Electric vehicle charging compatible fault diagnosis method based on knowledge graph

Abstract

The invention discloses an electric vehicle charging compatible fault diagnosis method based on a knowledge graph, and the method comprises the following steps: constructing the knowledge graph, carrying out the fault diagnosis reasoning, and updating the knowledge graph, and the construction of the knowledge graph comprises the following steps: collecting multi-source data in the charging process of an electric vehicle, knowledge extraction is performed by applying a natural language processing technology and a data mining algorithm, knowledge from different sources is fused, a knowledge graph is constructed in a graph form, nodes represent entities, edges represent relationships between the entities, and attributes are attached to the nodes or the edges. According to the method, the knowledge graph is constructed, the electric vehicle charging multi-source data is comprehensively collected, knowledge is extracted and fused by using natural language processing and a data mining algorithm, and during fault diagnosis reasoning, input data is preprocessed based on combination of rules and machine learning, for example, abnormal value detection by using Z-score, missing value filling by using linear interpolation and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle fault diagnosis, and specifically to a knowledge graph-based electric vehicle charging compatibility fault diagnosis method. Background Art

[0002] As the number of electric vehicles has exploded, the scale of charging infrastructure construction has continued to expand. In order to improve product competitiveness, different car companies have continuously innovated in the design of electric vehicles. Charging pile manufacturers are also actively developing new charging technologies to meet market demand. This has made the technical combination between electric vehicles and charging piles increasingly diverse, and the probability of charging compatibility failures has increased significantly. At the same time, users have increasing expectations for the convenience and reliability of electric vehicle charging. Once a charging compatibility failure occurs, it will not only affect the user's normal travel, but may also cause safety hazards. Therefore, the need for efficient and accurate charging compatibility fault diagnosis methods has become a key issue that the industry urgently needs to solve.

[0003] Currently, the diagnosis of electric vehicle charging compatibility failures has the following main defects:

[0004] 1. Traditional diagnostic methods are often based on a single data source or simple rule matching, which fails to fully consider the complex electrical and communication aspects of the charging process, as well as the interaction between the vehicle and the charging station. For example, fault diagnosis based solely on charging voltage and current thresholds can easily overlook compatibility differences between different vehicle models and charging station brands, leading to misdiagnosis or missed diagnosis.

[0005] 2. Existing technologies cannot effectively integrate a large amount of dispersed charging fault knowledge, nor can they deeply explore the potential relationship between fault symptoms and causes. When encountering new or combined fault conditions, it is difficult to quickly and accurately locate the fault and analyze the cause. For example, when a communication fault and an electrical fault occur simultaneously, it is difficult to determine the causal relationship between the two and their combined impact on charging compatibility.

[0006] 3. With the rapid development of electric vehicle technology and the continuous updating of charging pile standards, traditional diagnostic methods are difficult to adapt to new models, charging protocols and fault types in a timely manner. Traditional diagnostic methods lack effective response measures for new technologies adopted by new electric vehicles, such as wireless charging technology and new battery management systems.

[0007] In summary, it is necessary to propose a knowledge graph-based electric vehicle charging compatibility fault diagnosis method to solve the above-mentioned problems. Summary of the Invention

[0008] In view of the fact that traditional diagnostic methods pointed out in the above background technology are mostly based on a single data source or simple rule matching, it is difficult to fully consider the technical problems of complex electrical, communication, and vehicle-charging pile interaction factors in the charging process. The purpose of the present invention is to provide an electric vehicle charging compatibility fault diagnosis method based on a knowledge graph.

[0009] To achieve the purpose of the present invention, the technical solution provided by the present invention is as follows:

[0010] A method for diagnosing electric vehicle charging compatibility faults based on a knowledge graph includes the following steps:

[0011] S1: Constructing a knowledge graph: Collect multi-source data from the electric vehicle charging process, apply natural language processing technology and data mining algorithms to extract knowledge, integrate knowledge from different sources, and construct a knowledge graph in the form of a graph, where nodes represent entities, edges represent relationships between entities, and attributes are attached to nodes or edges; the multi-source data includes vehicle parameters, charging pile parameters, and fault data;

[0012] S2: Fault diagnosis reasoning: Input fault data, vehicle parameters, and charging pile parameter information into the constructed knowledge graph. Using a combination of rule-based reasoning and machine learning-based reasoning, the knowledge graph is matched and analyzed to output the cause of the fault and the corresponding solution.

