Automobile fault detection method and system based on fusion of sensor data

Abstract

This application relates to the field of fault detection and control technology, disclosing a method and system for automotive fault detection that fuses sensor data. The method includes: modeling a multi-system sensor network using a vehicle system topology mapping algorithm to obtain a sensor topology correlation matrix; synchronously processing the sensor topology correlation matrix based on vehicle dynamics constraint equations to obtain a synchronized multi-sensor dataset; integrating the synchronized multi-sensor dataset into a fault propagation path perception network to obtain a vehicle system fault feature vector; detecting the fault feature vector based on a vehicle physical model constraint optimizer to obtain a fault state identification result; and tracing and controlling the fault state identification result based on a fault propagation graph model to obtain a system adjustment control signal. This application improves the accuracy of root cause identification for complex coupled faults and the targeted control and adjustment.

Description

Technical Field

[0001] The present application relates to the field of fault detection and control technology, and in particular to a method and system for detecting automobile faults by fusing sensor data. Background Art

[0002] Modern vehicles, as complex mechatronic products, are equipped with numerous sensors to monitor the operating status of various subsystems, including the engine, brakes, steering, and transmission. The massive amounts of data generated by these sensors provide a rich source of information for vehicle fault detection. Traditional vehicle fault detection methods primarily rely on single sensors or simple combinations of multiple sensors for fault identification, determining system anomalies by setting fixed thresholds or employing basic statistical analysis methods. Existing sensor data fusion fault detection technologies include model-based, signal processing-based, and machine learning-based approaches, which can identify common vehicle fault types under specific conditions. Model-based approaches utilize the mathematical model and physical constraints of the vehicle system for fault detection, signal processing-based approaches extract fault signatures through techniques such as frequency domain analysis and wavelet transforms, and machine learning-based approaches employ algorithms such as neural networks and support vector machines to build fault classification models. Current multi-sensor data fusion primarily utilizes traditional fusion algorithms such as weighted averaging, Kalman filtering, and Bayesian reasoning. Fusion quality is enhanced through time alignment and data cleaning during the data preprocessing phase.

[0003] However, existing technologies have significant shortcomings, primarily manifested in a lack of in-depth modeling of the complex coupling relationships between automotive systems and effective analysis of multi-system fault propagation mechanisms. Traditional methods treat each sensor as an independent information source, ignoring the physical connection relationships and functional dependencies between the vehicle's subsystems. This makes it impossible to fully utilize the correlation information between systems to improve fault detection accuracy. Existing time synchronization methods often use a fixed timestamp alignment strategy, failing to consider the impact of changes in vehicle dynamics under different driving conditions on sensor response delays, resulting in a significant decrease in fusion effect under complex conditions. The fault detection process lacks modeling and analysis of the fault propagation paths and cascading effects between multiple systems, making it impossible to accurately identify the root cause of the fault and predict the development trend of the fault. In addition, existing methods lack constraint verification of the vehicle's physical model during feature extraction and fusion, which can easily produce detection results that violate physical laws, affecting the reliability and credibility of fault diagnosis. Summary of the Invention

[0004] The present application provides a vehicle fault detection method and system that fuses sensor data, which are used to improve the accuracy of root cause identification of complex coupled faults and the pertinence of control and adjustment.

[0005] In a first aspect, the present application provides a method for detecting vehicle faults by fusing sensor data, the method comprising:

[0006] The multi-system sensor network is modeled and processed by the automobile system topology mapping algorithm to obtain the sensor topology correlation matrix;

[0007] Performing adaptive time window synchronization processing on the sensor topology correlation matrix according to the vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set;

[0008] Inputting the synchronized multi-sensor data set into a fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector;

[0009] Performing abnormality detection processing on the vehicle system fault feature vector based on the vehicle physical model constraint optimizer to obtain a fault state identification result;

[0010] The fault state identification result is traced and located and a control instruction is generated according to the fault propagation graph model to obtain a system adjustment control signal.

[0011] Optionally, the method of performing correlation modeling on the multi-system sensor network by using an automobile system topology mapping algorithm to obtain a sensor topology correlation matrix includes:

[0012] Through CAN bus network topology scanning, the physical connection relationship of the sensor nodes of the engine, brake, steering and transmission systems is identified and processed to obtain the connection topology diagram between the systems;

[0013] Perform weight assignment processing on the inter-system connection topology diagram according to the functional dependency relationship of each subsystem to obtain a system-related weighted topology diagram;

[0014] Performing sensor node mapping processing on the system association weight topology map based on the sensor three-dimensional spatial coordinates and signal propagation paths to obtain a sensor space distribution topology map;

[0015] Numerical quantification of the sensor spatial distribution topology map is performed using the sensor physical correlation strength and signal propagation delay parameters to obtain a sensor correlation metric matrix;

[0016] Matrix normalization and symmetry adjustment are performed on the sensor correlation metric matrix to obtain the sensor topology correlation matrix.

[0017] Optionally, performing adaptive time window synchronization processing on the sensor topology correlation matrix according to the vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set includes:

[0018] The vehicle speed, acceleration and angular velocity sensors are used to collect and process the current vehicle operating state parameters in real time to obtain the vehicle dynamics state vector;

[0019] Dynamically adjusting and calculating a preset basic time window length according to the vehicle dynamics state vector to obtain an adaptive time window parameter;

[0020] Performing propagation delay compensation calculation processing on each sensor data based on the distance between sensors and the signal propagation path in the sensor topology association matrix to obtain a sensor delay correction coefficient;

[0021] Applying the sensor delay correction coefficient to the raw data of multiple system sensors to perform time sequence alignment and calibration processing to obtain time sequence synchronized sensor data;

[0022] The time-series synchronized sensor data is segmented and processed for data integrity verification according to the adaptive time window parameters to obtain the synchronized multi-sensor data set.

[0023] Optionally, inputting the synchronized multi-sensor data set into a fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector includes:

[0024] Inputting the synchronized multi-sensor data set into the sensor topology convolution layer of the fault propagation path perception network to perform convolution feature extraction based on the topology structure of the vehicle system to obtain a topology perception sensor feature map;

[0025] Performing time-dependent fault propagation path prediction processing on the topology-aware sensor feature graph through the fault propagation modeling layer of the fault propagation path awareness network to obtain a fault propagation path probability matrix;

[0026] The multi-system coupling analysis layer based on the fault propagation path perception network calculates the inter-system coupling strength and performs fault cascade effect modeling on the fault propagation path probability matrix to obtain a multi-system coupling fault feature tensor;

[0027] Inputting the multi-system coupling fault feature tensor into the adaptive feature fusion layer of the fault propagation path perception network to perform dynamic feature weighted fusion processing based on the fault severity to obtain a fault severity weighted feature vector;

[0028] Vehicle physical constraint verification and feature validity screening are performed on the fault severity weighted feature vector to obtain the vehicle system fault feature vector.

[0029] Optionally, the multi-system coupling analysis layer based on the fault propagation path perception network performs inter-system coupling strength calculation and fault cascade effect modeling processing on the fault propagation path probability matrix to obtain a multi-system coupling fault feature tensor, including:

[0030] The fault propagation path probability matrix is ​​quantified by the coupling strength between the engine-transmission, brake-ABS, and steering-suspension systems through the inter-system physical coupling degree calculation module to obtain a system coupling strength coefficient matrix;

[0031] According to the fault cascade propagation time series model, the system coupling strength coefficient matrix is ​​processed to calculate the fault propagation delay and attenuation coefficient between different systems to obtain the fault cascade propagation parameter matrix;

[0032] Based on the multi-system fault interaction impact analysis algorithm, the fault cascade propagation parameter matrix is ​​modeled and processed for mutual impact and amplification effect between faults to obtain a fault interaction impact feature matrix;

[0033] Performing three-dimensional tensor reconstruction and multi-dimensional fault feature encoding processing on the fault interaction influence feature matrix to obtain an initial multi-system coupling fault feature tensor;

[0034] The initial multi-system coupling fault feature tensor is subjected to tensor decomposition and feature dimension optimization processing to obtain the multi-system coupling fault feature tensor.

