Charging pile fault diagnosis method, device, equipment and storage medium

Abstract

The present invention relates to a charging pile fault diagnosis method, device, equipment and storage medium, comprising the following steps: based on the fault characteristic data, fault identification is performed on the components of the target charging pile to obtain potential fault components; based on the component map, a map is constructed for the potential fault components to obtain a potential fault component association map; based on the potential fault component association map, the influence relationship between the potential fault components is analyzed to obtain a fault propagation path; through a preset Bayesian network model, the failure probability of the potential fault component is calculated based on the fault propagation path; the potential fault component corresponding to the largest failure probability is taken as the fault source, and the fault type of the fault source is analyzed through a preset classification algorithm, and the maintenance strategy of the target charging pile is formulated based on the fault type, thereby solving the technical problem that the diagnosis of charging pile faults mainly relies on manual inspections and user feedback, and has slow response and poor accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of charging piles, and in particular to a charging pile fault diagnosis method, device, equipment and storage medium. Background Art

[0002] With the popularity of electric vehicles, charging piles are one of the important infrastructures of electric vehicles. Their stability and reliability are directly related to the travel experience of electric vehicle users. However, in actual use, charging piles frequently fail, which not only affects the charging efficiency of electric vehicles, but also may bring safety hazards. At present, the diagnosis of charging pile failures mainly relies on manual inspections and user feedback. This method has the problems of slow response and poor accuracy, and cannot meet the growing demand for electric vehicle charging.

[0003] There are also technical challenges in the research background. Existing charging pile fault diagnosis methods mostly use traditional fault detection methods, such as current and voltage monitoring. Although these methods can detect some obvious faults, they have limited ability to analyze the deep-seated fault causes. In addition, due to the complexity of the charging pile system, the interaction between different components may cause fault propagation, and a single fault detection method is difficult to fully and accurately locate the source of the fault. Therefore, it is particularly urgent to develop a method that can quickly and accurately diagnose charging pile faults.

[0004] Finally, with the development of big data and artificial intelligence technology, new ideas and technical support are provided for charging pile fault diagnosis. The use of big data analysis can dig out abnormal patterns in the operation of charging piles, and artificial intelligence algorithms, especially the application of machine learning and Bayesian network models, can effectively improve the accuracy and efficiency of fault diagnosis. However, how to effectively apply these advanced technologies to charging pile fault diagnosis is still a problem that needs in-depth research. The above-mentioned research method attempts to achieve accurate diagnosis of charging pile faults by combining feature extraction, fault identification, fault propagation path analysis and Bayesian network models of fault status data, thereby providing a scientific basis for the maintenance of charging piles. Summary of the invention

[0005] The main purpose of the present invention is to provide a charging pile fault diagnosis method, device, equipment and storage medium, which solves the technical problem that the diagnosis of charging pile faults mainly relies on manual inspections and user feedback, and has slow response and poor accuracy.

[0006] To achieve the above object, the present invention provides a charging pile fault diagnosis method, comprising the following steps:

[0007] Acquire fault status data of each component of the target charging pile, and perform feature extraction on the fault status data to obtain fault feature data;

[0008] Performing fault identification on components of the target charging pile based on the fault characteristic data to obtain potential fault components;

[0009] Acquire a component map of the target charging pile from a database, and construct a map of the potential faulty components based on the component map to obtain a potential faulty component association map;

[0010] Analyze the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path;

[0011] Calculating the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model;

[0012] The potential fault component corresponding to the greatest fault probability is taken as the fault source, and the fault type of the fault source is analyzed through a preset classification algorithm, and a maintenance strategy for the target charging pile is formulated based on the fault type.

[0013] Furthermore, the acquisition of fault status data of each component of the target charging pile and feature extraction of the fault status data to obtain fault feature data include:

[0014] The fault parameters of each component in the target charging pile are collected by using multiple sensors arranged inside the target charging pile to obtain original fault state parameters;

[0015] Performing parameter clustering on the original fault state parameters to obtain clustered fault parameters; wherein the clustered fault parameters include voltage parameters, current parameters and temperature parameters;

[0016] Performing time-frequency analysis on the clustered fault parameters to obtain fault time-frequency characteristic data;

[0017] Performing dimension reduction processing on the fault time-frequency characteristic data to obtain dimension-reduced fault parameters;

[0018] Feature extraction is performed on the dimension-reduced fault parameters to obtain fault feature data.

[0019] Further, the performing fault identification on the components of the target charging pile based on the fault characteristic data to obtain potential fault components includes:

[0020] Acquire historical charging pile fault feature data, perform tensor decomposition on the historical charging pile fault feature data to obtain a three-dimensional tensor feature matrix, and perform dimensionality reduction processing on the three-dimensional tensor feature matrix to obtain a core tensor and a factor matrix;

[0021] Performing non-negative matrix decomposition on the core tensor and factor matrix to obtain a historical fault feature subspace; wherein the historical fault feature subspace includes typical fault mode information of various components of the historical charging pile;

[0022] Performing fault identification learning on the historical fault feature subspace through a preset initial fault identification algorithm to obtain a fault identification algorithm;

[0023] Performing time-frequency fault identification on the fault feature data based on the fault identification algorithm to obtain time-frequency fault feature data;

[0024] Based on the time-frequency fault characteristic data, fault identification is performed on components of the target charging pile to obtain potential fault components.

