Fault identification method and device for vehicle-mounted power supply, vehicle and storage medium

Abstract

The invention discloses a fault identification method and device for a vehicle-mounted power supply, a vehicle and a storage medium. The method comprises the steps that an operation parameter set of a vehicle-mounted power supply in a vehicle in the current operation process is obtained, the operation parameter set comprises a plurality of sub-operation parameter sets collected by a plurality of sensors in the vehicle, and each sub-operation parameter set comprises a plurality of operation parameters collected at the current moment; the operation parameters are used for representing the current operation state of the vehicle-mounted power supply; the operation parameter set is preprocessed, a standardized time series data set is obtained, and the standardized time series data set is used for representing the operation state, changing along with time, of the vehicle-mounted power supply; and performing fault diagnosis on the standardized time sequence data set by using the trained fault diagnosis model to obtain a fault diagnosis result. The technical problem of low fault identification accuracy of the vehicle-mounted power supply is solved.

Description

Technical Field

[0001] This application relates to the field of network models, and more specifically, to a method, apparatus, vehicle, and storage medium for fault identification of vehicle power supplies. Background Technology

[0002] Currently, with the continuous development of science and technology, vehicles are becoming increasingly important in social life. The on-board power supply is one of the core components of new energy vehicles, enabling the conversion of AC and DC energy between high and low voltage systems. The reliability of the on-board power supply has a significant impact on the safety performance of new energy vehicles; a failure could lead to economic losses or even accidents. New energy vehicles face complex usage scenarios such as challenging road conditions, power grid anomalies, and harsh natural environments, posing new challenges to on-board power supply fault diagnosis. Traditional diagnostic methods suffer from slow response times, making real-time monitoring and early warning difficult. They also have significant limitations in addressing the diversity and complexity of fault diagnosis, resulting in low accuracy in on-board power supply fault identification.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides a method, apparatus, vehicle, and storage medium for fault identification of vehicle power supplies, in order to at least solve the technical problem of low accuracy in fault identification of vehicle power supplies.

[0005] According to one aspect of the embodiments of this application, a fault identification method for an on-board power supply is provided. The method may include: acquiring an operating parameter set of the on-board power supply in the vehicle during the current operation process, wherein the operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, and the sub-operating parameter sets include multiple operating parameters collected at multiple current times, the operating parameters being used to characterize the current operating state of the on-board power supply; preprocessing the operating parameter set to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating state of the on-board power supply changing over time; and using a trained fault diagnosis model to perform fault diagnosis on the standardized time series dataset to obtain a fault diagnosis result, wherein the fault diagnosis model is trained based on a historical operating parameter set of the on-board power supply during historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set, the historical operating parameter set including multiple sub-historical operating parameter sets collected by multiple sensors, the sub-historical operating parameter sets including multiple historical operating parameters collected at multiple historical times, the historical operating parameters being used to characterize the historical operating state of the on-board power supply.

[0006] Optionally, a trained fault diagnosis model is used to perform fault diagnosis on a standardized time series dataset to obtain fault diagnosis results. This includes: using a graph attention network model in the fault diagnosis model to extract features from the standardized time series dataset to obtain spatiotemporal feature data corresponding to the sensor; using a temporal convolutional network model in the fault diagnosis model to extract features from the spatiotemporal feature data to obtain trend features among multiple operating parameters in the operating parameter set; fusing the spatiotemporal feature data and trend features to obtain fused features; and identifying the fused features to obtain fault diagnosis results.

[0007] Optionally, the method may further include: acquiring a historical operating parameter set; preprocessing the historical operating parameter set to obtain a multidimensional time series dataset, wherein the multidimensional time series dataset is used to characterize the operating status of the vehicle power supply over time during historical operation; constructing a fault knowledge graph of the vehicle power supply based on the multidimensional time series dataset, wherein nodes in the fault knowledge graph are used to characterize sensors, and edges in the fault knowledge graph are used to characterize the degree of correlation between historical operating parameters collected by different sensors; and training a fault diagnosis model using the fault knowledge graph and historical fault diagnosis results.

[0008] Optionally, based on a multidimensional time-series dataset, a fault knowledge graph of the vehicle power supply is constructed, including: converting historical operating parameters in the multidimensional time-series dataset into multiple state vectors; identifying the similarity between multiple state vectors, where the similarity is used to characterize the degree of association between multiple state vectors; using the similarity, determining at least one target operating parameter adjacent to the historical operating parameter from the multidimensional time-series dataset, and the connection weight between the target operating parameter and the historical operating parameter, where the connection weight is positively correlated with the degree of association; using the sensors that collect historical operating parameters and the sensors that collect target operating parameters as nodes, and using the connection weight, constructing edges between nodes to obtain the fault knowledge graph.

[0009] Optionally, a fault diagnosis model is trained using a fault knowledge graph and historical fault diagnosis results. This includes: acquiring neighbor information, same-modality node information, and different-modality node information from the fault knowledge graph. Neighbor information is used to characterize the correlation between historical operating parameters collected by sensors in adjacent nodes; same-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of similar types; and different-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of different types. An initial graph attention network model is constructed using the neighbor information, same-modality node information, and different-modality node information. The initial graph attention network model is then trained using historical fault diagnosis results to obtain the fault diagnosis model.

[0010] Optionally, the initial graph attention network model is trained using historical fault diagnosis results to obtain a fault diagnosis model. This includes: extracting features from the sub-historical operating parameter sets corresponding to nodes in the fault knowledge graph to obtain trend features of the sub-historical operating parameter sets, wherein the trend features are used to characterize the changing trends of historical operating parameters at different historical moments in the sub-historical operating parameter sets; constructing an initial temporal convolutional network model using the trend features; and training the initial graph attention network model using the initial temporal convolutional network model and historical fault diagnosis results to obtain a fault diagnosis model.

[0011] Optionally, the initial graph attention network model is trained using the initial temporal convolutional network model and historical fault diagnosis results to obtain a fault diagnosis model. This includes: using the initial graph attention network model to extract features from the historical operating parameter set to obtain first feature data; using the initial temporal convolutional network model to extract features from the first feature data to obtain second feature data; fusing the first and second feature data to obtain third feature data; identifying the third feature data to obtain an identification result; constructing a loss function based on the identification result and historical fault diagnosis results; using the loss function to update the initial graph attention network model and the initial temporal convolutional network model respectively to obtain a graph attention network model and a temporal convolutional network model; and combining the graph attention network model and the temporal convolutional network model to obtain the fault diagnosis model.

[0012] According to another aspect of the embodiments of this application, a fault identification device for an on-board power supply is also provided. The device may include: an acquisition unit, configured to acquire a set of operating parameters of the on-board power supply in the vehicle during its current operation, wherein the set of operating parameters includes multiple sub-sets of operating parameters collected by multiple sensors in the vehicle, each sub-set including multiple operating parameters collected at a current time, and the operating parameters characterizing the current operating state of the on-board power supply; a processing unit, configured to preprocess the set of operating parameters to obtain a standardized time-series dataset, wherein the standardized time-series dataset characterizes the operating state of the on-board power supply changing over time; and a diagnosis unit, configured to use a trained fault diagnosis model to perform fault diagnosis on the standardized time-series dataset and obtain a fault diagnosis result, wherein the fault diagnosis model is trained based on a historical set of operating parameters of the on-board power supply during historical operation and historical fault diagnosis results corresponding to the historical operating parameter sets, the historical operating parameter sets including multiple sub-historical operating parameter sets collected by multiple sensors, each sub-historical operating parameter set including multiple historical operating parameters collected at multiple historical times, and the historical operating parameters characterizing the historical operating state of the on-board power supply.

[0013] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored program, wherein, when the program is running, it controls the device where the computer-readable storage medium is located to execute the vehicle power supply fault identification method of the embodiments of this application.

[0014] According to another aspect of the embodiments of this application, a processor is also provided for running a program, wherein the program is executed by the processor to perform the vehicle power supply fault identification method of the embodiments of this application.

[0015] According to another aspect of the embodiments of this application, a program product is also provided, the program product including computer instructions, wherein when the computer instructions are executed by a processor, they implement the vehicle power supply fault identification method of the embodiments of this application.