[0035] Optionally, the vehicle physical model-based constraint optimizer performs abnormality detection processing on the vehicle system fault feature vector to obtain a fault state identification result, including:

[0036] Performing physical rationality verification processing on the vehicle system fault feature vector using an engine thermodynamic model, a brake system dynamics model, and an electrical system Ohm's law constraint model to obtain a physical constraint violation metric parameter;

[0037] Constructing a constraint penalty term for the anomaly detection objective function according to the physical constraint violation metric parameter to obtain a physical constraint anomaly detection loss function;

[0038] Based on the improved particle swarm optimization algorithm, a global optimal solution search process is performed on the physical constraint anomaly detection loss function to obtain an optimal anomaly detection threshold parameter set;

[0039] Inputting the vehicle system fault feature vector into an anomaly detection classifier configured with the optimal anomaly detection threshold parameter set to perform fault type discrimination processing to obtain a preliminary fault state classification result;

[0040] Confidence assessment and uncertainty quantification are performed on the preliminary fault state classification result to obtain the fault state identification result.

[0041] Optionally, performing source tracing and control instruction generation processing on the fault state identification result according to the fault propagation graph model to obtain a system adjustment control signal includes:

[0042] Perform fault root node location processing on the fault state identification result by using a fault propagation path backtracking algorithm to obtain a fault root sensor node identifier;

[0043] According to the fault root sensor node identifier, the fault propagation graph model is used to calculate the fault impact range and trace the propagation link to obtain a fault impact system mapping table;

[0044] Based on the fault severity classification rules, the fault impact system mapping table is evaluated for fault risk level and prioritized to obtain a hierarchical fault handling strategy table;

[0045] Inputting the hierarchical fault handling strategy table into the control strategy library for corresponding control instruction matching and parameter adjustment processing to obtain an initial system adjustment instruction set;

[0046] The initial system adjustment instruction set is subjected to execution feasibility verification and instruction optimization processing to obtain the system adjustment control signal.

[0047] In a second aspect, the present application provides a vehicle fault detection system that fuses sensor data, the vehicle fault detection system that fuses sensor data comprising:

[0048] A modeling module is used to perform correlation modeling on the multi-system sensor network through the automobile system topology mapping algorithm to obtain the sensor topology correlation matrix;

[0049] a synchronization module, configured to perform adaptive time window synchronization processing on the sensor topology correlation matrix according to a vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set;

[0050] A fusion module, configured to input the synchronized multi-sensor data set into a fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector;

[0051] A detection module, configured to perform abnormality detection processing on the vehicle system fault feature vector based on a vehicle physical model constraint optimizer to obtain a fault state identification result;

[0052] The tracing module is used to perform tracing and locating of the fault state identification result and generate control instructions according to the fault propagation graph model to obtain a system adjustment control signal.

[0053] In a third aspect, a vehicle fault detection device for fusing sensor data is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the vehicle fault detection device for fusing sensor data executes the above-mentioned vehicle fault detection method for fusing sensor data.

[0054] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, enables the computer to execute the above-mentioned automobile fault detection method for fusing sensor data.

[0055] The technical solution provided in this application uses an automotive system topology mapping algorithm to perform correlation modeling on a multi-system sensor network to obtain a sensor topology correlation matrix. This overcomes the limitation of existing technologies that treat sensors as independent information sources, fully explores the physical connection relationships and functional dependencies between various automotive subsystems, and establishes a complete sensor space correlation model. This provides a system-level topological foundation for multi-sensor data fusion, significantly improving the utilization efficiency of inter-sensor correlation information and the accuracy of fault detection. Based on the vehicle dynamics constraint equations, the sensor topology correlation matrix is ​​adaptively time-windowed to obtain a synchronized multi-sensor data set. This overcomes the shortcomings of the existing fixed timestamp alignment strategy. By dynamically adjusting the time window length and compensating for propagation delays, it effectively eliminates the impact of sensor response delay differences under different driving conditions on data fusion quality, ensuring high consistency of multi-sensor data in the time dimension and laying a reliable data foundation for subsequent feature extraction and fault detection. The method uses synchronized multi-sensor datasets to feed a fault propagation path perception network, performing multi-level feature fusion processing to generate vehicle system fault feature vectors. This method overcomes the single feature extraction limitations of traditional methods. Through the synergistic effects of the sensor topology convolution layer, fault propagation modeling layer, multi-system coupling analysis layer, and adaptive feature fusion layer, it deeply explores the propagation patterns and coupling mechanisms of faults across multiple vehicle systems, forming a fault feature representation with rich semantic information, significantly improving the feature recognition and classification accuracy of complex coupled faults. Anomaly detection processing of vehicle system fault feature vectors based on a vehicle physical model constraint optimizer generates fault state identification results, effectively avoiding the physically unreasonable detection results that can occur with purely data-driven methods in existing technologies. Joint verification using the engine thermodynamic model, the brake system dynamics model, and the electrical system Ohm's law constraint model ensures that fault detection results conform to automotive physics and engineering common sense, significantly improving the reliability and credibility of fault diagnosis. According to the fault propagation graph model, the fault state identification results are traced and located, and control instructions are generated to obtain system adjustment control signals, which makes up for the shortcomings of existing technologies in lacking fault root location and active control capabilities. The fault root cause is accurately located through the fault propagation path backtracing algorithm, and targeted control instructions are generated based on fault severity classification and control strategy library matching, realizing closed-loop processing from fault detection to active control, providing a complete solution for automobile fault management, and greatly improving the safety and reliability of automobile systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0057] Figure 1 A schematic diagram of an embodiment of a method for detecting automobile faults by fusing sensor data in an embodiment of the present application;

[0058] Figure 2 This is a schematic diagram of an embodiment of a vehicle fault detection system that fuses sensor data in an embodiment of the present application;

[0059] Figure 3 4 is a schematic block diagram of the structure of an automobile fault detection device that fuses sensor data in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The embodiments of the present application provide a method and system for automobile fault detection that fuses sensor data. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or that are inherent to these processes, methods, products or devices.

[0061] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 An embodiment of the automobile fault detection method of fusing sensor data in the embodiment of the present application includes:

[0062] Step S101: performing correlation modeling on a multi-system sensor network using an automobile system topology mapping algorithm to obtain a sensor topology correlation matrix;

[0063] Step S102: performing adaptive time window synchronization processing on the sensor topology correlation matrix according to the vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set;

[0064] Step S103: Input the synchronized multi-sensor data set into the fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector;

[0065] Step S104: performing abnormality detection processing on the vehicle system fault feature vector based on the vehicle physical model constraint optimizer to obtain a fault state identification result;

[0066] Step S105 : performing source tracing and control instruction generation processing on the fault state identification result according to the fault propagation graph model to obtain a system adjustment control signal.

[0067] It is understandable that the execution subject of this application can be a vehicle fault detection system that integrates sensor data, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0068] Specifically, the automotive system topology mapping algorithm first identifies the physical connections between sensor nodes in the engine, brake, steering, and transmission systems through a CAN bus network topology scan, constructing a topological map of inter-system connections. This topological map reflects the functional dependencies between subsystems, and by assigning weights, a system-wide weighted topological map is formed. Based on the sensor's three-dimensional spatial coordinates and signal propagation paths, the sensor nodes are mapped into the topological map to form a sensor spatial distribution topological map. The sensor physical correlation strength and signal propagation delay parameters are then numerically quantified, and a sensor topological correlation matrix is ​​derived through matrix normalization and symmetry adjustment. The elements of this matrix represent the degree of correlation between sensors, resolving the problem of traditional methods that ignores the spatial relationships between sensors.