[0025] Furthermore, constructing a map of the potential faulty components based on the component map to obtain a potential faulty component association map includes:

[0026] Performing structural analysis on the component atlas of the target charging pile to obtain structural relationship analysis results between the components;

[0027] Constructing a component topology relationship matrix based on the structural relationship analysis results;

[0028] Based on the component topology relationship matrix, the interaction mode between the components of the target charging pile is captured to obtain the interaction mode between the components; wherein the interaction mode between the components is the action relationship and connection mode between the components;

[0029] Modeling the influence relationship of the potential fault components based on the interaction mode between the components to obtain the fault relationship of the potential fault components;

[0030] By using a preset graph attention network, based on the fault relationship of the potential fault component, the potential fault component is weighted in the component graph of the target charging pile to obtain a fault component weight coefficient;

[0031] The corresponding potential faulty components are connected based on the faulty component weight coefficients to obtain a potential faulty component association map.

[0032] Furthermore, the analysis of the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path includes:

[0033] Based on the potential fault component association graph, the node centrality of each potential fault component is calculated to obtain the centrality index of each potential fault component; wherein the node centrality includes degree centrality, closeness centrality and betweenness centrality;

[0034] Analyze the influence relationship between potential fault components based on the centrality index;

[0035] Based on the influence relationship between the potential fault components, the potential fault components in the fault component association map are divided into communities to identify potential fault components with similar fault propagation characteristics, and obtain a dynamic community structure of potential fault components;

[0036] A causal relationship analysis is performed on potential fault components in the dynamic community structure through a preset causal reasoning model to obtain a fault propagation causal path, and the fault propagation causal path is used as the fault propagation path.

[0037] Furthermore, the calculation of the failure probability of the potential fault component based on the fault propagation path by using a preset Bayesian network model includes:

[0038] Structural modeling of the fault propagation path is performed using a preset Bayesian network model to obtain a Bayesian network structure of the fault propagation;

[0039] Based on the Bayesian network structure, fault parameter estimation is performed on each potential fault component in the fault propagation path to obtain a conditional probability distribution of each potential fault component;

[0040] Performing a causal effect analysis on the potential fault components based on the conditional probability distribution to obtain a causal effect analysis result; wherein the causal effect analysis result is a causal influence result of each potential fault component;

[0041] Based on the causal effect analysis result, the fault expansion of the potential fault component in the time dimension is performed to obtain a dynamic fault propagation probability;

[0042] A robustness calculation is performed on the potential fault component based on the dynamic fault propagation probability to obtain the failure probability of the potential fault component.

[0043] Furthermore, analyzing the fault type of the fault source by a preset classification algorithm and formulating a maintenance strategy for the target charging pile based on the fault type includes:

[0044] Performing multi-dimensional feature extraction on the fault source corresponding to the maximum fault probability to obtain a multi-modal fault feature;

[0045] Classifying the multi-modal fault features into fault types using a preset classification algorithm to obtain a preliminary fault type;

[0046] Performing knowledge reasoning on the preliminary fault type to obtain a fault type report;

[0047] Preliminarily designing a maintenance strategy for the target charging pile based on the fault type report to obtain a preliminary maintenance plan;

[0048] Performing simulation verification on the fault source based on the preliminary maintenance plan to obtain a simulation verification result;

[0049] The simulation verification result is evaluated to obtain a simulation verification evaluation value. If the simulation verification evaluation value is less than a preset simulation verification evaluation value, the preliminary maintenance plan is optimized and adjusted based on the simulation verification result to obtain a maintenance strategy for the target charging pile.

[0050] The present invention also provides a charging pile fault diagnosis device, comprising:

[0051] An acquisition module is used to acquire the fault status data of each component of the target charging pile, and perform feature extraction on the fault status data to obtain fault feature data;

[0052] An identification module, used to identify faults of components of the target charging pile based on the fault characteristic data to obtain potential fault components;

[0053] A construction module, used for obtaining a component map of a target charging pile from a database, and constructing a map of the potential faulty components based on the component map to obtain a potential faulty component association map;

[0054] An analysis module, configured to analyze the influence relationship between the potential fault components based on the potential fault component association map to obtain a fault propagation path;

[0055] A calculation module, used to calculate the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model;

[0056] A formulation module is used to take the potential fault component corresponding to the greatest fault probability as the fault source, analyze the fault type of the fault source through a preset classification algorithm, and formulate a maintenance strategy for the target charging pile based on the fault type.