[0016] According to another aspect of the embodiments of this application, a vehicle is also provided, which can be used to perform the vehicle power supply fault identification method of the embodiments of this application.

[0017] In this embodiment, the operating parameter set of the vehicle's on-board power supply during the current operation is obtained. This operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, each sub-operating parameter set including multiple operating parameters collected at the current time. These operating parameters characterize the current operating state of the on-board power supply. The operating parameter set is preprocessed to obtain a standardized time-series dataset, which characterizes the operating state of the on-board power supply over time. A trained fault diagnosis model is used to diagnose faults in the standardized time-series dataset, yielding fault diagnosis results. The fault diagnosis model is trained based on historical operating parameter sets of the on-board power supply during historical operation and corresponding historical fault diagnosis results. The historical operating parameter sets include multiple sub-historical operating parameter sets collected by multiple sensors, each sub-historical operating parameter set including multiple historical operating parameters collected at multiple historical times. These historical operating parameters characterize the historical operating state of the on-board power supply. In other words, in this embodiment of the application, a fault diagnosis model is pre-trained based on historical operating parameters. When it is necessary to diagnose a fault in the vehicle power supply, an operating parameter set can be obtained. The operating parameter set is preprocessed to obtain a standardized time series dataset. This standardized time series dataset can be used to characterize the operating status of the vehicle power supply over time. The fault diagnosis model is used to diagnose the fault in the standardized time series dataset to obtain the fault diagnosis result. This achieves the technical effect of improving the fault identification accuracy of the vehicle power supply and solves the technical problem of low fault identification accuracy of the vehicle power supply. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of a fault identification method for an on-board power supply according to an embodiment of this application; Figure 2 This is a schematic diagram of data acquisition and preprocessing according to an embodiment of this application; Figure 3 This is a flowchart of an automotive on-board power supply fault diagnosis based on a multimodal spatiotemporal graph attention network model according to an embodiment of this application; Figure 4 This is a flowchart illustrating the construction of a fault diagnosis model according to an embodiment of this application; Figure 5 This is a schematic diagram of a fault identification device for an on-board power supply according to an embodiment of this application; Figure 6 This is a structural block diagram of a computer terminal according to an embodiment of this application; Figure 7 This is a block diagram of an electronic device for a fault identification method of an on-board power supply according to an embodiment of this application. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] Currently, with the continuous development of science and technology, vehicles are becoming increasingly important in social life. The on-board power supply is one of the core components of new energy vehicles, enabling the conversion of AC and DC energy between high and low voltage systems. The reliability of the on-board power supply has a significant impact on the safety performance of new energy vehicles; a failure could lead to economic losses or even accidents. New energy vehicles face complex usage scenarios such as challenging road conditions, power grid anomalies, and harsh natural environments, posing new challenges to on-board power supply fault diagnosis. Traditional diagnostic methods suffer from slow response times, making real-time monitoring and early warning difficult, and still have significant limitations in addressing the diversity and complexity of fault diagnosis.

[0022] In this embodiment, a fault diagnosis model is pre-trained based on historical operating parameters. When fault diagnosis of the vehicle power supply is required, an operating parameter set can be obtained. The operating parameter set is preprocessed to obtain a standardized time series dataset. This standardized time series dataset can be used to characterize the operating status of the vehicle power supply over time. The fault diagnosis model is used to diagnose the fault on the standardized time series dataset to obtain the fault diagnosis result. This achieves the technical effect of improving the fault identification accuracy of the vehicle power supply and solves the technical problem of low fault identification accuracy of the vehicle power supply.

[0023] Optionally, data can be nonlinearly mapped using deep learning multi-layer neural networks to automatically extract features and identify complex patterns, adapting to vehicle power supply conditions and achieving more accurate, efficient, and intelligent fault detection, identification, prediction, and maintenance.

[0024] According to an embodiment of this application, an embodiment of a fault identification method for vehicle power supply is provided. The steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0025] Figure 1 This is a flowchart of a fault identification method for an on-board power supply according to an embodiment of this application. Figure 1 As shown, the method may include the following steps:

[0026] Step S102: Obtain the set of operating parameters of the vehicle power supply during the current operation process. The set of operating parameters includes multiple sub-sets of operating parameters collected by multiple sensors in the vehicle. Each sub-set of operating parameters includes multiple operating parameters collected at the current time. The operating parameters are used to characterize the current operating state of the vehicle power supply.

[0027] In the technical solution provided in step S102 of this application, the on-board power supply can be a device related to power conversion and management in the vehicle, including but not limited to: a battery management system, an on-board charger, an inverter, a DC-DC converter, etc., used to convert the DC power from the battery into AC power required for vehicle drive and other electrical systems, or to convert AC power into DC power required for battery charging. The aforementioned set of operating parameters can include sub-sets of operating parameters collected by multiple sensors in the vehicle. These sub-sets of operating parameters include multiple operating parameters collected at the current moment. The operating parameters can include key information such as the input voltage, input current, output current, output voltage, temperature, charging time, and status information of the on-board power supply. It should be noted that this is merely an example and does not impose specific limitations on the type of on-board power supply or the type of operating parameters.

[0028] Optionally, multiple operating parameters collected by the sensor at multiple current times can be obtained to obtain the sub-operating parameter set corresponding to the sensor. Multiple sub-operating parameter sets of multiple sensors can be obtained to obtain the operating parameter set.

[0029] Step S104: Preprocess the set of operating parameters to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating status of the vehicle power supply over time.

[0030] In the technical solution provided in step S104 of this application, the set of operating parameters can be preprocessed to obtain a standardized time series dataset. This standardized time series dataset can be a standardized multidimensional time series dataset, which can be used to characterize the operating status of the vehicle power supply over time.

[0031] Optionally, the processed standardized time series dataset not only retains the information of the original sensor data, but also ensures the comparability of the data over time through preprocessing steps such as normalization, so that each operating parameter can clearly reflect the changes in the operating status of the vehicle power supply at different time points.

[0032] Optionally, the above preprocessing methods may include, but are not limited to: data cleaning, noise reduction, missing value imputation, normalization, etc. It should be noted that this is only an example and no specific restrictions are placed on the preprocessing methods.

[0033] Figure 2 This is a schematic diagram of data acquisition and preprocessing according to an embodiment of this application, such as... Figure 2As shown, the data acquisition module 201 can acquire sub-operational parameter sets collected by different sensors to obtain an operational parameter set. The operational parameter set can be transmitted to the preprocessing module 202, where the denoising unit 2021, missing value imputation unit 2022, and normalization unit 2023 in the preprocessing module are used to perform denoising, missing value imputation, and normalization processing on the operational parameter set to obtain a standardized time series dataset (which can be simply referred to as standardized data).

[0034] Step S106: Using the trained fault diagnosis model, fault diagnosis is performed on the standardized time series dataset to obtain fault diagnosis results. The fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set. The historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors. The sub-historical operating parameter sets include multiple historical operating parameters collected at multiple historical moments. The historical operating parameters are used to characterize the historical operating status of the vehicle power supply.

[0035] In the technical solution provided in step S106 of this application, the fault diagnosis model can be a deep learning model, such as a multimodal spatiotemporal graph attention network model (MST-GAT), which can include some or all features of a graph attention network (GAT) model and a temporal convolutional network (TCN) model. It can be a model trained based on the historical operating parameter set of the vehicle power supply during historical operation and the corresponding historical fault diagnosis results. The historical operating parameter set can include multiple sub-historical operating parameter sets collected by multiple sensors during historical operation. These sub-historical operating parameter sets can include multiple historical operating parameters collected at multiple historical moments. These historical operating parameters can include information such as input voltage, input current, output current, output voltage, temperature, charging time, and status information during historical operation. These historical operating parameters can be used to characterize the historical operating state of the vehicle power supply. It should be noted that this is only an example and does not impose specific limitations on the type of fault diagnosis model or the type of historical operating parameters.