[0013] S3: Knowledge graph update: Real-time monitoring of data during electric vehicle charging and fault diagnosis results. When new fault types appear or existing fault causes and solutions change, the knowledge graph update algorithm is used to integrate the new knowledge into the existing knowledge graph.

[0014] Furthermore, in S1, the vehicle parameters include the manufacturing year of the vehicle, the motor type, and the charging history data of the vehicle, and the charging history data covers the start time, end time, charging amount, and charging location information of each charging.

[0015] Furthermore, in S1, a deep learning model based on the attention mechanism is used to perform entity alignment and relationship completion on multi-source data. The entity types in the electric vehicle charging field include four categories: vehicle parameters, charging pile specifications, charging protocol versions, and environmental parameters. There are no less than 15 types of relationships between entities.

[0016] Furthermore, in S2, fault diagnosis reasoning adopts a two-layer reasoning architecture. The first layer performs rule matching in the knowledge graph through the Cypher query language, and the second layer uses a graph convolutional neural network to capture fault propagation patterns. The machine learning model uses negative sampling technology to generate virtual fault cases during training to enhance generalization capabilities.

[0017] Furthermore, in S2, data preprocessing is first performed on the input multi-source data, including the identification and correction of outliers, filling of missing values, and smoothing of time series data using sliding window technology to ensure the data quality of the input knowledge graph and improve the accuracy of diagnostic reasoning;

[0018] The specific steps of data preprocessing are:

[0019] 1) Outlier processing: The Z-score method is used to detect data points that deviate from the mean ±3σ, and outliers are replaced by the median to retain the original distribution characteristics;

[0020] 2) Missing value filling: Linear interpolation is used for continuous missing data, and discrete missing fields are filled with the historical mean of the same model to ensure data integrity;

[0021] 3) Time series smoothing: Apply a 5-point sliding average window to process time series data, balancing noise suppression and dynamic feature retention.

[0022] Furthermore, in S3, an embedding vector update mechanism containing meta-path constraints automatically triggers a meta-learning-based parameter initialization process when a new fault type occurs, and verifies the confidence threshold of the newly added knowledge through a Bayesian network to ensure graph consistency.

[0023] Furthermore, in S3, after the confidence threshold of the newly added knowledge is verified by the Bayesian network, an incremental update strategy is adopted to update only the affected local knowledge graph, reducing the computational complexity and time cost of the update. At the same time, a consistency check is performed on the updated knowledge graph. If inconsistency is found, a backtracking mechanism is used to undo the most recent update operation and re-verify and update it.

[0024] Furthermore, in S3, the knowledge graph update algorithm includes:

[0025] S31: Data Monitoring and Screening: Classify the real-time monitored electric vehicle charging process data and fault diagnosis results based on data source, type, and timestamp, and filter out data that will cause knowledge graph updates, such as data corresponding to new fault types or data related to existing fault causes and solutions that have changed;

[0026] S32: Knowledge Extraction and Conversion: For the filtered data, use natural language processing technology and information extraction methods to extract entity, relationship and attribute information, and convert the extracted information into a format acceptable to the knowledge graph to ensure consistency with the data structure of the existing knowledge graph;

[0027] S33: Conflict detection and processing: Compare the newly extracted knowledge with the existing knowledge graph to detect whether there is a conflict; if there is a conflict, use a rule-based and similarity matching method to handle it;

[0028] S34: Knowledge Fusion and Update: Integrate the processed new knowledge into the existing knowledge graph. For new entities, create new nodes in the knowledge graph; for new relationships, add new edges; for attribute updates, modify the attribute values ​​of the corresponding nodes or edges. During the update process, adjust the embedding vectors of entities and relationships according to the embedding vector update mechanism of meta-path constraints to ensure the semantic consistency of the knowledge graph.

[0029] S35: Verification and evaluation: Use the Bayesian network to verify the confidence of the newly added knowledge and set a confidence threshold. Only when the confidence of the newly added knowledge exceeds the threshold, the update is confirmed to be valid. At the same time, by comparing the performance of the knowledge graph in fault diagnosis reasoning before and after the update, the update effect is evaluated. If the effect is not good, consider readjusting the update strategy.