[0057] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0058] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0059] The charging pile fault diagnosis method provided by the present invention comprises the following steps: acquiring fault status data of each component of the target charging pile, and extracting features of the fault status data to obtain fault feature data; performing fault identification on the components of the target charging pile based on the fault feature data to obtain potential fault components; constructing a map of the potential fault components based on the component map to obtain a potential fault component association map; analyzing the influence relationship between the potential fault components based on the potential fault component association map to obtain a fault propagation path; calculating the fault probability of the potential fault component based on the fault propagation path through a preset Bayesian network model; taking the potential fault component corresponding to the maximum fault probability as the fault source, and analyzing the fault type of the fault source through a preset classification algorithm, and formulating a maintenance strategy for the target charging pile based on the fault type. Through the above-mentioned technical means, the technical problem that the diagnosis of charging pile faults mainly relies on manual inspections and user feedback, and has slow response and poor accuracy is solved, and the potential fault component corresponding to the maximum fault probability is taken as the fault source, and its fault type is further analyzed, and a maintenance strategy is formulated accordingly. This process not only reduces maintenance costs, but also extends the service life of the equipment, ensuring the long-term stable operation of the charging pile system. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of the steps of a charging pile fault diagnosis method according to an embodiment of the present invention;

[0061] Figure 2 is a structural block diagram of a charging pile fault diagnosis device in one embodiment of the present invention;

[0062] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0063] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0065] like Figure 1 As shown, Figure 1 It is a schematic diagram of the steps of a charging pile fault diagnosis method in one embodiment of the present invention;

[0066] In one embodiment of the present invention, a charging pile fault diagnosis method is provided, comprising the following steps:

[0067] Step S1, acquiring fault status data of each component of a target charging pile, and performing feature extraction on the fault status data to obtain fault feature data.

[0068] Specifically, the process of obtaining the fault status data of each component of the target charging pile, extracting features from the fault status data, and obtaining fault feature data is one of the core links in the charging pile fault diagnosis method. First, it is necessary to collect the working status information of each component through various sensors installed inside the charging pile. These sensors can monitor multiple parameters including current, voltage, temperature, etc. For example, in an electric vehicle charging station, when an electric vehicle is connected to the charging pile to start charging, the current sensor in the charging pile will continuously monitor the current intensity in the charging circuit, the voltage sensor will record the change of the supply voltage during the charging process, and the temperature sensor is used to detect the temperature of the charging interface and internal electronic components to ensure that they are within the normal working range. The data collected by these sensors constitutes the basis of the fault status data. Once these raw data are collected, the next step is to extract features from the fault status data to obtain more useful information. Feature extraction refers to extracting key indicators or patterns that can reflect the operating status of the system from a large amount of raw data. For example, by analyzing the changing trends of current and voltage, it can be found whether there is a risk of overload or short circuit; by analyzing the temperature data, it can be evaluated whether overheating occurs. This process usually involves signal processing techniques, such as filtering, Fourier transform, etc., as well as data analysis methods, such as statistical analysis, pattern recognition, etc. The goal of feature extraction is to filter out the characteristic quantities that best characterize the health status of the charging pile from the original data. These characteristic quantities are the fault characteristic data. In this way, not only can the amount of data be effectively reduced, which is convenient for subsequent processing, but also the accuracy and efficiency of fault diagnosis can be improved. Ultimately, these fault characteristic data will be used in subsequent fault identification and analysis steps to help technicians quickly locate the problem and take corresponding maintenance measures in a timely manner.

[0069] Step S2: performing fault identification on components of the target charging pile based on the fault characteristic data to obtain potential fault components.

[0070] Specifically, based on the fault feature data, fault identification of the components of the target charging pile is performed to obtain potential faulty components, which is a key step in the fault diagnosis process of the charging pile. After completing the collection and feature extraction of the fault status data, the next task is to use these fault feature data to identify which components may have faults. This process usually involves applying machine learning algorithms or pattern recognition techniques to train models to identify the differences between specific fault modes and normal operating modes. For example, in the actual application scenario of an electric vehicle charging station, assume that we have collected a series of fault feature data from multiple sensors such as current, voltage, and temperature. After preliminary processing, these data have been converted into a series of feature vectors reflecting the health status of the charging pile. Next, these feature vectors can be analyzed by pre-trained classifiers (such as support vector machines, random forests, or neural networks). When training these models, historical data sets containing known fault cases and normal operating cases are used so that the models can learn to distinguish different fault types. When the model receives new fault feature data, it can predict which components may have problems based on the previously learned knowledge. For example, if the model detects a sudden increase in current and an accompanying rise in temperature over a period of time, this may be a sign of poor contact or a short circuit somewhere in the charging circuit, and the model will mark the components involved in that circuit as potential faulty components. In this way, not only can potentially faulty components be identified quickly and accurately, but the false alarm rate can also be reduced, improving the overall efficiency of fault diagnosis. Ultimately, this process provides an important foundation for subsequent fault propagation path analysis and fault source location, allowing maintenance personnel to perform more targeted maintenance, thereby shortening fault handling time and improving the reliability and user experience of charging piles.