[0036] Optionally, the aforementioned fault diagnosis model combines the advantages of graph attention networks and temporal convolutional networks. Therefore, the fault diagnosis model can not only handle spatial graph relationships but also temporal sequence information. Moreover, it can simultaneously process and fuse data from different modalities (such as video, audio, and text). In MST-GAT, the graph attention mechanism is responsible for fusing spatial node relationships, while the temporal convolutional network is responsible for extracting long-term dependencies in the time series. The two work together, along with the inter-modal attention mechanism, enabling the model to capture fault signs from multiple angles and levels in complex and ever-changing environments, such as in vehicle power supply fault diagnosis scenarios, thereby improving the accuracy and timeliness of diagnosis.

[0037] Optionally, the Graph Attention Network (GAT) described above can be used to process node features in graph data structures. It adaptively determines the information transfer weights between nodes through an attention mechanism to aggregate neighbor node features. GAT can effectively capture the spatial relationships between nodes in the graph and their importance within the graph. The Temporal Convolutional Network (TCN) described above can be used to process time-series data. It utilizes causal convolution and residual connections to capture long-term dependencies, while ensuring that the model's predictions of future information in the sequence are not affected by factors beyond the current time step.

[0038] In this embodiment, the Multimodal Spatiotemporal Graph Attention Network (MST-GAT) model is used to extract features from time series data containing multiple types and perform fault diagnosis to handle the correlation and dependency of different types of data (i.e., multimodal) at different time points (i.e., time series).

[0039] Optionally, the MST-GAT model, through an attention mechanism, can focus on the sensor data and time points most relevant to the current fault. This means that even with large amounts of data and multiple sensors, the model can identify which information is most critical. For example, during a typical charging cycle, if the battery suddenly stops charging, the model will automatically focus on the data from the temperature and input current sensors, as their changes may be directly related to the cause of the fault. In this way, the MST-GAT model can identify faults faster and more accurately than traditional threshold-based or statistical methods, because it does not rely solely on a single indicator exceeding a certain threshold, but rather makes diagnoses through complex data correlation and pattern recognition.

[0040] Optionally, the MST-GAT model uses deep learning to automatically extract and fuse the spatiotemporal features of multimodal sensor data, and uses an attention mechanism to highlight key information, thereby achieving efficient and accurate fault diagnosis of the vehicle power system.

[0041] Optionally, in another aspect, this application also provides a fault diagnosis device for on-board power supply of new energy vehicles. The device may include: a data acquisition unit for acquiring data of the on-board power supply system of new energy vehicles during a target time period; a preprocessing unit for preprocessing the data to obtain standardized time-series data; a model inference unit for inputting the standardized time-series data into a trained multimodal spatiotemporal graph attention network model to obtain fault diagnosis results; and a result output unit for outputting fault diagnosis results and an interpretability analysis report.

[0042] Optionally, the fault diagnosis module implements real-time anomaly detection and fault classification to determine whether a fault has occurred. If a fault is determined to have occurred, a fault diagnosis report is generated.

[0043] To provide a more intuitive understanding of the above-mentioned vehicle power supply fault identification method, the overall process of the method is provided. Figure 3 This is a flowchart illustrating an automotive on-board power supply fault diagnosis method based on a multimodal spatiotemporal graph attention network model, according to an embodiment of this application. Figure 3 As shown, the method may include the following steps:

[0044] Step S302: Obtain data such as input voltage, input current, temperature, and status of the vehicle power supply.

[0045] In this embodiment, relevant data of the new energy vehicle's on-board power system during the target time period are acquired, such as input voltage, input current, temperature, and status data.

[0046] Step S304: Preprocess the acquired data to obtain a standardized time series dataset.

[0047] In this embodiment, the obtained data is preprocessed to obtain a standardized time series dataset, which can be a standardized multidimensional time series dataset.

[0048] Step S306: Input the data into the created fault diagnosis model and output the diagnosis report.

[0049] In this embodiment, a standardized time-series dataset can be input into a trained multimodal spatiotemporal graph attention network model to obtain fault diagnosis results. Furthermore, a diagnostic report can be output based on the fault diagnosis results.

[0050] Through steps S102 to S106 of this application, the operating parameter set of the vehicle power supply during the current operation process is obtained. The operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, each sub-operating parameter set including multiple operating parameters collected at the current time. These operating parameters characterize the current operating state of the vehicle power supply. The operating parameter set is preprocessed to obtain a standardized time-series dataset, which characterizes the operating state of the vehicle power supply over time. A trained fault diagnosis model is used to diagnose faults in the standardized time-series dataset, yielding fault diagnosis results. The fault diagnosis model is trained based on historical operating parameter sets of the vehicle power supply during historical operation and corresponding historical fault diagnosis results. The historical operating parameter sets include multiple sub-historical operating parameter sets collected by multiple sensors, each sub-historical operating parameter set including multiple historical operating parameters collected at multiple historical times. These historical operating parameters characterize the historical operating state of the vehicle power supply. This achieves the technical effect of improving the fault identification accuracy of the vehicle power supply and solves the technical problem of low fault identification accuracy of the vehicle power supply.

[0051] The method described in this embodiment will be further described below.

[0052] As an optional implementation, step S106 involves using a trained fault diagnosis model to perform fault diagnosis on a standardized time series dataset to obtain fault diagnosis results. This includes: using a graph attention network model in the fault diagnosis model to extract features from the standardized time series dataset to obtain spatiotemporal feature data corresponding to the sensor; using a temporal convolutional network model in the fault diagnosis model to extract features from the spatiotemporal feature data to obtain trend features among multiple operating parameters in the operating parameter set; fusing the spatiotemporal feature data and trend features to obtain fused features; and identifying the fused features to obtain fault diagnosis results.

[0053] In this embodiment, the graph attention network model described above can be a graph attention network (GAT) model, which can be a network model pre-constructed based on a historical set of operating parameters. The spatiotemporal feature data described above can be a spatiotemporal feature representation corresponding to each node. The trend features described above can be features of the long-term trend of the sub-operating parameter set, which can be used to characterize the change process of the sub-operating parameter set.

[0054] Optionally, the graph attention network model in the fault diagnosis model is used to extract features from the standardized time series dataset to obtain the spatiotemporal feature data corresponding to the sensor; the temporal convolutional network model (TCN) in the fault diagnosis model is used to extract features from the spatiotemporal feature data to obtain the trend features among multiple operating parameters in the operating parameter set; the spatiotemporal feature data and the trend features are fused to obtain the fused features; and the fused features are identified to obtain the fault diagnosis result.

[0055] Optionally, the data processing of the multimodal graph attention network may include the following steps: First, MST-GAT processes the spatial graph structure data, i.e., the relationships between various sensor nodes, through a graph attention network. This process aggregates the neighbor information, same-modal node information, and different-modal node information of each node, using a multi-head attention mechanism to capture the inter-node associations at different levels and generate a spatiotemporal feature representation for each node. Next, a temporal convolutional network is used to process the node features derived from GAT to capture long-term temporal dependencies. In this step, TCN slides across the time series using one-dimensional causal dilated convolutions, while maintaining the stability of the model through residual connections, extracting features representing long-term trends.

[0056] Furthermore, after processing the spatial and temporal information respectively, MST-GAT needs to fuse the feature representations of these two dimensions together. Fusion methods include, but are not limited to, concatenation, weighted summation, or fusion through a specially designed network layer such as a bilinear pooling layer. In this way, the MST-GAT model can simultaneously consider the local spatial relationships and historical temporal dynamics of each node. The output after feature fusion enters a classification layer, such as a fully connected layer, for fault detection or classification. This step can map the fused features to fault categories using trained weight parameters to obtain fault diagnosis results.

[0057] For example, suppose there is a set of sensor data from an onboard power supply, including temperature, voltage, and current, each constituting a mode in a sensor network graph. The MST-GAT model first processes the sensor data for each mode independently, capturing spatial node relationships (i.e., correlations between different sensors) through GAT. Then, it uses TCN to extract the time-series features of each sensor mode's data (e.g., how temperature changes over time). After extracting the spatial and temporal features, the MST-GAT model can fuse this information to obtain a comprehensive time-space feature representation. This representation can include information from multiple modes of data and the spatiotemporal correlations between the data. Finally, the fused features are passed through a classification layer to generate a fault diagnosis result. This fault diagnosis result can be a probability distribution, used to determine the likelihood of each potential fault category, as well as the location, region, and severity of each potential fault category. It should be noted that this is only an example and does not impose specific limitations on the content included in the fault diagnosis result.