[0069] The sensor topology correlation matrix is ​​synchronized with an adaptive time window using vehicle dynamics constraint equations. The operating state parameters collected in real time by the speed, acceleration, and angular velocity sensors form the vehicle dynamics state vector, which dynamically adjusts the length of a preset basic time window to form adaptive time window parameters. Based on the inter-sensor distances and signal propagation paths in the sensor topology correlation matrix, a propagation delay compensation coefficient is calculated for each sensor data. This correction coefficient is applied to the raw data from multiple sensor systems for timing alignment and calibration. The data is then segmented according to the adaptive time window parameters and its integrity verified, ultimately resulting in a synchronized multi-sensor dataset. This dynamic synchronization mechanism effectively addresses the issue of varying sensor response delays under different operating conditions.

[0070] The fault propagation path perception network uses a multi-level architecture for feature fusion processing. The sensor topology convolution layer extracts convolutional features based on the vehicle system topology, generating a topology-aware sensor feature map. The fault propagation modeling layer predicts the time-dependent fault propagation paths based on this feature map and generates a fault propagation path probability matrix. The multi-system coupling analysis layer quantifies the coupling strength between the engine-transmission, brake-ABS, and steering-suspension systems through an inter-system physical coupling calculation module. It calculates the fault propagation delay and attenuation coefficient using a fault cascade propagation time series model. The multi-system fault interaction impact analysis algorithm models the mutual influence and amplification effects between faults. A multi-system coupling fault feature tensor is generated through three-dimensional tensor reconstruction and feature encoding. The adaptive feature fusion layer performs dynamic feature weighted fusion based on fault severity. Vehicle system fault feature vectors are obtained through vehicle physical constraint verification and feature validity screening.

[0071] The vehicle physical model constraint optimizer verifies the physical plausibility of the vehicle system fault feature vector using an engine thermodynamic model, a brake system dynamics model, and an Ohm's law constraint model for the electrical system. It generates physical constraint violation metrics. These metrics construct the constraint penalty term of the anomaly detection objective function, forming a physical constraint anomaly detection loss function. An improved particle swarm optimization algorithm searches for the global optimal solution of this loss function, obtaining the optimal anomaly detection threshold parameter set. An anomaly detection classifier configured with this parameter set identifies the fault type of the vehicle system fault feature vector and, through confidence assessment and uncertainty quantification, determines the fault state identification result. This physical constraint mechanism ensures that the detection results conform to the laws of vehicle physics, avoiding the physically unreasonable results of purely data-driven approaches.

[0072] The fault state identification results are used through a fault propagation path backtracing algorithm to locate the fault root node, obtaining the fault root sensor node identifier. Based on this identifier, the fault propagation graph model is used to calculate the fault impact range and trace the propagation link, forming a fault-affected system mapping table. Fault severity classification rules are used to assess the hazard level and prioritize this mapping table, generating a hierarchical fault handling strategy table. The control strategy library matches corresponding control instructions and adjusts parameters based on this strategy table. After feasibility verification and instruction optimization, the system adjustment control signal is generated. This tracing mechanism overcomes the limitation of existing technologies that can only identify a single fault type. Through fault propagation chain analysis, the root cause of the fault is accurately located and a targeted control strategy is generated.

[0073] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0074] Through CAN bus network topology scanning, the physical connection relationship of the sensor nodes of the engine, brake, steering and transmission systems is identified and processed to obtain the connection topology diagram between the systems;

[0075] According to the functional dependency of each subsystem, the connection topology diagram between systems is weighted to obtain the system association weight topology diagram;

[0076] Based on the sensor's three-dimensional spatial coordinates and signal propagation path, the system's associated weight topology map is mapped to sensor nodes to obtain a sensor spatial distribution topology map.

[0077] The sensor physical correlation strength and signal propagation delay parameters are numerically quantified on the sensor spatial distribution topology map to obtain the sensor correlation metric matrix;

[0078] The sensor correlation metric matrix is ​​normalized and symmetry adjusted to obtain the sensor topological correlation matrix.

[0079] Specifically, CAN bus network topology scanning reads the device identifiers and network addresses of the vehicle's electronic control unit (ECU) to identify the distribution of sensor nodes in various systems, such as the engine control module, anti-lock braking system (ABS), electric power steering (EPS), and transmission control module. The scanning process traverses all node addresses on the CAN bus, obtaining information such as each sensor node's device ID, system type, and communication protocol version. It also records the data frame transmission paths and communication dependencies between nodes, constructing an inter-system connectivity topology consisting of nodes and edges. Nodes represent subsystems, and edges represent communication links between systems.

[0080] The functional dependency analysis of each subsystem is based on vehicle dynamics and control theory, determining the degree of coupling between different systems. The engine system directly influences the transmission's operating state through torque output. The braking system and ABS (anti-lock braking system) coordinate to control wheel slip. The steering system and suspension system jointly influence vehicle handling stability. Weighting is assigned based on the strength of the functional coupling between systems. The higher functional dependency between the engine and transmission is assigned a weight of 0.9, the collaborative control relationship between the braking system and ABS is assigned a weight of 0.8, and the steering and suspension systems' control coupling is assigned a weight of 0.7. The resulting system association weight topology contains weighted edges, each reflecting the closeness of the functional dependency between systems.

[0081] Sensor node mapping is based on the sensor's physical installation location within the vehicle's three-dimensional coordinate system, mapping specific sensor nodes to the system-association weighted topology. The water temperature sensor, intake pressure sensor, and oxygen sensor in the engine compartment are mapped to the engine system node. The wheel speed sensor and brake pressure sensor near the wheel are mapped to the brake system node. The steering wheel assembly's angle sensor and torque sensor are mapped to the steering system node. Signal propagation path analysis considers the physical path length and transmission delay as sensor signals travel through hardware pathways such as wiring harnesses, connectors, and ECUs to the data collection point. The sensor spatial distribution topology map adds spatial location information and signal propagation path data to the system-association weighted topology map.

[0082] Numerical quantification converts the sensor physical correlation strength and signal propagation delay parameters into computable numerical matrix elements. The sensor physical correlation strength is calculated based on factors such as the physical distance between sensors, the number of shared hardware connections, and the degree of signal coupling. The closer the distance, the more shared connections, and the stronger the signal coupling, the greater the correlation strength. The signal propagation delay parameter is calculated based on the signal propagation speed and distance in different media. The propagation speed of electrical signals in copper wire is approximately two-thirds the speed of light, while the propagation speed of optical signals in optical fiber is close to the speed of light. The physical correlation strength between each pair of sensors is used as the element value at the intersection of the matrix rows and columns, and the signal propagation delay is used as the time correction parameter to form the sensor correlation metric matrix.

[0083] Matrix normalization scales the element values ​​in the sensor correlation metric matrix to a standard range between 0 and 1, performing normalization by dividing by the maximum element value in the matrix. Symmetry adjustment ensures that the matrix satisfies the mathematical properties of a symmetric matrix, i.e., the value of the element in the i-th row and j-th column of the matrix is ​​equal to the value of the element in the j-th row and i-th column. A symmetric matrix is ​​obtained by calculating the average of the matrix and its transpose. The sensor topology correlation matrix obtained after normalization and symmetry adjustment has a standardized numerical range and good mathematical properties. The matrix element values ​​represent the degree of correlation between corresponding sensor pairs, with larger values ​​indicating stronger correlation and smaller values ​​indicating weaker correlation.