[0071] Step S3, obtaining a component map of the target charging pile from a database, and constructing a map of the potential faulty components based on the component map to obtain a potential faulty component association map.

[0072] Specifically, the component map of the target charging pile is obtained from the database, and the potential fault components are constructed based on the component map to obtain the potential fault component association map, which is an important part of the charging pile fault diagnosis method. In the previous steps, we have identified the components that may have faults through the analysis of fault feature data. However, in order to more accurately understand the relationship between these potential fault components and their impact on the entire system, it is necessary to further use the component map of the charging pile for in-depth analysis. First, the component map of the target charging pile is obtained from the database. The component map mentioned here is actually a document or data model that records in detail the internal structure of the charging pile, the connection mode between each component and its function. For example, in the application scenario of an electric vehicle charging station, the component map of the charging pile will include key components such as the power input port, the main control board, the charging interface, the cooling system, and the electrical connection and physical position relationship between them. This information is usually provided by the manufacturer of the charging pile and stored in a special database for later maintenance and fault diagnosis. After obtaining the component map, the next step is to construct the map of the identified potential fault components based on this map to obtain the potential fault component association map. This means that we need to put the previously identified potential fault components in the structural framework of the entire charging pile and analyze their connections with other components. For example, if the charging interface is found to be overheated through the analysis of fault feature data, then it is necessary to check the component map to understand the specific connection method between the charging interface and the main control board and the power input port, and whether these components share certain circuits or cooling channels. The purpose of this is to reveal the possible causal relationship between potential fault components and their impact on the performance of other components or the entire system. By constructing a potential fault component association map, it can not only help technicians understand the fault propagation path more intuitively, but also provide strong support for subsequent fault propagation path analysis and fault source location. For example, if it is found that the overheated charging interface is associated with the decline in the efficiency of the cooling system, then it can be preliminarily determined that the problem with the cooling system may be the root cause of the overheating of the charging interface. This component map-based analysis method helps to improve the accuracy and efficiency of fault diagnosis, thereby speeding up troubleshooting, reducing the downtime of the charging pile, and ensuring the continuity and stability of charging services.

[0073] Step S4, analyzing the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path.

[0074] Specifically, the influence relationship between each potential fault component is analyzed based on the potential fault component association graph to obtain the fault propagation path. This process is a key step in the charging pile fault diagnosis method. By combining graph theory and causal reasoning technology, the modeling and analysis of the complex influence relationship between potential fault components is realized. First, by constructing a potential fault component association graph, each potential fault component and the connection relationship between them are visualized. For example, in an electric vehicle charging station, components such as the charging interface, cooling system, and main control board are nodes, and the connection relationship between them is an edge. Next, by analyzing the node centrality in the graph, such as degree centrality, closeness centrality, and betweenness centrality, the importance of each component in the network is evaluated. For example, the degree centrality of the charging interface is high, indicating that it is directly connected to multiple other components and may be a key node for fault propagation. Based on these centrality indicators, the mutual influence relationship between each potential fault component can be further analyzed. For example, if the temperature increase of the charging interface causes the load of the cooling system to increase, which in turn affects the temperature of the main control board, then it can be considered that the charging interface has a strong influence on the cooling system and the main control board. Through these analyses, potential fault components with similar fault propagation characteristics can be identified to form a dynamic community structure. Finally, through the preset causal reasoning model, the causal relationship analysis of the potential fault components in these dynamic community structures is carried out to obtain the specific fault propagation path. For example, through the causal reasoning model, it can be determined that the increase in the temperature of the charging interface causes the increase in the cooling system load, which in turn affects the temperature of the main control board. This causal path is the fault propagation path. Through this series of steps, not only can the critical path of fault propagation be accurately identified, but it also provides a scientific basis for locating the source of the fault and formulating maintenance strategies.

[0075] Step S5, calculating the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model.