[0058] Optionally, the internal processing logic of the MST-GAT model extracts and processes spatial relationships and time-series features through a graph attention network and a temporal convolutional network, respectively. These features are then fused together, and a classifier makes the final fault diagnosis decision. This method effectively utilizes the spatiotemporal characteristics of the data, improving the accuracy and efficiency of diagnosis.

[0059] Optionally, MST-GAT processes spatial graph structure data, i.e., the relationships between various sensor nodes, using a Graph Attention Network (GAT). This process aggregates neighbor information, same-modality node information, and different-modality node information for each node, using a multi-head attention mechanism to capture the inter-node associations at different levels, generating a spatiotemporal feature representation for each node. Next, a Temporal Convolutional Network (TCN) is used to process the node features derived from GAT to capture long-term temporal dependencies. In this step, the TCN slides across the time series using one-dimensional causal dilated convolutions, while maintaining model stability through residual connections, extracting features representing long-term trends.

[0060] As an optional implementation, the method may further include: acquiring a historical operating parameter set; preprocessing the historical operating parameter set to obtain a multidimensional time series dataset, wherein the multidimensional time series dataset is used to characterize the operating status of the vehicle power supply over time during historical operation; constructing a fault knowledge graph of the vehicle power supply based on the multidimensional time series dataset, wherein nodes in the fault knowledge graph are used to characterize sensors, and edges in the fault knowledge graph are used to characterize the degree of correlation between historical operating parameters collected by different sensors; and training a fault diagnosis model using the fault knowledge graph and historical fault diagnosis results.

[0061] In this embodiment, historical operating parameters are collected, such as key information like input voltage, input current, output current, output voltage, temperature, charging time, and status information of the vehicle power supply during historical operation. Simultaneously, operating status and historical fault data are recorded to obtain a historical operating parameter set. The historical operating parameter set undergoes data cleaning, noise reduction, and missing value imputation, and multiple data points are normalized to form a multidimensional time series dataset. Based on the multidimensional time series dataset, a fault knowledge graph is constructed. A deep learning model for fault diagnosis of the vehicle power supply system is established through a multimodal spatiotemporal graph attention network construction process, resulting in a network structure that integrates multimodal spatiotemporal features and performs joint optimization. The network topology is constructed through adaptive graph structure learning.

[0062] Optionally, the historical operating parameter set is organized into a graph after preprocessing to obtain a fault knowledge graph. In this graph, nodes represent different sensors, and edges represent the correlations between sensors. For example, there may be a strong correlation between input voltage and input current, thus resulting in strong edge weights. A multimodal graph attention network mechanism is used to consider not only information from a single mode (e.g., voltage) but also the relationships between different modes (e.g., voltage and temperature) to synthesize data from various sensors. The model learns the features at each time point while considering the continuity of the time series, identifying patterns of feature changes before and after a fault. By simultaneously optimizing spatial and temporal features, the model ensures that it captures both spatial multi-sensor correlations and temporal dynamic changes.

[0063] Optionally, nodes in a fault knowledge graph can represent sensors or data sources, rather than the specific data values ​​collected by the sensors. Each node can be considered a feature entity, carrying data information collected by the corresponding sensor. When constructing the graph structure, nodes represent sensors, and the edge weights between nodes reflect the correlation or similarity between sensor data. For example, in the scenario of vehicle power supply fault diagnosis, the nodes of the graph may include "input voltage sensor," "input current sensor," and "output voltage sensor," etc. These nodes are connected by edges, and the weights of the edges are determined by the correlation between sensor data. This structure allows the model to consider the mutual influence between sensors when processing data, thereby more effectively extracting multimodal spatiotemporal features for fault diagnosis.

[0064] As an optional implementation, a fault knowledge graph of the vehicle power supply is constructed based on a multidimensional time series dataset, including: converting historical operating parameters in the multidimensional time series dataset into multiple state vectors; identifying the similarity between the multiple state vectors, wherein the similarity is used to characterize the degree of association between the multiple state vectors; using the similarity, determining at least one target operating parameter adjacent to the historical operating parameter from the multidimensional time series dataset, and the connection weight between the target operating parameter and the historical operating parameter, wherein the connection weight is positively correlated with the degree of association; using the sensors that collect historical operating parameters and the sensors that collect target operating parameters as nodes, and using the connection weight, constructing edges between the nodes to obtain the fault knowledge graph.

[0065] In this embodiment, a fault knowledge graph of the vehicle power supply needs to be constructed based on a multi-dimensional time series dataset. The construction of this graph structure may include: initializing a learnable embedding vector for each node, and selecting multiple similar nodes (e.g., the Top K similar nodes) as neighbors by calculating the cosine similarity between nodes to obtain a matrix, which can serve as the fault knowledge graph.

[0066] Optionally, a node can be created for each sensor. Next, the connection methods between nodes need to be determined. The connection weights between nodes can be determined based on correlation analysis of sensor data, the correlation between nodes can be calculated using methods such as cosine similarity, and the Top K similar nodes can be selected as adjacent nodes.

[0067] Optionally, each sensor is identified as a node in the graph. Edge weights are determined based on correlation analysis between sensor data (i.e., historical operating parameters collected by the sensors). The correlation between sensor data can be calculated using methods such as cosine similarity and Pearson correlation coefficient, and then the connection strength between each pair of sensor nodes is determined based on the calculation results.

[0068] Optionally, historical operating parameters in the multidimensional time series dataset are converted into multiple state vectors; the similarity between the multiple state vectors is identified, where the similarity characterizes the degree of association between the multiple state vectors; using the similarity, at least one target operating parameter adjacent to the historical operating parameter is determined from the multidimensional time series dataset, based on the fact that the sensor that collects the target operating parameter can be considered as an adjacent node of the sensor that collects the historical operating parameter. Connection weights can also be added between the target operating parameter and the historical operating parameter, where the connection weights are positively correlated with the degree of association; using the sensors that collect the historical operating parameter and the sensors that collect the target operating parameter as nodes, and using the connection weights, edges are constructed between the nodes to obtain a fault knowledge graph.

[0069] As an optional implementation, a fault diagnosis model is trained using a fault knowledge graph and historical fault diagnosis results. This includes: acquiring neighbor information, same-modality node information, and different-modality node information from the fault knowledge graph. Neighbor information is used to characterize the correlation between historical operating parameters collected by sensors in adjacent nodes; same-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of similar types; and different-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of different types. An initial graph attention network model is constructed using the neighbor information, same-modality node information, and different-modality node information. The initial graph attention network model is then trained using historical fault diagnosis results to obtain the fault diagnosis model.

[0070] In this embodiment, the fault diagnosis model may include a graph attention network, which can be constructed using a fault knowledge graph. The neighbor information, same-modality node information, and different-modality node information of the aggregated nodes can be used to capture and fuse local and global information in the dataset, as well as the correlations between different data modalities.

[0071] Optionally, the aforementioned neighbor information can be aggregated neighbor information. In a graph data structure, the neighbor information of nodes can reflect the interaction between local environments or related entities. For example, in vehicle power supply fault diagnosis, sensor nodes for input voltage and input current may be neighbors, and the data association between them can reveal the real-time status of the power input. By aggregating neighbor information, the model can learn the spatial dependencies between sensor data, thereby better understanding local fault modes. The aforementioned homomodal node information can refer to multiple sources of the same type of data, such as temperature data collected by multiple temperature sensors. Aggregating homomodal node information can capture the differences and trends of the same type of data at different locations or at different times. The aforementioned heteromodal node information can be used to fuse the relationships between different data types, such as temperature data and voltage data. In many cases, faults are caused by abnormal combinations of different modal data. For example, if the temperature rises abnormally while the current drops abnormally during charging of a vehicle power supply, it may indicate a fault in the charging circuit or the battery itself. By aggregating heteromodal node information, the model can learn cross-modal dependencies, which is crucial for a comprehensive understanding of the causes of faults.