[0084] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0085] The vehicle speed, acceleration and angular velocity sensors are used to collect and process the current vehicle operating state parameters in real time to obtain the vehicle dynamics state vector;

[0086] Dynamically adjust and calculate the preset basic time window length according to the vehicle dynamics state vector to obtain the adaptive time window parameter;

[0087] Based on the distance between sensors and the signal propagation path in the sensor topology correlation matrix, the propagation delay compensation calculation of each sensor data is performed to obtain the sensor delay correction coefficient;

[0088] Apply the sensor delay correction coefficient to the raw data of multi-system sensors to perform timing alignment and calibration to obtain timing synchronized sensor data;

[0089] The time-series synchronized sensor data is segmented and processed according to the adaptive time window parameters, and the data integrity is verified to obtain a synchronized multi-sensor dataset.

[0090] Specifically, the vehicle speed sensor converts the speed signal from the transmission output shaft or wheels to obtain vehicle speed data. The acceleration sensor detects changes in the vehicle's acceleration along the three axes based on the principle of inertial measurement. The angular velocity sensor uses a gyroscope to measure the vehicle's angular velocity around the vertical axis. Real-time acquisition and processing converts the analog signals from these three sensors into digital signals using an analog-to-digital converter. The sensor values ​​are continuously read at a fixed sampling frequency to form a vehicle dynamics state vector containing parameters such as vehicle speed, longitudinal acceleration, lateral acceleration, vertical acceleration, and yaw rate.

[0091] The parameters in the vehicle dynamics state vector directly reflect the characteristics of the current vehicle operating conditions. Sensor response characteristics and data synchronization requirements vary under different operating conditions. Dynamic adjustment calculation processing determines the current vehicle operating condition based on the combined state of vehicle speed and acceleration. When driving at a low and constant speed, sensor data changes slowly, requiring a longer time window to capture complete information. During sudden acceleration or sharp turns, sensor data changes drastically, requiring a shorter time window to maintain timeliness. A preset basic time window length is used as a standard reference value. The window length is dynamically adjusted through a weighted calculation of the velocity change rate and acceleration amplitude in the vehicle dynamics state vector. The more drastic the speed change and the greater the acceleration, the shorter the time window. Ultimately, adaptive time window parameters that adapt to the current operating conditions are obtained.

[0092] The sensor topology association matrix contains information about the physical distance between each sensor and the length of the signal propagation path. Propagation delay compensation is calculated based on the propagation speed characteristics of the signal in different transmission media. The propagation speed of electrical signals in copper wires is approximately 200 million meters per second, while the propagation speed of optical signals in optical fibers is close to 300 million meters per second. The propagation of digital signals in the CAN bus also requires consideration of protocol processing delays. The propagation delay compensation calculation is based on the physical distance between each pair of sensors divided by the signal propagation speed of the corresponding transmission medium to obtain the theoretical propagation delay. Hardware processing delay and protocol conversion delay are then added to obtain the total propagation delay. The difference in propagation delay from different sensors to the data acquisition center constitutes the sensor delay correction factor, which represents the time offset of each sensor relative to the reference sensor.

[0093] Timing alignment applies sensor delay correction coefficients as time offsets to the raw data from multiple sensor systems. The data's time base is adjusted by adding or subtracting the corresponding correction coefficients from the timestamps of each sensor data. The calibration process selects the sensor with the smallest propagation delay as the time base. The remaining sensor data is then shifted forward or backward based on its delay correction coefficient, ensuring that all sensor data corresponds to the same physical moment. This alignment ensures consistency in the temporal dimension of sensor data, eliminating time deviations caused by differences in signal propagation paths and resulting in time-synchronized sensor data.

[0094] The segmentation process divides continuous time-series synchronized sensor data into fixed time segments based on adaptive time window parameters. Each data segment contains synchronized data from all sensors within that time window. The data integrity verification process checks each data segment for missing sensor data, outliers, or communication errors. Data quality is determined by verifying data continuity, range rationality, and checksum correctness. Integrity verification includes checking multiple dimensions, such as whether the number of data sampling points meets expectations, whether the values ​​are within the sensor's measurement range, and whether changes between adjacent sampling points exceed physical limits. Data segments that pass verification constitute a synchronized multi-sensor dataset that ensures time synchronization and data integrity.

[0095] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0096] The synchronized multi-sensor dataset is input into the sensor topology convolution layer of the fault propagation path perception network to perform convolution feature extraction based on the vehicle system topology structure, and a topology-aware sensor feature map is obtained.

[0097] The fault propagation modeling layer of the fault propagation path perception network performs time-dependent fault propagation path prediction processing on the topology perception sensor feature graph to obtain the fault propagation path probability matrix;

[0098] The multi-system coupling analysis layer based on the fault propagation path perception network calculates the inter-system coupling strength and models the fault cascade effect of the fault propagation path probability matrix to obtain the multi-system coupling fault feature tensor;

[0099] The multi-system coupling fault feature tensor is input into the adaptive feature fusion layer of the fault propagation path perception network for dynamic feature weighted fusion processing based on the fault severity to obtain the fault severity weighted feature vector;

[0100] The fault severity weighted feature vector is processed by vehicle physical constraint verification and feature validity screening to obtain the vehicle system fault feature vector.

[0101] Specifically, the sensor topology convolution layer uses convolution kernels based on the vehicle's system topology to extract features from synchronized multi-sensor datasets. These kernels are designed to respect the physical connectivity and functional dependencies between vehicle subsystems. Unlike traditional two-dimensional convolutions in image processing, convolution is performed on a graph based on the adjacency relationships defined by the sensor topology association matrix. The eigenvalue of each sensor node is derived by aggregating the weighted eigenvalues ​​of its topological neighbors, with the weight coefficients derived from the corresponding association strength values ​​in the sensor topology association matrix. Convolutional feature extraction combines the original data value of each sensor with the data values ​​of its topological neighbors. The weights reflect the degree of physical correlation between sensors. The aggregated eigenvalues ​​incorporate both the local information of the sensor itself and the associated information of its neighbors. The topology-aware sensor feature graph maintains the same topology as the original sensor network, but the eigenvalues ​​of each node incorporate the associated information of surrounding sensors, forming a topology-aware feature representation. The fault propagation modeling layer uses a recurrent neural network architecture to perform temporal dependency analysis on the topology-aware sensor feature graph, exploring the propagation patterns and path characteristics of faults over time. The time-dependent fault propagation path prediction process feeds the topology-aware sensor feature graph into a recurrent neural network in a time series. The network memorizes the feature information of historical moments through hidden states and learns the temporal pattern of fault propagation from the initial faulty sensor node to other related sensor nodes. Based on the feature graph states at the current and historical moments, the prediction process calculates the probability of each sensor node failing in the future and the transition probability of a fault propagating from one sensor node to another. The rows of the fault propagation path probability matrix represent fault source sensor nodes, and the columns represent fault target sensor nodes. The matrix element values ​​represent the probability of a fault propagating from a source node to a target node, with larger probability values ​​indicating a higher probability of propagation.

[0102] The multi-system coupling analysis layer uses the inter-system physical coupling calculation module to conduct an in-depth analysis of the fault propagation path probability matrix, quantifying the coupling strength between different systems, such as the engine-transmission, brake-ABS, and steering-suspension systems. The inter-system coupling strength calculation is based on a statistical analysis of the fault propagation probability values ​​between sensor nodes belonging to different systems. The average fault propagation probability between sensor nodes within the same system is calculated as the intra-system coupling strength, and the average fault propagation probability between sensor nodes in different systems is calculated as the inter-system coupling strength. Fault cascading effect modeling analyzes the chain propagation mechanism of faults across multiple systems and identifies possible cascading fault paths by constructing an inter-system fault propagation link graph. Cascading effect modeling considers the time delay, attenuation coefficient, and amplification factor of fault propagation, establishing a complete propagation link model from the initial fault system to the final affected system. The multi-system coupling fault feature tensor combines the fault propagation path probability matrix, the system coupling strength matrix, and the cascading effect parameters into a three-dimensional tensor structure. The first dimension represents the sensor node index, the second dimension represents the time step, and the third dimension represents the fault feature type.