[0076] Specifically, calculating the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model is an important part of the charging pile fault diagnosis method. In the previous stage, we have obtained the fault propagation path by analyzing the potential fault component association map and clarified the mutual influence relationship between the potential fault components. Next, using the Bayesian network model, the probability of failure of these components can be further quantified, thereby providing a more accurate basis for determining the source of the fault. Bayesian network is a probabilistic graphical model that can represent the conditional dependency between variables and is very suitable for dealing with problems with uncertainty. In the application scenario of charging pile fault diagnosis, we can regard each potential fault component as a node in the Bayesian network, and the edges between the nodes represent the causal relationship or influence relationship between the components. For example, suppose that in the previous analysis, we have identified three potential fault components, such as overheating of the charging interface, reduced efficiency of the cooling system, and increased temperature of the main control board, and have clarified the fault propagation path between them. Then, in the Bayesian network, overheating of the charging interface can be used as a starting node, connected to reduced efficiency of the cooling system through an edge, and the latter is connected to increased temperature of the main control board. After the Bayesian network model is built, it is necessary to estimate the parameters of the model, that is, to determine the conditional probability distribution of each node given the state of its parent node. These parameters can be learned from historical data or manually set based on the knowledge of domain experts. For example, by analyzing historical fault records, it can be found that the probability of cooling system efficiency reduction when the charging interface is overheated; similarly, the probability of the main control board temperature increase when the cooling system efficiency decreases can also be calculated. These conditional probability distributions constitute the core of the Bayesian network, enabling the model to infer the probability of unknown events based on known information. In practical applications, when an abnormality is detected in the charging pile, the fault feature data currently collected can be input into the Bayesian network model, and the model will automatically calculate the probability of failure of each potential fault component. For example, if the temperature of the charging interface is currently detected to be abnormally high, this information can be input into the model, and the model will calculate the probability of cooling system efficiency reduction and main control board temperature increase based on the known fault propagation path and conditional probability distribution. By comparing these probability values, the most likely source of the fault can be determined, thereby guiding maintenance personnel to conduct targeted inspections and repairs. In summary, by using the preset Bayesian network model and calculating the failure probability of the potential faulty components based on the fault propagation path, not only can a scientific method be provided to evaluate the failure risk of each component, but also a strong support can be provided for the precise positioning of the source of the fault, thereby improving the accuracy and efficiency of charging pile fault diagnosis, ensuring the stable operation of charging facilities, and improving user experience.

[0077] Step S6, taking the potential fault component corresponding to the greatest fault probability as the fault source, analyzing the fault type of the fault source through a preset classification algorithm, and formulating a maintenance strategy for the target charging pile based on the fault type.

[0078] Specifically, the potential fault component corresponding to the largest fault probability is taken as the fault source, and the fault type of the fault source is analyzed by a preset classification algorithm, and the maintenance strategy of the target charging pile is formulated based on the fault type. This process is a key step in the charging pile fault diagnosis method. By combining fault probability analysis and classification algorithm, the accuracy and effectiveness of the maintenance strategy are ensured. First, the potential fault component corresponding to the largest fault probability is taken as the fault source. In the previous steps, we have calculated the fault probability of each potential fault component through the Bayesian network model. Assuming that the fault probability of overheating of the charging interface is the highest, reaching 70%, then the overheating of the charging interface can be determined as the fault source. This step ensures the accuracy of fault diagnosis, because the component with the highest fault probability is most likely to be the starting point of the entire fault chain. Next, the fault type of the fault source is analyzed by a preset classification algorithm. The purpose of the classification algorithm is to map the multimodal features of the fault source to a specific fault type. For example, classifiers such as support vector machines (SVM), random forests, or neural networks can be used to analyze the multimodal fault features of the charging interface. These features may include current fluctuations, temperature changes, vibration frequencies, etc. By training the classifier, the fault type of the charging interface can be classified into poor contact, aging of internal wires, loose plugs, etc. For example, assuming that the fault type of the charging interface is determined to be poor contact through classification algorithm analysis, this step provides a clear direction for the subsequent maintenance strategy. Formulate a maintenance strategy for the target charging pile based on the fault type. After determining the source of the fault and its specific fault type, a detailed maintenance plan needs to be formulated. For example, if the fault type is poor contact, the preliminary maintenance plan may include measures such as tightening the plug, cleaning the contact surface, and checking the wear of the jack. When formulating a maintenance strategy, factors such as the severity of the fault, the difficulty of maintenance, and the possible maintenance cost need to be considered. For example, if the cause of the poor contact is a loose plug, then a simple measure of tightening the plug can be taken; but if the cause is aging of the internal wire, the entire charging interface may need to be replaced, or even the entire charging system may need to be overhauled. Through this series of steps, not only can the source of the fault and its specific fault type be accurately identified, but also an effective maintenance strategy can be formulated to ensure the rapid recovery and stable operation of the charging pile. For example, in the actual application scenario of an electric vehicle charging station, if the fault type of the charging interface is determined to be poor contact through the above method, maintenance personnel can quickly take measures to tighten the plug and clean the contact surface, thereby quickly restoring the normal operation of the charging pile, reducing downtime, and ensuring the continuity and stability of charging services. This process not only improves the accuracy and efficiency of fault diagnosis, but also provides a scientific basis for the maintenance and management of charging piles, and improves user experience.

[0079] In a specific embodiment, the acquiring of the fault status data of each component of the target charging pile and the feature extraction of the fault status data to obtain the fault feature data include:

[0080] The fault parameters of each component in the target charging pile are collected by using multiple sensors arranged inside the target charging pile to obtain original fault state parameters;

[0081] Performing parameter clustering on the original fault state parameters to obtain clustered fault parameters; wherein the clustered fault parameters include voltage parameters, current parameters and temperature parameters;

[0082] Performing time-frequency analysis on the clustered fault parameters to obtain fault time-frequency characteristic data;

[0083] Performing dimension reduction processing on the fault time-frequency characteristic data to obtain dimension-reduced fault parameters;

[0084] Feature extraction is performed on the dimension-reduced fault parameters to obtain fault feature data.