[0072] Optionally, by utilizing neighbor information, same-modal node information, and different-modal node information, a more comprehensive and detailed feature representation can be constructed. These features not only include the data information of individual nodes but also incorporate the spatial correlations and temporal dependencies between nodes, as well as the interactions between data from different modalities. Such a feature representation can provide richer context for fault diagnosis tasks, helping the model to more accurately identify faults and improve the real-time performance and accuracy of diagnosis.

[0073] In this stage, the MST-GAT model can use multi-head attention, intra-modal attention, and inter-modal attention to aggregate basic neighbor information, intra-modal node information, and inter-modal node information, respectively, thereby obtaining node spatial features that fuse complex modal relationships. This process can be viewed as the model learning how to extract the most valuable information from different perspectives and different sensor types.

[0074] Optionally, the above-mentioned initial graph attention network model can be an initial multimodal graph attention network, which can be used to aggregate the neighbor information, same-modal node information and different-modal node information of nodes using a multi-head attention mechanism, so as to extract spatiotemporal information from different perspectives and modes. This model can be used to determine the interaction between different modal data, such as the relationship between temperature and current, and how the two jointly affect the occurrence of faults.

[0075] Optionally, after constructing an initial graph attention network model using neighbor information, same-modality node information, and different-modality node information, the initial graph attention network model can be updated using a loss function to obtain a trained graph attention network model.

[0076] As an optional implementation, the initial graph attention network model is trained using historical fault diagnosis results to obtain a fault diagnosis model. This includes: extracting features from the sub-historical operating parameter sets corresponding to nodes in the fault knowledge graph to obtain trend features of the sub-historical operating parameter sets, wherein the trend features are used to characterize the changing trends of historical operating parameters at different historical moments in the sub-historical operating parameter sets; constructing an initial temporal convolutional network model using the trend features; and training the initial graph attention network model using the initial temporal convolutional network model and historical fault diagnosis results to obtain the fault diagnosis model.

[0077] In this embodiment, feature extraction is performed on the sub-historical operating parameter sets corresponding to nodes in the fault knowledge graph to obtain trend features of the sub-historical operating parameter sets. These trend features can be used to characterize the changing trends of historical operating parameters at different historical moments in the sub-historical operating parameter sets, and can serve as high-order spatiotemporal fusion features characterizing long-term trends. Using the trend features, an initial temporal convolutional network model is constructed. Using the initial temporal convolutional network model and historical fault diagnosis results, an initial graph attention network model is trained to obtain a fault diagnosis model.

[0078] Optionally, a one-dimensional causal dilated convolution is employed to capture long-term temporal dependencies, and residual connections are introduced to stabilize training, thereby obtaining high-order spatiotemporal fusion features representing long-term trends. One-dimensional causal dilated convolution is used to capture long-term dependencies in time series. This method can effectively capture the impact of past data on the current state because many failures accumulate gradually rather than occurring instantaneously.

[0079] Optionally, after constructing the initial temporal convolutional network model, a loss function can be used to update the initial temporal convolutional network model to obtain a trained temporal convolutional network model. Based on the trained graph attention network model and the temporal convolutional network model, a trained fault diagnosis model can be constructed.

[0080] For example, the structure of the initial temporal convolutional network model is predefined, which may include, but is not limited to, one-dimensional convolutional layers, dilated convolutions, and residual connections. The preprocessed trend features are input into the pre-built model for training. The model parameters are updated using optimization algorithms such as backpropagation and gradient descent to minimize the difference between the predicted fault categories and the actual fault categories, thus obtaining the initial graph attention network model.

[0081] As an optional implementation, an initial graph attention network model is trained using an initial temporal convolutional network model and historical fault diagnosis results to obtain a fault diagnosis model. This includes: using the initial graph attention network model to extract features from a historical set of operating parameters to obtain first feature data; using the initial temporal convolutional network model to extract features from the first feature data to obtain second feature data; fusing the first and second feature data to obtain third feature data; identifying the third feature data to obtain an identification result; constructing a loss function based on the identification result and historical fault diagnosis results; using the loss function to update the initial graph attention network model and the initial temporal convolutional network model respectively to obtain a graph attention network model and a temporal convolutional network model; and combining the graph attention network model and the temporal convolutional network model to obtain the fault diagnosis model.

[0082] In this embodiment, after constructing the initial graph attention network model and the initial temporal convolutional network model, a loss function can be constructed based on the historical operating parameter set and historical fault diagnosis results. The loss function is used to update the model parameters of the initial graph attention network model and the initial temporal convolutional network model to obtain the trained graph attention network model and temporal convolutional network model. The graph attention network model and the temporal convolutional network model can be combined to obtain the fault diagnosis model.

[0083] In this embodiment, a weighted joint loss function can be used, comprising reconstruction loss and classification loss. The reconstruction loss measures the accuracy of the model in reconstructing the input data, while the classification loss measures the accuracy of the model in classifying faults. Furthermore, an L2 regularization term can be added to prevent overfitting. The first feature data mentioned above can be fused multimodal features, which can be spatiotemporal features, including both temporal and spatial features, and can be used to characterize diverse information from different sensors. The second feature data mentioned above can be trend features or time series features.

[0084] Optionally, a loss function can be defined by combining mean squared error and binary cross-entropy to simultaneously optimize the performance of data reconstruction and fault classification.

[0085] In this embodiment, a weighted joint loss function is defined as the output, thereby providing a multi-task learning framework for the subsequent model training stage that can simultaneously optimize reconstruction and prediction capabilities. Mean squared error and binary cross-entropy are used as the loss functions, and the overall loss function is shown below:

[0086] in: For reconstruction loss. For classification loss. This is for L2 regularization. These are the weighting coefficients. This represents the original input data for the i-th sample. This represents the i-th sample data output by the model reconstruction. N is the batch size. Let be the true label of the i-th sample. This represents the probability distribution predicted by the model.

[0087] Optionally, a phased training approach can be adopted to improve accuracy by splitting the training order of tasks, adjusting the learning rate, and the number of rounds.

[0088] Figure 4 This is a flowchart illustrating the construction of a fault diagnosis model according to an embodiment of this application, such as... Figure 4 As shown, constructing a fault diagnosis model may include the following steps:

[0089] Step S402: Construct the graph structure.

[0090] In this embodiment, a set of historical operating parameters is obtained; the set of historical operating parameters is preprocessed to obtain a multidimensional time series dataset; based on the multidimensional time series dataset, a fault knowledge graph of the vehicle power supply is constructed to complete the construction of the graph structure.

[0091] Step S404: Construct the graph attention network model.

[0092] In this embodiment, neighbor information, same-modal node information, and different-modal node information of the fault knowledge graph are obtained; using the neighbor information, same-modal node information, and different-modal node information, an initial graph attention network model is constructed.

[0093] Step S406: Construct a temporal convolutional network model.

[0094] In this embodiment, feature extraction is performed on the sub-historical operating parameter sets corresponding to the nodes in the fault knowledge graph to obtain the trend features of the sub-historical operating parameter sets; using the trend features, an initial temporal convolutional network model is constructed.

[0095] Step S408: Define the loss function.

[0096] In this embodiment, a loss function is predefined. Based on the loss function, the initial graph attention network model and the initial temporal convolutional network model are updated to obtain the graph attention network model and the temporal convolutional network model. The graph attention network model and the temporal convolutional network model are combined to obtain the fault diagnosis model.

[0097] This application's embodiments utilize a Multimodal Spatiotemporal Graph Attention Network (MST-GAT) model to effectively extract temporal features from multimodal sensor data, adaptively fuse multimodal information, and highlight key features through a spatiotemporal attention mechanism, thereby significantly improving the accuracy and real-time performance of fault diagnosis. Simultaneously, the diagnostic report provides maintenance personnel with detailed fault analysis information, enhancing the credibility and practicality of the diagnostic results.