[0030] Furthermore, S3 further includes a dynamic update mechanism driven by user feedback. When the end user submits a new fault diagnosis case or corrects an existing diagnosis result through the interactive interface, the system automatically triggers the knowledge fusion and update process, integrates the user feedback information with the real-time monitoring data, integrates the multi-source update basis through the confidence weighted algorithm, and generates an update log to record the user contribution for the optimization of subsequent update strategies.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. This application constructs a knowledge graph to comprehensively collect multi-source data on electric vehicle charging, uses natural language processing and data mining algorithms to extract and integrate knowledge, and combines rule-based and machine learning in fault diagnosis reasoning, and pre-processes input data, such as using Z-score to detect outliers and linear interpolation to fill missing values. After actual testing, the fault diagnosis accuracy rate is improved by more than 20% after adopting a two-layer reasoning architecture, effectively avoiding the misdiagnosis and missed diagnosis problems caused by traditional methods that rely only on a single data source or simple rule matching and ignore the compatibility differences between vehicle models and charging pile brands.

[0033] 2. This application uses a powerful knowledge graph update mechanism to monitor charging data and diagnostic results in real time. Once a new fault type appears or existing knowledge changes, the new knowledge is integrated into the graph through processes such as data monitoring and screening, knowledge extraction and conversion, and conflict detection and processing. There is a meta-path constraint embedded vector update mechanism to ensure semantic consistency, and the confidence is verified through a Bayesian network. It also includes a user feedback-driven dynamic update mechanism, which can keep up with the development of electric vehicle technology and charging facilities, promptly incorporate new fault knowledge, and continuously maintain diagnostic effectiveness and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A schematic diagram of the flow chart of the electric vehicle charging compatibility fault diagnosis method based on the knowledge graph provided in an embodiment of the present invention;

[0035] Figure 2 A framework diagram of the electric vehicle charging compatibility fault diagnosis method based on knowledge graph provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following is a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts are within the scope of protection of the present invention.

[0037] like Figure 1 and Figure 2 This embodiment provides a method for diagnosing electric vehicle charging compatibility faults based on a knowledge graph, comprising the following steps:

[0038] S1: Constructing a knowledge graph: Collect multi-source data from the electric vehicle charging process, apply natural language processing technology and data mining algorithms to extract knowledge, integrate knowledge from different sources, and construct a knowledge graph in the form of a graph, where nodes represent entities, edges represent relationships between entities, and attributes are attached to nodes or edges;

[0039] S2: Fault diagnosis reasoning: Input information such as charging compatibility fault phenomena, vehicle and charging pile parameters into the constructed knowledge graph. Using a combination of rule-based reasoning and machine learning-based reasoning, the knowledge graph is matched and analyzed to output the cause of the fault and the corresponding solution.

[0040] S3: Knowledge graph update: Real-time monitoring of data during electric vehicle charging and fault diagnosis results. When new fault types appear or existing fault causes and solutions change, the knowledge graph update algorithm is used to integrate the new knowledge into the existing knowledge graph.

[0041] In this embodiment, the method can integrate multi-source data, mine hidden relationships therein, and present complex charging field knowledge in a structured manner by constructing a knowledge graph. This makes fault diagnosis no longer rely on single data or simple rules, greatly improving diagnostic accuracy. In the fault diagnosis reasoning stage, based on the combination of rules and machine learning, it can quickly match fault phenomena, deeply analyze and locate causes, and provide accurate solutions. The knowledge graph update mechanism ensures that the method can keep pace with the development of electric vehicle technology and charging facilities, promptly incorporate new fault types, causes and solutions, and continuously maintain the effectiveness and adaptability of diagnosis.

[0042] Preferably, the above steps are described in detail below:

[0043] The construction of the knowledge graph specifically includes the following:

[0044] (1) Multi-source data collection

[0045] Vehicle parameter collection: Through the vehicle management system interface, the manufacturing year and motor type information of electric vehicles of different brands and models are obtained. For charging history data, the data recorded by the vehicle's built-in driving recorder or battery management system (BMS) is used to cover the start time, end time, charge amount, and charging location of each charge. For example, the charging records of a certain brand of electric vehicle's BMS system are regularly collected over the past year, including detailed information on 500 charging events.