[0085] Specifically, the acquisition of the fault state data of each component of the target charging pile and the feature extraction of the fault state data to obtain the fault feature data include collecting fault parameters of each component in the target charging pile through multiple sensors arranged inside the target charging pile to obtain the original fault state parameters; clustering the original fault state parameters to obtain clustered fault parameters; wherein the clustered fault parameters include voltage parameters, current parameters and temperature parameters; performing time-frequency analysis on the clustered fault parameters to obtain fault time-frequency feature data; performing dimension reduction processing on the fault time-frequency feature data to obtain dimension reduction fault parameters; and feature extraction on the dimension reduction fault parameters to obtain fault feature data. This process is the most basic and most critical part of the charging pile fault diagnosis method, which directly affects the accuracy of subsequent fault identification and maintenance strategy formulation. First, the fault parameters of each component in the target charging pile are collected through multiple sensors arranged inside the target charging pile to obtain the original fault state parameters. In actual application scenarios, for example, in electric vehicle charging stations, various types of sensors such as current sensors, voltage sensors, temperature sensors, etc. are installed inside the charging piles, which are responsible for real-time monitoring of the working status of various components of the charging piles. The current sensor can detect the current change in the charging circuit, the voltage sensor records the voltage level of the power supply line, and the temperature sensor monitors the temperature of the charging interface and internal electronic components. The data collected by these sensors constitute the basis of the original fault state parameters. For example, when an electric vehicle is connected to a charging pile and starts charging, the current sensor will continuously monitor the current intensity in the charging circuit, the voltage sensor will record the change of the power supply voltage during the charging process, and the temperature sensor is used to detect the temperature of the charging interface and internal electronic components to ensure that they are within the normal working range. Next, the original fault state parameters are clustered to obtain clustered fault parameters. This process aims to classify a large amount of raw data according to certain rules for subsequent analysis and processing. Clustered fault parameters mainly include voltage parameters, current parameters and temperature parameters. Clustering methods can use K-means algorithms, hierarchical clustering, etc., through which similar parameter values ​​are classified into one category. For example, current parameters can be divided into currents within the normal range and abnormally high or low currents, voltage parameters can be divided into stable voltages and voltages with large fluctuations, and temperature parameters can be divided into normal working temperatures and abnormally high temperatures. Through clustering, the original data can be simplified into several representative parameters, laying the foundation for subsequent feature extraction. Subsequently, the clustered fault parameters are subjected to time-frequency analysis to obtain fault time-frequency characteristic data. Time-frequency analysis is an important signal processing technology that can simultaneously display the characteristics of a signal in two dimensions, time and frequency. In charging pile fault diagnosis, the timing information and frequency components of the fault can be extracted from the clustered fault parameters through time-frequency analysis.For example, for current parameters, time-frequency analysis can reveal the periodic characteristics of current fluctuations. If a certain frequency component is found to be significantly enhanced, it may mean that there is some periodic interference in the charging circuit. For temperature parameters, time-frequency analysis can help identify the law of temperature changes. If the temperature is found to rise rapidly in a short period of time, it may indicate that a certain component has overheated. Through time-frequency analysis, clustered fault parameters can be converted into fault time-frequency feature data with more diagnostic significance. Next, the fault time-frequency feature data is subjected to dimensionality reduction processing to obtain dimensionality-reduced fault parameters. The purpose of dimensionality reduction processing is to reduce the dimension of the data and retain the most important feature information, thereby improving the efficiency and accuracy of subsequent analysis. Common dimensionality reduction methods include principal component analysis (PCA), linear discriminant analysis (LDA), etc. Through dimensionality reduction, high-dimensional fault time-frequency feature data can be converted into low-dimensional dimensionality-reduced fault parameters. For example, through PCA analysis, the multidimensional feature space of current parameters, voltage parameters and temperature parameters can be projected into a two-dimensional or three-dimensional space, retaining several principal components that best reflect the fault characteristics. This not only reduces the complexity of the data, but also facilitates visualization and further analysis. Finally, feature extraction is performed on the reduced-dimensionality fault parameters to obtain fault feature data. Feature extraction refers to selecting the most representative and discriminative features from the reduced-dimensionality fault parameters for subsequent fault identification. Feature extraction methods may include statistical analysis, pattern recognition, and the like. For example, statistical features such as average value, variance, and peak value may be extracted from the reduced-dimensionality current parameters, features such as fluctuation amplitude and frequency component may be extracted from the voltage parameters, and features such as maximum value and rate of change may be extracted from the temperature parameters. These feature quantities can effectively reflect the working status of each component of the charging pile and provide key information for subsequent fault identification and analysis. Through this series of steps, not only can the most valuable fault feature data be extracted from a large number of original fault state parameters, but also a solid data foundation can be provided for charging pile fault diagnosis. Ultimately, these fault feature data will be used for subsequent fault identification, fault propagation path analysis, and fault source location, helping technicians to quickly and accurately diagnose and repair charging pile faults, ensure the stable operation of charging facilities, and improve user experience.