[0098] Optionally, a multi-scale spatiotemporal attention diagnostic model is used to extract complex spatiotemporal dependent features from the multimodal data of the vehicle power system. Compared with traditional methods, this model can more sensitively capture early and hidden faults and adaptively fuse multi-scale information to improve the accuracy and real-time performance of fault diagnosis. Simultaneously, it supports online learning, enabling continuous model optimization as the vehicle operates and adapting to changes in data distribution caused by system variations.

[0099] In this embodiment, a fault diagnosis model is pre-trained based on historical operating parameters. When fault diagnosis of the vehicle power supply is required, an operating parameter set can be obtained. The operating parameter set is preprocessed to obtain a standardized time series dataset. This standardized time series dataset can be used to characterize the operating status of the vehicle power supply over time. The fault diagnosis model is used to diagnose the fault on the standardized time series dataset to obtain the fault diagnosis result, thereby achieving the technical effect of improving the fault identification accuracy of the vehicle power supply and solving the technical problem of low fault identification accuracy of the vehicle power supply.

[0100] According to an embodiment of this application, a fault identification device for an on-board power supply is also provided. It should be noted that the fault identification device for an on-board power supply in this embodiment can be used to execute the fault identification method for an on-board power supply described above in this application.

[0101] Figure 5 This is a schematic diagram of a fault identification device for an on-board power supply according to an embodiment of this application. Figure 5 As shown, the vehicle power supply fault identification device 50 may include: an acquisition unit 502, a processing unit 504, and a diagnostic unit 506.

[0102] The acquisition unit 502 is used to acquire the set of operating parameters of the vehicle power supply during the current operation process. The set of operating parameters includes multiple sub-sets of operating parameters collected by multiple sensors in the vehicle. Each sub-set of operating parameters includes multiple operating parameters collected at the current time. The operating parameters are used to characterize the current operating state of the vehicle power supply.

[0103] The processing unit 504 is used to preprocess the set of operating parameters to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating status of the vehicle power supply over time.

[0104] The diagnostic unit 506 is used to perform fault diagnosis on a standardized time series dataset using a trained fault diagnosis model to obtain fault diagnosis results. The fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set. The historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors. Each sub-historical operating parameter set includes multiple historical operating parameters collected at multiple historical moments. The historical operating parameters are used to characterize the historical operating state of the vehicle power supply.

[0105] The vehicle power supply fault identification device of this embodiment acquires the set of operating parameters of the vehicle power supply during its current operation through an acquisition unit. The set of operating parameters includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, each sub-operating parameter set including multiple operating parameters collected at the current time. These operating parameters characterize the current operating state of the vehicle power supply. A processing unit preprocesses the operating parameter set to obtain a standardized time-series dataset, which characterizes the operating state of the vehicle power supply over time. A diagnosis unit uses a trained fault diagnosis model to diagnose faults in the standardized time-series dataset, obtaining fault diagnosis results. The fault diagnosis model is trained based on historical operating parameter sets of the vehicle power supply during its historical operation and corresponding historical fault diagnosis results. The historical operating parameter sets include multiple sub-historical operating parameter sets collected by multiple sensors, each sub-historical operating parameter set including multiple historical operating parameters collected at multiple historical times. These historical operating parameters characterize the historical operating state of the vehicle power supply. This achieves the technical effect of improving the accuracy of vehicle power supply fault identification and solves the technical problem of low fault identification accuracy in vehicle power supplies.

[0106] Embodiments of this application may provide a computer terminal, which may be any computer terminal device in a group of computer terminals. Optionally, in this embodiment, the aforementioned computer terminal may also be replaced by a mobile terminal or other terminal device.

[0107] Optionally, in this embodiment, the computer terminal may be located in at least one of a plurality of network devices in a computer network.

[0108] In this embodiment, the aforementioned computer terminal can execute the program code for the following steps in the vehicle power supply fault identification method: obtaining the operating parameter set of the vehicle power supply during its current operation, wherein the operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, and the sub-operating parameter sets include multiple operating parameters collected at multiple current times, the operating parameters being used to characterize the current operating state of the vehicle power supply; preprocessing the operating parameter set to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating state of the vehicle power supply changing over time; using a trained fault diagnosis model to perform fault diagnosis on the standardized time series dataset to obtain a fault diagnosis result, wherein the fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during its historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set, the historical operating parameter set including multiple sub-historical operating parameter sets collected by multiple sensors, the sub-historical operating parameter sets including multiple historical operating parameters collected at multiple historical times, the historical operating parameters being used to characterize the historical operating state of the vehicle power supply.

[0109] Optionally, Figure 6 This is a structural block diagram of a computer terminal according to an embodiment of this application, such as... Figure 6 As shown, the computer terminal 608 may include one or more (only one is shown in the figure) processors 602, memory 604, and transmission devices 606.

[0110] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the vehicle power supply fault identification method and device in this application embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned vehicle power supply fault identification method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to computer terminal 608 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0111] The processor can access information and applications stored in the memory via a transmission device to perform the following steps: acquiring the operating parameter set of the vehicle's on-board power supply during its current operation, wherein the operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, each sub-operating parameter set including multiple operating parameters collected at current times, and the operating parameters characterizing the current operating state of the on-board power supply; preprocessing the operating parameter set to obtain a standardized time-series dataset, wherein the standardized time-series dataset characterizes the operating state of the on-board power supply over time; and using a trained fault diagnosis model to perform fault diagnosis on the standardized time-series dataset to obtain fault diagnosis results, wherein the fault diagnosis model is trained based on historical operating parameter sets of the on-board power supply during historical operation and corresponding historical fault diagnosis results, wherein the historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors, each sub-historical operating parameter set including multiple historical operating parameters collected at multiple historical times, and the historical operating parameters characterizing the historical operating state of the on-board power supply.

[0112] Those skilled in the art will understand that Figure 6 The structure shown is for illustrative purposes only. The computer terminal 608 can also be a smartphone (such as an Android phone, an iOS phone, etc.), a tablet computer, a handheld computer, a mobile internet device (MID), a PAD, and other terminal devices. Figure 6 This does not limit the structure of the computer terminal 608 described above. For example, the computer terminal 608 may also include components that are more advanced than those described above. Figure 6 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 6 The different configurations shown.

[0113] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0114] According to an embodiment of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored program, wherein the program executes the vehicle power supply fault identification method in the above embodiments.

[0115] Optionally, in this embodiment, the computer-readable storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0116] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: obtaining the operating parameter set of the vehicle power supply during the current operation process, wherein the operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, the sub-operating parameter sets include multiple operating parameters collected at multiple current times, and the operating parameters are used to characterize the current operating state of the vehicle power supply; preprocessing the operating parameter set to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating state of the vehicle power supply changing over time; using a trained fault diagnosis model to perform fault diagnosis on the standardized time series dataset to obtain fault diagnosis results, wherein the fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set, the historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors, the sub-historical operating parameter sets include multiple historical operating parameters collected at multiple historical times, and the historical operating parameters are used to characterize the historical operating state of the vehicle power supply.

[0117] Optionally, the aforementioned computer-readable storage medium may also execute program code that performs the following steps: using a graph attention network model in the fault diagnosis model to extract features from a standardized time series dataset to obtain spatiotemporal feature data corresponding to the sensor; using a temporal convolutional network model in the fault diagnosis model to extract features from the spatiotemporal feature data to obtain trend features among multiple operating parameters in the operating parameter set; fusing the spatiotemporal feature data and trend features to obtain fused features; and identifying the fused features to obtain fault diagnosis results.

[0118] Optionally, the aforementioned computer-readable storage medium may also execute program code that performs the following steps: acquiring a set of historical operating parameters; preprocessing the set of historical operating parameters to obtain a multidimensional time-series dataset, wherein the multidimensional time-series dataset is used to characterize the operating state of the vehicle power supply over time during historical operation; constructing a fault knowledge graph of the vehicle power supply based on the multidimensional time-series dataset, wherein nodes in the fault knowledge graph are used to characterize sensors, and edges in the fault knowledge graph are used to characterize the degree of correlation between historical operating parameters collected by different sensors; and training a fault diagnosis model using the fault knowledge graph and historical fault diagnosis results.