[0046] Charging pile parameter collection: Connect to the charging pile management platform to obtain specifications such as the charging pile's brand, model, power, and charging protocol version. At the same time, deploy sensors to collect environmental parameters around the charging pile, such as temperature, humidity, and altitude. Taking a public charging station as an example, parameters of 10 charging piles of different brands in the station are collected, and environmental parameter data is updated every hour.

[0047] Fault data collection: Collect feedback from electric vehicle users when they encounter faults during charging, including descriptions of the fault phenomenon, the status of the vehicle and charging pile at the time of the fault, etc. At the same time, obtain fault repair records from the after-sales databases of automobile manufacturers and charging pile suppliers, including the cause of the fault and the solution. Within one month, a total of 200 user-reported fault information and 150 after-sales repair records were collected.

[0048] (2) Knowledge Extraction

[0049] Application of natural language processing technology: Use the named entity recognition (NER) algorithm to process text data such as fault phenomenon descriptions and extract entities. For example, from the sentence "The vehicle displays charging interruption during charging, and the charging pile indicator light flashes abnormally", extract fault phenomenon entities such as "charging interruption" and "charging pile indicator light flashes abnormally". Use the relationship extraction algorithm to determine the relationship between entities, such as causal relationship, association relationship, etc. For example, from the sentence "Charging interruption due to poor contact of the charging interface", extract the causal relationship between "poor contact of the charging interface" and "charging interruption".

[0050] Application of data mining algorithms: Cluster analysis, association rule mining and other algorithms are used to extract knowledge from collected structured data, such as vehicle parameters, charging pile parameters and fault data. For example, cluster analysis revealed that electric vehicles of a certain brand and model are prone to unstable charging power when using charging piles with a certain version of the charging protocol, thereby establishing a correlation between the vehicle, charging pile and fault.

[0051] (3) Entity alignment and relationship completion of deep learning models based on attention mechanism

[0052] A deep learning model based on the attention mechanism is used to perform entity alignment and relationship completion on multi-source heterogeneous data. Taking vehicle parameters and charging pile parameters as examples, the model learns the feature representations of entities in different data sources, such as the brand, model, battery capacity and other features of the vehicle, and the brand, model, power and other features of the charging pile, and uses the attention mechanism to focus on the importance of different features to entity alignment and relationship completion, so as to accurately align the same entities in different data sources and complete the possible missing relationships between entities. In actual applications, after model processing, the accuracy of entity alignment reached more than 95%, and the recall rate of relationship completion reached more than 90%.

[0053] (4) Knowledge fusion and graph construction

[0054] The extracted and aligned knowledge is integrated to eliminate knowledge conflicts and redundancies. For example, different descriptions of the same fault cause in different data sources are standardized. Then, a graph database (such as Neo4j) is used to construct a knowledge graph in the form of a graph. Nodes represent entities, edges represent the relationships between entities, and attributes are attached to nodes or edges. For example, a "car model A" node is created and connected to nodes such as the charging interface standard, common fault phenomena, and applicable charging pile brands corresponding to the car model through edges. Corresponding attributes are added to the nodes and edges, such as the manufacturing year of the car model and the power of the charging pile.

[0055] The fault diagnosis reasoning specifically includes the following:

[0056] (1) Data preprocessing

[0057] Outlier processing: Taking charging current data as an example, the Z-score method is used to detect outliers in the charging current data of a certain vehicle model collected over a period of time. The mean and standard deviation of the data are calculated, and data points that deviate from the mean by ±3σ are identified as outliers. For example, in a set of 1,000 charging current data points, 5 outliers are detected. Then, these outliers are replaced by the median to retain the distribution characteristics of the original data.

[0058] Missing value filling: For continuously missing data, such as missing temperature data within a certain period of time during the charging process, linear interpolation is used to fill in the missing data. The temperature estimate for the missing time period is calculated based on the temperature values ​​at the previous and next time points. For discrete missing fields, such as missing vehicle model information, the historical mean of the same vehicle model is used to fill in the missing data. For example, in a fault record, if the vehicle model information is missing, the common parameters of the vehicle model can be obtained as fill-in values ​​by querying the information of other vehicles of the same brand and model.