[0103] The adaptive feature fusion layer dynamically weights the multi-system coupled fault feature tensor based on fault severity. Severity assessment is graded based on the fault's impact on vehicle safety and performance. Braking system faults are considered safety-critical faults and receive the highest weight, engine performance degradation is considered performance-impacting faults and receives a medium weight, and air conditioning system abnormalities are considered comfort-impact faults and receive a lower weight. Dynamic feature weighting fusion combines the different feature dimensions in the multi-system coupled fault feature tensor according to their corresponding severity weights, highlighting the importance of safety-critical fault features and suppressing the interference of minor fault features. The fusion process utilizes an attention mechanism to dynamically calculate the weight coefficient for each feature dimension. Weight calculation is based on a comprehensive assessment of the eigenvalue's magnitude, changing trends, and historical fault patterns. The fault severity weighted feature vector reduces the three-dimensional tensor to a one-dimensional vector, where each element corresponds to a weighted fault feature value for a sensor node.

[0104] Vehicle physical constraint verification processes validate the fault severity weighted feature vector based on automotive engineering principles, ensuring that fault detection results conform to vehicle physics and engineering common sense. Physical constraint verification includes checking multiple dimensions, such as the matching relationship between engine speed and vehicle speed, the correspondence between brake pressure and deceleration, and the consistency between steering angle and vehicle trajectory. The verification process compares the fault feature values ​​in the feature vector with pre-set physical constraints, identifying and correcting or eliminating abnormal feature values ​​that violate physical laws. Feature validity screening uses statistical analysis and information gain calculations to evaluate the contribution of each feature dimension to fault detection, eliminating redundant and noisy features while retaining valid features with significant discriminatory power for fault identification. The screening process utilizes a recursive feature elimination algorithm to gradually eliminate feature dimensions with low contributions until the remaining feature set reaches a pre-set feature count or contribution threshold. The vehicle system fault feature vector contains high-quality fault features that have undergone physical constraint verification and validity screening. Each feature dimension has clear physical meaning and strong fault identification capabilities.

[0105] In a specific embodiment, the process of performing inter-system coupling strength calculation and fault cascading effect modeling on the fault propagation path probability matrix based on the multi-system coupling analysis layer of the fault propagation path awareness network may specifically include the following steps:

[0106] The fault propagation path probability matrix is ​​quantified by the coupling strength between the engine-transmission, brake-ABS, and steering-suspension systems through the inter-system physical coupling degree calculation module to obtain the system coupling strength coefficient matrix;

[0107] According to the fault cascade propagation time series model, the system coupling strength coefficient matrix is ​​processed to calculate the fault propagation delay and attenuation coefficient between different systems, and the fault cascade propagation parameter matrix is ​​obtained;

[0108] Based on the multi-system fault interaction impact analysis algorithm, the fault cascade propagation parameter matrix is ​​modeled to process the mutual impact and amplification effect between faults, and the fault interaction impact characteristic matrix is ​​obtained;

[0109] The fault interaction effect feature matrix is ​​reconstructed into a three-dimensional tensor and multi-dimensional fault feature encoding is performed to obtain the initial multi-system coupling fault feature tensor;

[0110] The initial multi-system coupling fault feature tensor is subjected to tensor decomposition and feature dimension optimization to obtain the multi-system coupling fault feature tensor.

[0111] Specifically, the inter-system physical coupling calculation module quantifies the physical coupling strength between systems by analyzing the fault propagation probabilities between sensor nodes in different systems within the fault propagation path probability matrix. The engine-transmission system coupling strength is calculated based on a statistical analysis of the fault propagation probability from engine sensor nodes to transmission sensor nodes. All sensor nodes within the engine system are considered source nodes, and all sensor nodes within the transmission system are considered target nodes. The engine-transmission coupling strength coefficient is calculated by averaging the corresponding propagation probabilities. The brake-ABS system coupling strength is calculated by calculating the fault propagation probability from brake system sensors, such as the brake pressure sensor and brake pedal position sensor, to ABS system sensors, such as wheel speed sensors and pressure regulators. The steering-suspension system coupling strength is determined by analyzing the fault propagation probability from steering angle sensors and steering torque sensors to suspension position sensors and shock absorber sensors. The system coupling strength coefficient matrix is ​​organized with rows representing source systems and columns representing target systems. Matrix element values ​​reflect the degree of physical coupling between the corresponding systems. Larger values ​​indicate tighter coupling and a higher probability of fault propagation.

[0112] The fault cascade propagation time series model, based on system dynamics theory, establishes a mathematical model of the time delay and intensity attenuation of fault propagation between different systems. This model considers the system response characteristics and the temporal characteristics of the physical propagation mechanism. The propagation delay calculation process analyzes the time interval between the detection of an anomaly by a sensor in the source system and the corresponding change displayed by the sensor in the target system. This delay is related to factors such as the physical distance between the systems, the signal propagation path, and the control system response speed. The delay time for the propagation of an engine system fault to the transmission system is calculated by the time difference between the engine torque change and the transmission input shaft torque response. The delay time for the propagation of a brake system fault to the ABS system is determined by the response time from brake pedal application to ABS activation. The attenuation coefficient calculation process quantifies the degree of attenuation of the fault signal during inter-system propagation. This attenuation is related to factors such as the buffering mechanism, damping characteristics, and compensation capability of the control algorithm between the systems. The fault cascade propagation parameter matrix organizes the propagation delay and attenuation coefficient into a matrix format based on the corresponding system relationships. The matrix rows and columns correspond to the propagation relationships between the systems, and the element values ​​contain two parameters: the delay time and the attenuation coefficient.

[0113] The multi-system fault interaction impact analysis algorithm builds a mathematical model of the interactions between multi-system faults based on graph theory and complex network theory. This algorithm considers complex interactions such as synergistic effects, competitive effects, and amplification effects between faults. Fault interaction impact analysis constructs a system fault impact graph, with systems as nodes and fault propagation relationships as edges, to analyze how multiple simultaneous faults influence and interact with each other. Synergistic effect modeling analyzes the cumulative impact of simultaneous faults in multiple systems. For example, when an engine overheating fault and a cooling system fault occur simultaneously, they reinforce each other, resulting in a fault impact that exceeds the simple summation of individual faults. Competitive effect modeling analyzes the competition between different system faults for shared resources. For example, when an electrical system fault and an engine control fault occur simultaneously, they compete for limited power resources and controller processing capacity. Amplification effect modeling analyzes how certain system faults amplify other system faults. For example, a brake system fault increases engine braking load, amplifying the engine system's operating pressure and failure risk. The fault interaction impact characteristic matrix organizes the intensity parameters of various interaction effects according to system pairings. The matrix element values ​​quantify the impact of the corresponding fault interaction between the corresponding systems.

[0114] The three-dimensional tensor reconstruction process expands the fault interaction feature matrix into a three-dimensional data structure consisting of time, system, and fault type dimensions. The first dimension of the tensor corresponds to different vehicle subsystems, the second dimension corresponds to the time series, and the third dimension corresponds to the fault feature type. The multi-dimensional fault feature encoding process encodes different types of fault feature information into numerical elements within a tensor, including characteristic parameters such as fault probability, fault severity, fault duration, and fault impact range. The encoding process utilizes a combination of one-hot encoding and numerical normalization, converting discrete fault type labels into numerical vectors and normalizing continuous fault parameter values ​​to a standard numerical range. The tensor structure maintains the multidimensional correlation between fault features. The position index of each tensor element corresponds to a specific system-time-feature combination, and the element value reflects the strength of the fault feature for that combination. The initial multi-system coupled fault feature tensor contains complete multi-system fault interaction information, but is high in dimensionality and contains redundant information.