[0119] Optionally, the aforementioned computer-readable storage medium may also execute program code that performs the following steps: converting historical operating parameters in a multidimensional time series dataset into multiple state vectors; identifying the similarity between the multiple state vectors, wherein the similarity is used to characterize the degree of association between the multiple state vectors; using the similarity, determining at least one target operating parameter adjacent to the historical operating parameter from the multidimensional time series dataset, and the connection weight between the target operating parameter and the historical operating parameter, wherein the connection weight is positively correlated with the degree of association; using the sensors that collect historical operating parameters and the sensors that collect target operating parameters as nodes, and using the connection weight, constructing edges between the nodes to obtain a fault knowledge graph.

[0120] Optionally, the aforementioned computer-readable storage medium may also execute program code that performs the following steps: acquiring neighbor information, same-modality node information, and different-modality node information of the fault knowledge graph, wherein the neighbor information is used to characterize the correlation between historical operating parameters collected by sensors in adjacent nodes, the same-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of similar types, and the different-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of different types; constructing an initial graph attention network model using the neighbor information, same-modality node information, and different-modality node information; and training the initial graph attention network model using historical fault diagnosis results to obtain a fault diagnosis model.

[0121] Optionally, the aforementioned computer-readable storage medium may also execute program code that performs the following steps: extracting features from the sub-historical operating parameter sets corresponding to nodes in the fault knowledge graph to obtain trend features of the sub-historical operating parameter sets, wherein the trend features are used to characterize the changing trends of historical operating parameters at different historical moments in the sub-historical operating parameter sets; using the trend features to construct an initial temporal convolutional network model; and using the initial temporal convolutional network model and historical fault diagnosis results to train the initial graph attention network model to obtain a fault diagnosis model.

[0122] Optionally, the aforementioned computer-readable storage medium may also execute program code that performs the following steps: using an initial graph attention network model to extract features from a set of historical operating parameters to obtain first feature data; using an initial temporal convolutional network model to extract features from the first feature data to obtain second feature data; fusing the first and second feature data to obtain third feature data; identifying the third feature data to obtain an identification result; constructing a loss function based on the identification result and historical fault diagnosis results; using the loss function to update the initial graph attention network model and the initial temporal convolutional network model respectively to obtain a graph attention network model and a temporal convolutional network model; and combining the graph attention network model and the temporal convolutional network model to obtain a fault diagnosis model.

[0123] In this embodiment, a fault diagnosis model is pre-trained based on historical operating parameters. When fault diagnosis of the vehicle power supply is required, an operating parameter set can be obtained. The operating parameter set is preprocessed to obtain a standardized time series dataset. This standardized time series dataset can be used to characterize the operating status of the vehicle power supply over time. The fault diagnosis model is used to diagnose the fault on the standardized time series dataset to obtain the fault diagnosis result. This achieves the technical effect of improving the fault identification accuracy of the vehicle power supply and solves the technical problem of low fault identification accuracy of the vehicle power supply.

[0124] According to an embodiment of this application, a processor is also provided for running a program, wherein the fault identification method for the vehicle power supply in the above embodiment is executed when the program is run by the processor.

[0125] Optionally, in this embodiment, the computer terminal may be located in at least one of a plurality of network devices in a computer network.

[0126] In this embodiment, the aforementioned computer terminal can execute the program code for the following steps in the vehicle power supply fault identification method: obtaining the operating parameter set of the vehicle power supply during its current operation, wherein the operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, and the sub-operating parameter sets include multiple operating parameters collected at multiple current times, the operating parameters being used to characterize the current operating state of the vehicle power supply; preprocessing the operating parameter set to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating state of the vehicle power supply changing over time; using a trained fault diagnosis model to perform fault diagnosis on the standardized time series dataset to obtain a fault diagnosis result, wherein the fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during its historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set, the historical operating parameter set including multiple sub-historical operating parameter sets collected by multiple sensors, the sub-historical operating parameter sets including multiple historical operating parameters collected at multiple historical times, the historical operating parameters being used to characterize the historical operating state of the vehicle power supply.

[0127] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the vehicle power supply fault identification method and device in this application embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned vehicle power supply fault identification method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0128] The processor can access information and applications stored in the memory via a transmission device to perform the following steps: acquiring the operating parameter set of the vehicle's on-board power supply during its current operation, wherein the operating parameter set includes multiple sub-operating parameter sets collected by multiple sensors in the vehicle, each sub-operating parameter set including multiple operating parameters collected at current times, and the operating parameters characterizing the current operating state of the on-board power supply; preprocessing the operating parameter set to obtain a standardized time-series dataset, wherein the standardized time-series dataset characterizes the operating state of the on-board power supply over time; and using a trained fault diagnosis model to perform fault diagnosis on the standardized time-series dataset to obtain fault diagnosis results, wherein the fault diagnosis model is trained based on historical operating parameter sets of the on-board power supply during historical operation and corresponding historical fault diagnosis results, wherein the historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors, each sub-historical operating parameter set including multiple historical operating parameters collected at multiple historical times, and the historical operating parameters characterizing the historical operating state of the on-board power supply.

[0129] Optionally, the processor may also execute program code that performs the following steps: using the graph attention network model in the fault diagnosis model to extract features from the standardized time series dataset to obtain the spatiotemporal feature data corresponding to the sensor; using the temporal convolutional network model in the fault diagnosis model to extract features from the spatiotemporal feature data to obtain the trend features among multiple operating parameters in the operating parameter set; fusing the spatiotemporal feature data and the trend features to obtain the fused features; and identifying the fused features to obtain the fault diagnosis result.

[0130] Optionally, the processor may also execute program code for the following steps: acquiring a historical operating parameter set; preprocessing the historical operating parameter set to obtain a multidimensional time series dataset, wherein the multidimensional time series dataset is used to characterize the operating state of the vehicle power supply over time during historical operation; constructing a fault knowledge graph of the vehicle power supply based on the multidimensional time series dataset, wherein nodes in the fault knowledge graph are used to characterize sensors, and edges in the fault knowledge graph are used to characterize the degree of correlation between historical operating parameters collected by different sensors; and training a fault diagnosis model using the fault knowledge graph and historical fault diagnosis results.

[0131] Optionally, the processor may also execute program code that performs the following steps: converting historical operating parameters in a multidimensional time series dataset into multiple state vectors; identifying the similarity between the multiple state vectors, wherein the similarity is used to characterize the degree of association between the multiple state vectors; using the similarity, determining at least one target operating parameter adjacent to the historical operating parameter from the multidimensional time series dataset, and the connection weight between the target operating parameter and the historical operating parameter, wherein the connection weight is positively correlated with the degree of association; using the sensors that collect historical operating parameters and the sensors that collect target operating parameters as nodes, and using the connection weights, constructing edges between the nodes to obtain a fault knowledge graph.

[0132] Optionally, the processor may also execute program code for the following steps: acquiring neighbor information, same-modality node information, and different-modality node information of the fault knowledge graph, wherein the neighbor information is used to characterize the correlation between historical operating parameters collected by sensors in adjacent nodes, the same-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of similar types, and the different-modality node information is used to characterize the correlation between historical operating parameters collected by sensors of different types; constructing an initial graph attention network model using the neighbor information, same-modality node information, and different-modality node information; and training the initial graph attention network model using historical fault diagnosis results to obtain a fault diagnosis model.

[0133] Optionally, the processor may also execute program code for the following steps: extracting features from the sub-historical operating parameter sets corresponding to nodes in the fault knowledge graph to obtain trend features of the sub-historical operating parameter sets, wherein the trend features are used to characterize the changing trends of historical operating parameters at different historical moments in the sub-historical operating parameter sets; using the trend features to construct an initial temporal convolutional network model; and using the initial temporal convolutional network model and historical fault diagnosis results to train the initial graph attention network model to obtain a fault diagnosis model.