[0086] In a specific embodiment, the fault identification of the components of the target charging pile based on the fault characteristic data to obtain potential fault components includes:

[0087] Acquire historical charging pile fault feature data, perform tensor decomposition on the historical charging pile fault feature data to obtain a three-dimensional tensor feature matrix, and perform dimensionality reduction processing on the three-dimensional tensor feature matrix to obtain a core tensor and a factor matrix;

[0088] Performing non-negative matrix decomposition on the core tensor and factor matrix to obtain a historical fault feature subspace; wherein the historical fault feature subspace includes typical fault mode information of various components of the historical charging pile;

[0089] Performing fault identification learning on the historical fault feature subspace through a preset initial fault identification algorithm to obtain a fault identification algorithm;

[0090] Performing time-frequency fault identification on the fault feature data based on the fault identification algorithm to obtain time-frequency fault feature data;

[0091] Based on the time-frequency fault characteristic data, fault identification is performed on components of the target charging pile to obtain potential fault components.

[0092] Specifically, the fault identification of the components of the target charging pile based on the fault feature data to obtain potential fault components includes obtaining historical charging pile fault feature data, performing tensor decomposition on the historical charging pile fault feature data to obtain a three-dimensional tensor feature matrix, and performing dimensionality reduction processing on the three-dimensional tensor feature matrix to obtain a core tensor and a factor matrix; performing non-negative matrix decomposition on the core tensor and factor matrix to obtain a historical fault feature subspace; wherein the historical fault feature subspace includes typical fault mode information of each component of the historical charging pile; performing fault identification learning on the historical fault feature subspace through a preset initial fault identification algorithm to obtain a fault identification algorithm; performing time-frequency fault identification on the fault feature data based on the fault identification algorithm to obtain time-frequency fault feature data; performing fault identification on the components of the target charging pile based on the time-frequency fault feature data to obtain potential fault components. This process is the core link in the charging pile fault diagnosis method. By combining historical data and modern data analysis technology, efficient identification of charging pile faults is achieved. First, historical charging pile fault feature data is obtained. This step requires extracting previous fault data from the operation and maintenance records of the charging pile, including information such as the time, location, fault component, and fault type of the fault. For example, in the actual application scenario of an electric vehicle charging station, the operation and maintenance personnel will regularly record the operating status and fault conditions of each charging pile, and these data will be saved in a database. By accessing these historical data, a large amount of fault feature data can be obtained, which will be used for subsequent fault mode analysis and training of fault identification models. Next, the historical charging pile fault feature data is tensor decomposed to obtain a three-dimensional tensor feature matrix. Tensor decomposition is an advanced data analysis technique that can decompose multidimensional data into multiple low-rank components. In charging pile fault diagnosis, historical fault feature data can be regarded as a three-dimensional tensor, in which the three dimensions represent time, component, and feature type. For example, the time dimension can represent fault records in different time periods, the component dimension represents different components of the charging pile (such as charging interface, main control board, cooling system, etc.), and the feature type dimension represents different fault features (such as current, voltage, temperature, etc.). Through tensor decomposition, this complex three-dimensional tensor can be decomposed into a core tensor and several factor matrices, which correspond to low-rank representations of time, components, and feature types, respectively. Then, the three-dimensional tensor feature matrix is ​​subjected to dimensionality reduction processing to obtain a core tensor and a factor matrix. The purpose of dimensionality reduction processing is to reduce the dimension of the data while retaining the most important feature information. In the process of tensor decomposition, by selecting a suitable rank parameter, the high-dimensional three-dimensional tensor feature matrix can be reduced to a core tensor and a factor matrix. The core tensor retains the main structural information of the historical fault data, while the factor matrix reflects the low-rank representation of time, components, and feature types.For example, by selecting an appropriate rank parameter, a large three-dimensional tensor feature matrix can be reduced to a smaller core tensor and several low-rank factor matrices, thereby greatly reducing the complexity of the data and improving the efficiency of subsequent analysis. Then, the core tensor and factor matrix are subjected to non-negative matrix decomposition to obtain a historical fault feature subspace. Non-negative matrix decomposition (NMF) is a commonly used dimensionality reduction and feature extraction technology, particularly suitable for processing non-negative data. Through NMF, the core tensor and factor matrix can be decomposed into several non-negative basis vectors and weight matrices, which together constitute the historical fault feature subspace. The historical fault feature subspace contains typical fault mode information of each component of the charging pile, which can be used for subsequent fault identification and pattern matching. For example, through NMF decomposition, typical fault modes such as overheating of the charging interface, abnormal temperature of the main control board, and reduced efficiency of the cooling system can be extracted, and these modes will serve as the basis for fault identification. Subsequently, the fault identification learning is performed on the historical fault feature subspace through the preset initial fault identification algorithm to obtain a fault identification algorithm. This process usually involves machine learning or deep learning technology, which trains a classifier or regression model to learn the characteristics of the fault mode from the historical fault feature subspace. For example, algorithms such as support vector machine (SVM), random forest (RandomForest) or neural network can be selected as the initial fault recognition algorithm. By inputting the data in the historical fault feature subspace into these algorithms for training, a fault recognition algorithm that can accurately identify the fault mode can be obtained. During the training process, the historical fault data needs to be divided into a training set and a test set, and the model parameters are continuously adjusted to achieve the best performance on the test set. Finally, the fault feature data is subjected to time-frequency fault recognition based on the fault recognition algorithm to obtain time-frequency fault feature data. In practical applications, when an abnormality is detected in the charging pile, the current fault feature data can be collected through sensors, which include parameters such as current, voltage, and temperature. These fault feature data are input into the trained fault recognition algorithm, and the algorithm automatically performs time-frequency analysis to extract time-frequency fault feature data. For example, through time-frequency analysis, the periodic characteristics of current fluctuations, transient changes in voltage, and rapid increases in temperature can be identified, and these characteristic data will be used for subsequent fault identification. Based on the time-frequency fault characteristic data, fault identification is performed on the components of the target charging pile to obtain potential fault components. By inputting the time-frequency fault characteristic data into the fault identification algorithm, the algorithm will match and classify the current fault characteristic data according to the typical fault modes in the historical fault characteristic subspace to determine which components are most likely to fail. For example, if the time-frequency fault characteristic data shows that the temperature of the charging interface increases abnormally and is accompanied by current fluctuations, the fault identification algorithm will mark the charging interface as a potential fault component.In this way, the source of the fault can be quickly and accurately identified, providing clear guidance to maintenance personnel, improving maintenance efficiency, and reducing the downtime of the charging pile. Through this series of steps, not only can typical fault modes be extracted from a large amount of historical fault data, but these modes can also be used to efficiently identify and classify the current fault feature data. Ultimately, this method not only improves the accuracy and efficiency of charging pile fault diagnosis, but also provides a scientific basis for the maintenance and management of charging piles, ensuring the stable operation of charging facilities and improving user experience.