[0115] Tensor decomposition uses singular value decomposition or non-negative matrix factorization algorithms to decompose the high-dimensional initial tensor into a combination of multiple low-dimensional tensors. The decomposition process retains key fault feature information while reducing data dimensionality and computational complexity. The decomposition algorithm uses iterative optimization to find the optimal low-dimensional tensor representation, minimizing the error between the reconstructed tensor and the original tensor. Feature dimension optimization determines the number of feature dimensions to retain based on feature importance assessment and information entropy calculation, eliminating redundant feature dimensions with low contribution to fault detection. The optimization process uses principal component analysis and feature selection algorithms to identify the most representative fault feature dimensions, ensuring that the optimized tensor maintains the integrity of the original fault information while maintaining high computational efficiency. After decomposition and optimization, the multi-system coupling fault feature tensor has a compact data structure and optimized feature representation, and each element in the tensor corresponds to important fault feature information.

[0116] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0117] The vehicle system fault feature vector is physically verified for rationality using the engine thermodynamics model, the brake system dynamics model, and the electrical system Ohm's law constraint model to obtain physical constraint violation metrics.

[0118] According to the physical constraint violation metric parameter, the constraint penalty term of the anomaly detection objective function is constructed to obtain the physical constraint anomaly detection loss function;

[0119] Based on the improved particle swarm optimization algorithm, the global optimal solution search process of the physical constraint anomaly detection loss function is carried out to obtain the optimal anomaly detection threshold parameter set;

[0120] The vehicle system fault feature vector is input into the anomaly detection classifier configured with the optimal anomaly detection threshold parameter set to perform fault type discrimination processing and obtain a preliminary fault status classification result;

[0121] The confidence evaluation and uncertainty quantification processing are performed on the preliminary fault state classification results to obtain the fault state identification results.

[0122] Specifically, the engine thermodynamic model establishes the energy conservation and entropy change relationships during the engine's operation based on the first and second laws of thermodynamics. The model incorporates physical constraints such as equations for the relationship between combustion chamber temperature and pressure, the relationship between coolant temperature and engine load, and the correlation between exhaust temperature and combustion efficiency. The brake system dynamics model establishes the relationship between braking force and vehicle deceleration based on Newton's laws of motion. The model incorporates dynamic constraints such as the linear relationship between brake pressure and braking torque, the nonlinear relationship between wheel slip and road adhesion, and the relationship between braking distance and the square of initial speed. The electrical system Ohm's law constraint model establishes the relationship between voltage, current, and resistance based on fundamental circuit laws. The model incorporates electrical constraints such as the relationship between battery voltage and load current, the proportional relationship between line resistance and voltage drop, and the relationship between power consumption and the square of current. The physical rationality verification process compares the eigenvalues ​​in the vehicle system fault feature vector with the constraints of the corresponding physical model and calculates the extent to which the eigenvalues ​​deviate from the physical constraints. The verification process checks whether the matching relationship between engine speed and torque conforms to the engine characteristic curve, whether the ratio of brake pressure to deceleration is within a reasonable range, and whether the product of voltage and current equals the actual power consumption. The physical constraint violation metric parameter quantifies the severity of each feature dimension violating the physical constraint. The greater the violation, the more inconsistent the feature value is with the physical law, and more attention needs to be paid in fault detection.

[0123] The constraint penalty term is constructed based on the Lagrange multiplier method and penalty function theory, integrating the physical constraint violation metric into the anomaly detection objective function. The penalty term imposes an additional penalty on detection results that violate the physical constraints. The penalty term is implemented as a squared penalty function, with the penalty intensity proportional to the square of the physical constraint violation metric; the greater the violation, the more severe the penalty. The anomaly detection objective function originally only considers data-driven classification error. The inclusion of the constraint penalty term requires a balance between classification accuracy and physical plausibility. The physical constraint anomaly detection loss function combines the original detection loss with the constraint penalty term in a weighted summation. The weight coefficient controls the importance of the physical constraint in the overall optimization objective. The gradient calculation of the loss function considers both the classification error gradient and the constraint violation gradient, ensuring that the optimization process improves detection accuracy while meeting physical constraint requirements.

[0124] The improved particle swarm optimization algorithm integrates adaptive inertia weighting and a diversity preservation mechanism based on the standard particle swarm optimization algorithm, enhancing global search capabilities and convergence stability. During the algorithm initialization phase, a swarm of particles is randomly generated within the parameter space, with each particle representing a candidate solution for a set of anomaly detection threshold parameters. The particle velocity update formula incorporates physical constraint gradient information, allowing the particles to consider both their historical optimal positions and physical constraints during the search process. The adaptive inertia weight is dynamically adjusted based on the current number of iterations and fitness value. A larger inertia weight is used in the early stages of the search to maintain global exploration capability, while a smaller inertia weight is used in the later stages to enhance local convergence accuracy. The diversity preservation mechanism monitors the dispersion of the particle swarm and, when particles become overly aggregated, introduces random perturbations to redistribute the particle distribution, preventing the algorithm from falling into local optima. The global optimal solution search process uses iterative optimization to find the parameter combination that minimizes the physical constraint anomaly detection loss function. The optimal anomaly detection threshold parameter set includes discrimination thresholds and confidence thresholds for various fault types.

[0125] Anomaly detection classifiers utilize machine learning algorithms such as support vector machines or random forests. When configured with an optimal set of anomaly detection threshold parameters, they are able to distinguish between normal states and various fault states. Fault type discrimination involves inputting a vehicle system fault feature vector into the classifier, which then calculates the classification probability based on the similarity between the feature vector and samples of each fault type. The discrimination process calculates the distance to the decision boundary for each fault type based on the feature vector's position in the high-dimensional feature space. The fault type with the closest distance is used as the preliminary classification result. The classifier output contains a classification probability value for each fault type, reflecting the likelihood that the input feature vector belongs to the corresponding fault type. The optimal threshold parameters determine the critical conditions for the classification decision. When the classification probability of a fault type exceeds the corresponding threshold, it is classified as that type of fault; otherwise, it is classified as normal or another fault type. The preliminary fault state classification result includes the fault type label and the corresponding classification probability value, laying the foundation for subsequent confidence assessment.

[0126] Confidence assessment calculates the confidence level of the classification result based on the probability distribution of the classifier output and the distance to the decision boundary. A high confidence level indicates a relatively reliable classification result, while a low confidence level indicates the existence of classification uncertainty. Assessment methods include calculating the difference between the maximum probability and the second-highest probability, analyzing the entropy of the probability distribution, and measuring the distance from the feature vector to the decision boundary, among other metrics. Uncertainty quantification uses Bayesian reasoning or Monte Carlo methods to estimate the uncertainty level of the classification result and quantify the classification uncertainty caused by factors such as data noise, model limitations, and parameter estimation errors. The quantified results are expressed as confidence intervals or uncertainty scores, providing a reliability reference for fault diagnosis decisions. The fault state identification result integrates the preliminary classification results, confidence assessment, and uncertainty quantification information to form a comprehensive diagnostic conclusion that includes the fault type, probability of occurrence, confidence level, and uncertainty range.