[0059] Time series smoothing: For time series data such as charge capacity and temperature, a 5-point sliding average window is applied for smoothing. For example, for charge capacity data recorded every 10 minutes, the average of the current data point and the two data points before and after it is taken as the smoothed data point. This suppresses noise while retaining the dynamic change characteristics of the data.

[0060] (2) Two-layer reasoning architecture

[0061] First-level rule matching: When a charging compatibility fault is detected, information such as the fault phenomenon, vehicle and charging pile related parameters, etc. is input into the constructed knowledge graph, and rule matching is performed in the knowledge graph through the Cypher query language. For example, if the fault phenomenon is "charging interruption", the knowledge graph is queried for fault causes and solutions related to "charging interruption", such as "poor contact of charging interface" and "charging pile communication failure", and corresponding solution suggestions are obtained, such as "check the connection status of the charging interface" and "restart the charging pile". At the same time, the confidence of the matching results is calculated. If the confidence is less than 60%, the second-level reasoning is automatically entered.

[0062] The second layer of graph convolutional neural network reasoning: Graph convolutional neural networks are used to capture fault propagation patterns. The node and edge information in the knowledge graph is used as input. The relationship characteristics between nodes are learned through convolution operations to predict possible fault causes. During model training, negative sampling technology is used to generate virtual fault cases to enhance the generalization ability of the model. For example, some node pairs that do not have fault relationships are randomly extracted from the knowledge graph as negative samples for training. In this way, the model can better identify node relationships in real fault situations and improve the accuracy of fault diagnosis. In actual tests, after adopting a two-layer reasoning architecture, the accuracy of fault diagnosis has increased by more than 20%.

[0063] The knowledge graph update specifically includes the following:

[0064] (1) Data monitoring and screening

[0065] Real-time monitoring of data and fault diagnosis results during the charging process of electric vehicles; continuous collection of new data through data acquisition equipment and fault feedback systems, classification of collected data, and division based on data source (vehicle, charging pile, user feedback, etc.), type (vehicle parameters, fault phenomenon, diagnostic results, etc.) and timestamp, to filter out data that may lead to knowledge graph updates, such as data corresponding to new fault types, or data related to and changing existing fault causes and solutions. For example, during a week of monitoring, it was found that a new type of electric vehicle had a never-before-recorded fault phenomenon "vehicle battery overheating alarm during charging" when using a new type of charging pile. This data was filtered out as candidate data for knowledge graph updates.

[0066] 2. Knowledge Extraction and Transformation

[0067] For the filtered data, natural language processing technology and information extraction methods are used to extract entity, relationship and attribute information. For example, from the sentence "The vehicle battery overheats and alarms during charging due to unstable power output of the charging pile", "vehicle battery overheating alarm" is extracted as the fault phenomenon entity, "unstable power output of the charging pile" is extracted as the fault cause entity, and the causal relationship between the two is also extracted. Then, the extracted information is converted into a format acceptable to the knowledge graph to ensure consistency with the data structure of the existing knowledge graph. For example, the newly extracted entity and relationship information is converted into node and edge creation statements in the graph database.

[0068] (3) Conflict detection and handling

[0069] Compare the newly extracted knowledge with the existing knowledge graph to detect whether there is a conflict. If there is a conflict, use a rule-based and similarity matching method to handle it. For example, the newly extracted fault cause is different from the description of the same fault cause in the existing knowledge graph, but through semantic similarity calculation, it is found that the two have similar meanings, so they are merged into the same fault cause. In actual operation, the semantic similarity threshold is set to 0.8, and the merge is performed when the similarity exceeds this threshold.

[0070] (4) Knowledge integration and updating

[0071] The processed new knowledge is integrated into the existing knowledge graph. For new entities, new nodes are created in the knowledge graph; for new relationships, new edges are added; for attribute updates, the attribute values ​​of the corresponding nodes or edges are modified. During the update process, the embedding vectors of entities and relationships are adjusted according to the embedding vector update mechanism of meta-path constraints to ensure the semantic consistency of the knowledge graph. For example, when a "new type of charging pile" entity is added, a corresponding node is created, and edges are added according to its relationship with other entities (such as vehicles, fault phenomena, etc.). At the same time, the embedding vector update mechanism of meta-path constraints is used to update the embedding vectors of related entities and relationships, so that the knowledge graph can accurately reflect the new semantic relationship.