[0134] Optionally, the processor may also execute program code that performs the following steps: using an initial graph attention network model to extract features from a set of historical operating parameters to obtain first feature data; using an initial temporal convolutional network model to extract features from the first feature data to obtain second feature data; fusing the first and second feature data to obtain third feature data; recognizing the third feature data to obtain a recognition result; constructing a loss function based on the recognition result and historical fault diagnosis results; using the loss function to update the initial graph attention network model and the initial temporal convolutional network model respectively to obtain a graph attention network model and a temporal convolutional network model; and combining the graph attention network model and the temporal convolutional network model to obtain a fault diagnosis model.

[0135] By adopting the embodiments of this application, a fault diagnosis model is pre-trained based on historical operating parameters. When it is necessary to diagnose a fault in the vehicle power supply, an operating parameter set can be obtained. The operating parameter set is preprocessed to obtain a standardized time series dataset. This standardized time series dataset can be used to characterize the operating status of the vehicle power supply over time. The fault diagnosis model is used to diagnose the fault in the standardized time series dataset to obtain the fault diagnosis result. This achieves the technical effect of improving the fault identification accuracy of the vehicle power supply and solves the technical problem of low fault identification accuracy of the vehicle power supply.

[0136] According to an embodiment of this application, a computer program product is also provided, which includes computer instructions, wherein when the computer instructions are executed by a processor, they implement the vehicle power supply fault identification method in the above embodiments.

[0137] Embodiments of this application may provide an electronic device that may include a memory and a processor.

[0138] Figure 7 This is a block diagram of an electronic device for a fault identification method of an on-board power supply according to an embodiment of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present application described and / or claimed herein.

[0139] like Figure 7As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded from storage unit 708 into random access memory (RAM) 703. RAM 703 can also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0140] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 704, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0141] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as data verification methods. For example, in some embodiments, the data verification method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the data verification method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform a data verification method by any other suitable means (e.g., by means of firmware).

[0142] According to an embodiment of this application, a fault identification method for vehicle power supply is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0143] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0144] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0145] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0146] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display, monitor); and a keyboard and pointing device (e.g., a mouse or pathball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0147] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0148] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0149] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0150] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0151] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0152] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0153] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0154] 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 this application, in essence, or the part 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0155] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for fault identification of vehicle power supply, characterized in that, include: The system acquires a set of operating parameters of the vehicle's on-board power supply during its current operation. The set of operating parameters includes multiple sub-sets of operating parameters collected by multiple sensors in the vehicle. Each sub-set of operating parameters includes multiple operating parameters collected at the current moment. These operating parameters are used to characterize the current operating state of the on-board power supply. The set of operating parameters is preprocessed to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating status of the vehicle power supply over time. Using a trained fault diagnosis model, fault diagnosis is performed on the standardized time series dataset to obtain fault diagnosis results. The fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during its historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set. The historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors. Each sub-historical operating parameter set includes multiple historical operating parameters collected at multiple historical moments. The historical operating parameters are used to characterize the historical operating state of the vehicle power supply.

2. The method according to claim 1, characterized in that, The process of using a trained fault diagnosis model to diagnose faults in the standardized time series dataset and obtaining fault diagnosis results includes: Using the graph attention network model in the fault diagnosis model, feature extraction is performed on the standardized time series dataset to obtain the spatiotemporal feature data corresponding to the sensor. By using the temporal convolutional network model in the fault diagnosis model, feature extraction is performed on the spatiotemporal feature data to obtain the trend features among multiple operating parameters in the operating parameter set; By fusing the spatiotemporal feature data and the trend features, a fused feature is obtained; The fusion features are identified to obtain the fault diagnosis results.

3. The method according to claim 1, characterized in that, The method further includes: Obtain the set of historical operating parameters; The historical operating parameter set is preprocessed to obtain a multidimensional time series dataset, wherein the multidimensional time series dataset is used to characterize the operating status of the vehicle power supply as a function of time during the historical operation. Based on the multidimensional time series dataset, a fault knowledge graph of the vehicle power supply is constructed, wherein the nodes in the fault knowledge graph are used to represent the sensors, and the edges in the fault knowledge graph are used to represent the degree of correlation between the historical operating parameters collected by different sensors. The fault diagnosis model is trained using the fault knowledge graph and the historical fault diagnosis results.

4. The method according to claim 3, characterized in that, The process of constructing a fault knowledge graph for the vehicle power supply based on the multidimensional time series dataset includes: The historical operating parameters in the multidimensional time series dataset are converted into multiple state vectors; Identify the similarity among multiple state vectors, wherein the similarity is used to characterize the degree of association among the multiple state vectors; Using the similarity, at least one target operating parameter adjacent to the historical operating parameter is determined from the multidimensional time series dataset, and the connection weight between the target operating parameter and the historical operating parameter is determined, wherein the connection weight is positively correlated with the degree of association; The fault knowledge graph is obtained by using the sensors that collect the historical operating parameters and the sensors that collect the target operating parameters as nodes, and constructing the edges between the nodes using the connection weights.

5. The method according to claim 3, characterized in that, The step of training the fault diagnosis model using the fault knowledge graph and the historical fault diagnosis results includes: Obtain neighbor information, same-modality node information, and different-modality node information of the fault knowledge graph. The neighbor information is used to characterize the correlation between the historical operating parameters collected by the sensors in adjacent nodes. The same-modality node information is used to characterize the correlation between the historical operating parameters collected by sensors of similar types. The different-modality node information is used to characterize the correlation between the historical operating parameters collected by sensors of different types. An initial graph attention network model is constructed using the neighbor information, the same-modality node information, and the different-modality node information. The initial graph attention network model is trained using the historical fault diagnosis results to obtain the fault diagnosis model.

6. The method according to claim 5, characterized in that, The step of training the initial graph attention network model using the historical fault diagnosis results to obtain the fault diagnosis model includes: Feature extraction is performed on the sub-historical operating parameter set corresponding to the node in the fault knowledge graph to obtain the trend feature of the sub-historical operating parameter set, wherein the trend feature is used to characterize the change trend of the historical operating parameter at different historical moments in the sub-historical operating parameter set; Using the aforementioned trend features, an initial temporal convolutional network model is constructed. The initial temporal convolutional network model and the historical fault diagnosis results are used to train the initial graph attention network model to obtain the fault diagnosis model.

7. The method according to claim 6, characterized in that, The step of training the initial graph attention network model using the initial temporal convolutional network model and the historical fault diagnosis results to obtain the fault diagnosis model includes: Using the initial graph attention network model, features are extracted from the historical operating parameter set to obtain the first feature data; Using the initial temporal convolutional network model, feature extraction is performed on the first feature data to obtain the second feature data; The first feature data and the second feature data are fused to obtain the third feature data; The third feature data is identified to obtain the identification result; Based on the identification results and the historical fault diagnosis results, a loss function is constructed. Using the loss function, the initial graph attention network model and the initial temporal convolutional network model are updated respectively to obtain the graph attention network model and the temporal convolutional network model; The fault diagnosis model is obtained by combining the graph attention network model and the temporal convolutional network model.

8. A fault identification device for vehicle power supply, characterized in that, include: The acquisition unit is used to acquire the set of operating parameters of the vehicle power supply during the current operation process. The set of operating parameters includes multiple sub-sets of operating parameters collected by multiple sensors in the vehicle. Each sub-set of operating parameters includes multiple operating parameters collected at the current time. The operating parameters are used to characterize the current operating state of the vehicle power supply. The processing unit is used to preprocess the set of operating parameters to obtain a standardized time series dataset, wherein the standardized time series dataset is used to characterize the operating status of the vehicle power supply over time. The diagnostic unit is used to perform fault diagnosis on the standardized time series dataset using a trained fault diagnosis model to obtain fault diagnosis results. The fault diagnosis model is trained based on the historical operating parameter set of the vehicle power supply during historical operation and the historical fault diagnosis results corresponding to the historical operating parameter set. The historical operating parameter set includes multiple sub-historical operating parameter sets collected by multiple sensors. The sub-historical operating parameter sets include multiple historical operating parameters collected at multiple historical moments. The historical operating parameters are used to characterize the historical operating state of the vehicle power supply.

9. A vehicle, characterized in that, Used to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 7.

Images