[0127] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0128] The fault state identification result is processed by the fault propagation path backtracking algorithm to locate the fault root node and obtain the fault root sensor node identifier;

[0129] According to the fault root sensor node identifier, the fault propagation graph model is used to calculate the fault impact range and trace the propagation link to obtain a fault impact system mapping table;

[0130] Based on the fault severity classification rules, the fault risk level is evaluated and the priority is sorted on the fault impact system mapping table to obtain a hierarchical fault handling strategy table;

[0131] Input the hierarchical fault handling strategy table into the control strategy library to perform corresponding control instruction matching and parameter adjustment processing to obtain the initial system adjustment instruction set;

[0132] The initial system adjustment instruction set is subjected to execution feasibility verification and instruction optimization processing to obtain the system adjustment control signal.

[0133] Specifically, the fault propagation path backtracking algorithm is based on the reverse search principle in graph theory. Starting from the sensor node corresponding to the fault status identification result, it traverses backward along the directed edges in the fault propagation graph to locate the initial fault location. The backtracking process employs a depth-first search strategy, recursively visiting all possible fault source nodes starting from the current faulty sensor node. The algorithm determines the true origin of the fault by analyzing the temporal and causal relationships of fault propagation. The algorithm maintains a node access status table that records the access status and fault occurrence timestamp of each sensor node. The direction and path of fault propagation are identified by comparing the order of these timestamps. The fault root node localization process uses a comprehensive judgment based on fault propagation probability and temporal logic to select the sensor node with the earliest fault timestamp and the highest probability of outward fault propagation as the root node. The localization process also considers the reliability and historical fault history of the sensor node to eliminate misidentifications caused by sensor failures. The fault root sensor node identifier contains information such as the node's unique number, system type, physical location coordinates, and fault occurrence time, providing accurate fault source location for subsequent impact analysis and control decisions.

[0134] The fault propagation graph model describes the fault propagation relationships between various vehicle systems in the form of a directed graph. Nodes in the graph represent sensors or subsystems, and directed edges indicate the direction and intensity of fault propagation. The fault impact range calculation process uses a graph-based breadth-first search algorithm, starting from the fault's root-cause sensor node and progressively expanding to search all potentially affected downstream nodes. The search process records the impact level and propagation delay of each affected node. The impact level is calculated based on the fault propagation probability and propagation path length. The propagation link tracing process constructs the complete propagation path from the fault's root-cause to each affected node. Each path includes intermediate nodes and corresponding propagation delays. The tracing algorithm uses dynamic programming to find the shortest and most probable fault propagation paths, providing analysis of various propagation scenarios for fault impact assessment. The fault-affected system mapping table organizes the impact analysis results in a tabular format, with rows corresponding to affected systems or sensor nodes and columns corresponding to attributes such as impact level, propagation delay, and propagation path. The mapping table also categorizes impact types, distinguishing between direct, indirect, and potential impacts, providing detailed impact analysis data for developing targeted mitigation strategies.

[0135] The fault severity classification system establishes a multi-tiered fault severity assessment system based on automotive safety standards and engineering experience. The classification criteria consider the fault's impact on vehicle safety, performance, and comfort. Safety-critical faults include faults that directly threaten driving safety, such as brake failure, steering loss, and sudden power loss, and are assigned the highest priority. Performance-impacting faults include faults that affect normal vehicle performance but do not directly threaten safety, such as engine power loss, transmission shifting anomalies, and reduced fuel efficiency, and are assigned a medium priority. Comfort-impacting faults include faults that only affect driving comfort, such as air conditioning malfunctions, audio system malfunctions, and seat adjustment failures, and are assigned a lower priority. The fault severity assessment process uses the impact scope and severity in the fault-affected system mapping table, combined with the classification rules, to determine the severity level of each affected system. Prioritization ranks all affected systems according to severity level and urgency, ensuring that safety-critical faults receive the highest handling priority. The graded fault handling strategy table integrates the severity level assessment and priority ranking results to develop a corresponding handling strategy and resource allocation plan for each affected system.

[0136] The control strategy library stores standard control response plans for various fault scenarios, including mappings between fault types and control instructions, parameter adjustment rules, and execution sequencing. The control instruction matching process searches the control strategy library for corresponding control response plans based on the fault type and severity specified in the hierarchical fault handling strategy table. This matching process utilizes a combination of pattern matching and similarity calculation. When no fully matching strategy is found, the most similar strategy is selected as a reference. The parameter adjustment process customizes the parameters in the standard control plan based on the specific characteristics of the current fault and the vehicle's operating status. The adjustment process considers factors such as the vehicle's current speed, load, and environmental conditions to ensure that the control instructions are adapted to actual operating conditions. Adjustment parameters include key control parameters such as controller gain coefficients, actuator action amplitudes, and response time constants. The initial system adjustment instruction set contains specific control instructions for each affected system. Each instruction includes the target system identifier, control action type, parameter settings, and execution time requirements.

[0137] The execution feasibility verification process checks whether each instruction in the initial set of system control instructions can be executed safely and efficiently under current conditions. This verification includes actuator availability checks, system state compatibility analysis, and safety constraint satisfaction assessment. The availability check confirms that the target actuator is in normal working order and capable of performing the specified action. The state compatibility analysis ensures that there are no conflicts or interference between multiple control instructions. The safety constraint assessment verifies that the execution of the control instructions will not introduce new safety risks or exacerbate existing faults. The instruction optimization process further optimizes instructions that pass feasibility verification, including execution timing optimization, parameter fine-tuning, and conflict resolution. Timing optimization determines the optimal execution sequence and time interval for each control instruction to avoid system overload caused by simultaneous execution of multiple instructions. Parameter fine-tuning fine-tunes instruction parameters based on real-time system status and expected control effects. Conflict resolution coordinates potential conflicts between multiple instructions. The system control signal integrates the optimized control instructions into a standard control signal that can be directly sent to the controllers of each subsystem. The signal format complies with the communication protocols and interface standards of automotive electronic systems.

[0138] The above describes the automobile fault detection method of fusing sensor data in the embodiment of the present application. The following describes the automobile fault detection system of fusing sensor data in the embodiment of the present application. Figure 2 In one embodiment of the present application, a vehicle fault detection system integrating sensor data includes:

[0139] A modeling module is used to perform correlation modeling on the multi-system sensor network through the automobile system topology mapping algorithm to obtain the sensor topology correlation matrix;

[0140] a synchronization module, configured to perform adaptive time window synchronization processing on the sensor topology correlation matrix according to a vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set;

[0141] A fusion module, configured to input the synchronized multi-sensor data set into a fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector;

[0142] A detection module, configured to perform abnormality detection processing on the vehicle system fault feature vector based on a vehicle physical model constraint optimizer to obtain a fault state identification result;

[0143] The tracing module is used to perform tracing and locating of the fault state identification result and generate control instructions according to the fault propagation graph model to obtain a system adjustment control signal.

[0144] above Figure 2 The automobile fault detection system for fusing sensor data in an embodiment of the present invention is described in detail from the perspective of modular functional entities. The automobile fault detection device for fusing sensor data in an embodiment of the present invention is described in detail from the perspective of hardware processing.

[0145] Reference Figure 3 In an embodiment of the present invention, a vehicle fault detection device that fuses sensor data is also provided. The vehicle fault detection device that fuses sensor data can be a server, and its internal structure can be as follows: Figure 3 As shown. The automobile fault detection device that fuses sensor data includes a processor, a memory, a display screen, an input device, a network interface and a database connected via a system bus. Among them, the computer-designed processor is used to provide computing and control capabilities. The memory of the automobile fault detection device that fuses sensor data includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the automobile fault detection device that fuses sensor data is used to store the corresponding data in this embodiment. The network interface of the automobile fault detection device that fuses sensor data is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0146] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the automobile fault detection device for fusing sensor data to which the solution of the present invention is applied.

[0147] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the automobile fault detection method for fusing sensor data.