[0072] (V) Verification and evaluation

[0073] The confidence of the newly added knowledge is verified using a Bayesian network, with a confidence threshold of 70%. The update is only considered valid when the confidence of the newly added knowledge exceeds this threshold. At the same time, the update effect is evaluated by comparing the performance of the knowledge graph in fault diagnosis reasoning before and after the update. For example, after updating the knowledge graph, a new set of fault cases is used for testing, and indicators such as fault diagnosis accuracy and recall rate before and after the update are compared. If the effect is not good, consider readjusting the update strategy, such as adjusting the knowledge extraction algorithm and optimizing the conflict detection rules.

[0074] (6) Dynamic update mechanism driven by user feedback

[0075] In the knowledge graph update, it further includes a dynamic update mechanism driven by user feedback. When the end user submits a new fault diagnosis case or corrects the existing diagnosis result through the interactive interface, the system automatically triggers the knowledge fusion and update process, integrates the user feedback information with the real-time monitoring data, and integrates the multi-source update basis through the confidence weighted algorithm. For example, for the fault cases reported by users, a higher confidence weight is given because the user's actual experience can better reflect the real situation. At the same time, an update log is generated to record the user's contribution for the optimization of subsequent update strategies. For example, the number of valid fault cases submitted by users, the number of corrections to the diagnosis results, and other information are recorded to reward users who actively contribute, and the update process of the knowledge graph is optimized according to the characteristics of user feedback.

[0076] Through the above specific implementation methods, the present invention can effectively construct a knowledge graph, perform fault diagnosis reasoning, and update the knowledge graph in real time, thereby improving the accuracy and efficiency of electric vehicle charging compatibility fault diagnosis and adapting to the continuous development of electric vehicle technology and charging facilities.

[0077] In summary, this knowledge graph-based electric vehicle charging compatibility fault diagnosis method, by constructing a knowledge graph, comprehensively collects multi-source data on electric vehicle charging, uses natural language processing and data mining algorithms to extract and integrate knowledge, and combines rule-based and machine learning in fault diagnosis reasoning, and preprocesses the input data, such as using Z-score to detect outliers and linear interpolation to fill missing values. After actual testing, the fault diagnosis accuracy is improved by more than 20% after adopting a two-layer reasoning architecture, effectively avoiding the misdiagnosis and missed diagnosis problems caused by traditional methods relying only on a single data source or simple rule matching, and ignoring the compatibility differences between vehicle models and charging pile brands.

[0078] In addition, the knowledge graph update mechanism is powerful, monitoring charging data and diagnostic results in real time. Once a new fault type appears or existing knowledge changes, the new knowledge will be integrated into the graph through processes such as data monitoring and screening, knowledge extraction and conversion, and conflict detection and processing. There is a meta-path constraint embedded vector update mechanism to ensure semantic consistency, and the confidence is verified through the Bayesian network. It also includes a user feedback-driven dynamic update mechanism, which can keep up with the development of electric vehicle technology and charging facilities, promptly incorporate new fault knowledge, and continuously maintain diagnostic effectiveness and adaptability.

[0079] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0080] Finally, it should be noted that the above embodiments are merely examples and illustrations of the present invention and are not intended to limit the present invention to the described embodiments. Furthermore, those skilled in the art will appreciate that the present invention is not limited to the above embodiments and that various variations and modifications may be made based on the teachings of the present invention, all of which fall within the scope of the present invention.

Claims

1. A knowledge graph-based electric vehicle charging compatibility fault diagnosis method, characterized by: The following steps are involved: S1: Constructing a knowledge graph: Collect multi-source data from the electric vehicle charging process, apply natural language processing technology and data mining algorithms to extract knowledge, integrate knowledge from different sources, and construct a knowledge graph in the form of a graph, where nodes represent entities, edges represent relationships between entities, and attributes are attached to nodes or edges; the multi-source data includes vehicle parameters, charging pile parameters, and fault data; S2: Fault diagnosis reasoning: Input fault data, vehicle parameters, and charging pile parameter information into the constructed knowledge graph. Using a combination of rule-based reasoning and machine learning-based reasoning, the knowledge graph is matched and analyzed to output the cause of the fault and the corresponding solution. S3: Knowledge graph update: Real-time monitoring of data during electric vehicle charging and fault diagnosis results. When new fault types appear or existing fault causes and solutions change, the knowledge graph update algorithm is used to integrate the new knowledge into the existing knowledge graph.

2. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: In S1, the vehicle parameters include the manufacturing year of the vehicle, the motor type, and the charging history data of the vehicle. The charging history data covers the start time, end time, charging amount, and charging location information of each charging.

3. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: In S1, a deep learning model based on the attention mechanism is used to perform entity alignment and relationship completion on multi-source data. The entity types in the electric vehicle charging field include four categories: vehicle parameters, charging pile specifications, charging protocol version, and environmental parameters. There are no less than 15 types of relationships between entities.

4. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: In S2, fault diagnosis reasoning adopts a two-layer reasoning architecture. The first layer uses the Cypher query language to match rules in the knowledge graph, and the second layer uses a graph convolutional neural network to capture fault propagation patterns. Negative sampling technology is used during machine learning model training to generate virtual fault cases to enhance generalization capabilities.

5. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: In S2, data preprocessing is first performed on the input multi-source data, including the identification and correction of outliers, filling in missing values, and smoothing of time series data using sliding window technology to ensure the data quality of the input knowledge graph and improve the accuracy of diagnostic reasoning. The specific steps of data preprocessing are: 1) Outlier processing: The Z-score method is used to detect data points that deviate from the mean ±3σ, and outliers are replaced by the median to retain the original distribution characteristics; 2) Missing value filling: Linear interpolation is used for continuous missing data, and discrete missing fields are filled with the historical mean of the same model to ensure data integrity; 3) Time series smoothing: Apply a 5-point sliding average window to process time series data, balancing noise suppression and dynamic feature retention.

6. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: In S3, an embedding vector update mechanism that includes meta-path constraints automatically triggers a meta-learning-based parameter initialization process when a new fault type occurs, and verifies the confidence threshold of the newly added knowledge through a Bayesian network to ensure graph consistency.

7. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 6 is characterized by: In S3, after the confidence threshold of the newly added knowledge is verified by the Bayesian network, an incremental update strategy is adopted to update only the affected local knowledge graph, reducing the computational complexity and time cost of the update. At the same time, the updated knowledge graph is checked for consistency. If inconsistency is found, a backtracking mechanism is used to undo the most recent update operation and re-verify and update it.

8. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: In S3, the knowledge graph update algorithm includes: S31: Data Monitoring and Screening: Classify the real-time monitored electric vehicle charging process data and fault diagnosis results based on data source, type, and timestamp, and filter out data that will cause knowledge graph updates, such as data corresponding to new fault types or data related to existing fault causes and solutions that have changed; S32: Knowledge Extraction and Conversion: For the filtered data, use natural language processing technology and information extraction methods to extract entity, relationship and attribute information, and convert the extracted information into a format acceptable to the knowledge graph to ensure consistency with the data structure of the existing knowledge graph; S33: Conflict detection and processing: Compare the newly extracted knowledge with the existing knowledge graph to detect whether there is a conflict; if there is a conflict, use a rule-based and similarity matching method to handle it; S34: Knowledge Fusion and Update: Integrate the processed new knowledge into the existing knowledge graph. For new entities, create new nodes in the knowledge graph; for new relationships, add new edges; for attribute updates, modify the attribute values ​​of the corresponding nodes or edges. During the update process, adjust the embedding vectors of entities and relationships according to the embedding vector update mechanism of meta-path constraints to ensure the semantic consistency of the knowledge graph. S35: Verification and evaluation: Use the Bayesian network to verify the confidence of the newly added knowledge and set a confidence threshold. Only when the confidence of the newly added knowledge exceeds the threshold, the update is confirmed to be valid. At the same time, by comparing the performance of the knowledge graph in fault diagnosis reasoning before and after the update, the update effect is evaluated. If the effect is not good, consider readjusting the update strategy.

9. The electric vehicle charging compatibility fault diagnosis method based on knowledge graph according to claim 1 is characterized by: S3 further includes a dynamic update mechanism driven by user feedback. When the end user submits a new fault diagnosis case or corrects an existing diagnosis result through the interactive interface, the system automatically triggers the knowledge fusion and update process, integrates the user feedback information with the real-time monitoring data, integrates the multi-source update basis through the confidence weighted algorithm, and generates an update log to record the user contribution for the optimization of subsequent update strategies.

Images