a self-healing strategy according to the control command sequence.

7. The method for predicting and self-healing automotive intelligent chassis faults as described in claim 6, characterized in that, The step of controlling the control chassis multi-subsystem to execute a self-healing strategy according to the control command sequence includes: When the steering torque sensor of the current new energy vehicle is detected to be faulty, the system automatically switches to the redundant sensor channel according to the steering self-healing command in the control command sequence, and generates auxiliary steering torque through differential braking within a preset steering response time. When a brake fluid pressure deficiency fault is detected in the current new energy vehicle, the drive motor provides regenerative braking force compensation within a preset motor response time according to the brake hydraulic self-healing command in the control command sequence. At the same time, the damping force of the shock absorber is adjusted in real time through the suspension continuous damping control system CDC to control the vehicle pitch angle change rate to meet the preset change rate threshold. If the communication bus of the current new energy vehicle is detected to be congested, the transmission priority of non-critical signals is dynamically downgraded according to the communication self-healing command in the control command sequence to ensure the transmission delay of critical steering or braking commands.

8. A self-healing device for predicting and resolving automotive intelligent chassis faults, characterized in that, The intelligent automotive chassis fault prediction and self-healing device includes: The spatiotemporal feature matrix construction module is used to collect bus network data, controller internal variables, physical sensor data and environmental perception data of current new energy vehicles in real time, and construct spatiotemporal feature matrix through spatiotemporal alignment and feature-level fusion. The self-healing decision instruction generation module is used to perform fault prediction and root cause localization based on the spatiotemporal feature matrix, determine the fault type, fault occurrence probability and fault root cause location, and generate corresponding self-healing decision instructions. The strategy execution module is used to coordinate and control the execution of the self-healing strategy by multiple subsystems of the chassis through a multi-objective optimization algorithm according to the self-healing decision instructions.

9. A self-healing device for predicting and resolving automotive intelligent chassis faults, characterized in that: The vehicle intelligent chassis fault prediction and self-healing device includes: a memory, a processor, and a vehicle intelligent chassis fault prediction and self-healing program stored in the memory and executable on the processor, wherein the vehicle intelligent chassis fault prediction and self-healing program is configured to implement the steps of the vehicle intelligent chassis fault prediction and self-healing method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a vehicle intelligent chassis fault prediction and self-healing program, which, when executed by a processor, implements the steps of the vehicle intelligent chassis fault prediction and self-healing method as described in any one of claims 1 to 7.

Images