[0148] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0149] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a vehicle fault detection device (which can be a personal computer, server, or network device, etc.) that integrates sensor data to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0150] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for detecting automobile faults by fusing sensor data, characterized in that: The method comprises: The multi-system sensor network is modeled and processed by the automobile system topology mapping algorithm to obtain the sensor topology correlation matrix; Performing adaptive time window synchronization processing on the sensor topology correlation matrix according to the vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set; Inputting the synchronous multi-sensor data set into the fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector, including: inputting the synchronous multi-sensor data set into the sensor topology convolution layer of the fault propagation path perception network for convolution feature extraction processing based on the automobile system topology structure to obtain a topology perception sensor feature map; performing time-dependent fault propagation path prediction processing on the topology perception sensor feature map through the fault propagation modeling layer of the fault propagation path perception network to obtain a fault propagation path probability matrix; performing inter-system coupling strength calculation and fault cascade effect modeling processing on the fault propagation path probability matrix based on the multi-system coupling analysis layer of the fault propagation path perception network to obtain a multi-system coupling fault feature tensor; inputting the multi-system coupling fault feature tensor into the adaptive feature fusion layer of the fault propagation path perception network for dynamic feature weighted fusion processing based on fault severity to obtain a fault severity weighted feature vector; performing vehicle physical constraint verification and feature validity screening processing on the fault severity weighted feature vector to obtain the vehicle system fault feature vector; Among them, the multi-system coupling analysis layer based on the fault propagation path perception network calculates the inter-system coupling strength and models the fault cascade effect on the fault propagation path probability matrix to obtain a multi-system coupling fault feature tensor, including: quantifying the coupling strength between the engine-transmission, brake-ABS, and steering-suspension systems on the fault propagation path probability matrix through the inter-system physical coupling degree calculation module to obtain a system coupling strength coefficient matrix; calculating the propagation delay and attenuation coefficient of the fault between different systems on the system coupling strength coefficient matrix according to the fault cascade propagation timing model to obtain a fault cascade propagation parameter matrix; modeling the mutual influence and amplification effect between faults on the fault cascade propagation parameter matrix based on the multi-system fault interaction impact analysis algorithm to obtain a fault interaction impact feature matrix; performing three-dimensional tensor reconstruction and multi-dimensional fault feature encoding on the fault interaction impact feature matrix to obtain an initial multi-system coupling fault feature tensor; performing tensor decomposition and feature dimension optimization on the initial multi-system coupling fault feature tensor to obtain the multi-system coupling fault feature tensor; Performing abnormality detection processing on the vehicle system fault feature vector based on the vehicle physical model constraint optimizer to obtain a fault state identification result; The fault state identification result is traced and located and a control instruction is generated according to the fault propagation graph model to obtain a system adjustment control signal.

2. The automobile fault detection method of fusing sensor data according to claim 1, characterized in that: The multi-system sensor network is subjected to correlation modeling processing by the automobile system topology mapping algorithm to obtain a sensor topology correlation matrix, including: Through CAN bus network topology scanning, the physical connection relationship of the sensor nodes of the engine, brake, steering and transmission systems is identified and processed to obtain the connection topology diagram between the systems; Perform weight assignment processing on the inter-system connection topology diagram according to the functional dependency relationship of each subsystem to obtain a system-related weighted topology diagram; Performing sensor node mapping processing on the system association weight topology map based on the sensor three-dimensional spatial coordinates and signal propagation paths to obtain a sensor space distribution topology map; Numerical quantification of the sensor spatial distribution topology map is performed using the sensor physical correlation strength and signal propagation delay parameters to obtain a sensor correlation metric matrix; Matrix normalization and symmetry adjustment are performed on the sensor correlation metric matrix to obtain the sensor topology correlation matrix.

3. The automobile fault detection method of fusing sensor data according to claim 1, characterized in that: The adaptive time window synchronization processing of the sensor topology correlation matrix according to the vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set includes: The vehicle speed, acceleration and angular velocity sensors are used to collect and process the current vehicle operating state parameters in real time to obtain the vehicle dynamics state vector; Dynamically adjusting and calculating a preset basic time window length according to the vehicle dynamics state vector to obtain an adaptive time window parameter; Performing propagation delay compensation calculation processing on each sensor data based on the distance between sensors and the signal propagation path in the sensor topology association matrix to obtain a sensor delay correction coefficient; Applying the sensor delay correction coefficient to the raw data of multiple system sensors to perform time sequence alignment and calibration processing to obtain time sequence synchronized sensor data; The time-series synchronized sensor data is segmented and processed for data integrity verification according to the adaptive time window parameters to obtain the synchronized multi-sensor data set.

4. The automobile fault detection method of fusing sensor data according to claim 1, characterized in that: The vehicle physical model-based constraint optimizer performs abnormality detection processing on the vehicle system fault feature vector to obtain a fault state identification result, including: Performing physical rationality verification processing on the vehicle system fault feature vector using an engine thermodynamic model, a brake system dynamics model, and an electrical system Ohm's law constraint model to obtain a physical constraint violation metric parameter; Constructing a constraint penalty term for the anomaly detection objective function according to the physical constraint violation metric parameter to obtain a physical constraint anomaly detection loss function; Based on the improved particle swarm optimization algorithm, a global optimal solution search process is performed on the physical constraint anomaly detection loss function to obtain an optimal anomaly detection threshold parameter set; Inputting the vehicle system fault feature vector into an anomaly detection classifier configured with the optimal anomaly detection threshold parameter set to perform fault type discrimination processing to obtain a preliminary fault state classification result; Confidence assessment and uncertainty quantification are performed on the preliminary fault state classification result to obtain the fault state identification result.

5. The automobile fault detection method of fusing sensor data according to claim 1, characterized in that: The method of performing source tracing and control instruction generation processing on the fault state identification result according to the fault propagation graph model to obtain a system adjustment control signal includes: Perform fault root node location processing on the fault state identification result by using a fault propagation path backtracking algorithm to obtain a fault root sensor node identifier; According to the fault root sensor node identifier, the fault propagation graph model is used to calculate the fault impact range and trace the propagation link to obtain a fault impact system mapping table; Based on the fault severity classification rules, the fault impact system mapping table is evaluated for fault risk level and prioritized to obtain a hierarchical fault handling strategy table; Inputting the hierarchical fault handling strategy table into the control strategy library for corresponding control instruction matching and parameter adjustment processing to obtain an initial system adjustment instruction set; The initial system adjustment instruction set is subjected to execution feasibility verification and instruction optimization processing to obtain the system adjustment control signal.

6. An automobile fault detection system integrating sensor data, characterized in that: A method for detecting an automobile fault by fusing sensor data according to any one of claims 1 to 5, wherein the automobile fault detection system fusing sensor data comprises: A modeling module is used to perform correlation modeling on the multi-system sensor network through the automobile system topology mapping algorithm to obtain the sensor topology correlation matrix; a synchronization module, configured to perform adaptive time window synchronization processing on the sensor topology correlation matrix according to a vehicle dynamics constraint equation to obtain a synchronized multi-sensor data set; A fusion module, configured to input the synchronized multi-sensor data set into a fault propagation path perception network for multi-level feature fusion processing to obtain a vehicle system fault feature vector; A detection module, configured to perform abnormality detection processing on the vehicle system fault feature vector based on a vehicle physical model constraint optimizer to obtain a fault state identification result; The tracing module is used to perform tracing and locating of the fault state identification result and generate control instructions according to the fault propagation graph model to obtain a system adjustment control signal.

7. An automobile fault detection device integrating sensor data, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the method for detecting automobile faults by fusing sensor data according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the automobile fault detection method for fusing sensor data according to any one of claims 1 to 5.

Images