[0093] In a specific embodiment, constructing a graph of the potential faulty components based on the component graph to obtain a potential faulty component association graph includes:

[0094] Performing structural analysis on the component atlas of the target charging pile to obtain structural relationship analysis results between the components;

[0095] Constructing a component topology relationship matrix based on the structural relationship analysis results;

[0096] Based on the component topology relationship matrix, the interaction mode between the components of the target charging pile is captured to obtain the interaction mode between the components; wherein the interaction mode between the components represents the functional relationship and connection mode between the components;

[0097] Modeling the influence relationship of the potential fault components based on the interaction mode between the components to obtain the fault relationship of the potential fault components;

[0098] By using a preset graph attention network, based on the fault relationship of the potential fault component, the potential fault component is weighted in the component graph of the target charging pile to obtain a fault component weight coefficient;

[0099] The corresponding potential faulty components are connected based on the faulty component weight coefficients to obtain a potential faulty component association map.

[0100] Specifically, the construction of the graph of the potential faulty component based on the component graph to obtain the potential faulty component association graph includes structural analysis of the component graph of the target charging pile to obtain the structural relationship analysis results between the components; constructing a component topological relationship matrix based on the structural relationship analysis results; capturing the interaction mode between the components of the target charging pile based on the component topological relationship matrix to obtain the interaction mode between the components; wherein the interaction mode between the components is the action relationship and connection mode between the components; modeling the influence relationship of the potential faulty component based on the interaction mode between the components to obtain the fault relationship of the potential faulty component; through a preset graph attention network, based on the fault relationship of the potential faulty component, the attention of the potential faulty component is weighted in the component graph of the target charging pile to obtain the faulty component weight coefficient; connecting the corresponding potential faulty components based on the faulty component weight coefficient to obtain the potential faulty component association graph. This process is an important step in the charging pile fault diagnosis method. By combining graph theory and deep learning technology, the modeling and analysis of the complex relationship between potential faulty components is realized. First, the component map of the target charging pile is structurally analyzed to obtain the structural relationship analysis results between the components. In the actual application scenario of the electric vehicle charging station, the component map of the charging pile records in detail the physical location, electrical connection and functional relationship of each component. For example, the component map of the charging pile may include key components such as the power input port, the main control board, the charging interface, the cooling system, and the connection method between them. By performing structural analysis on these component maps, the structural relationship between the components can be extracted, such as which components are directly connected and which components are indirectly connected through intermediate components. These structural relationships provide a basis for the subsequent topological relationship matrix construction. Next, a component topological relationship matrix is ​​constructed based on the structural relationship analysis results. The topological relationship matrix is ​​a mathematical tool used to represent the connection relationship between various components in the system. In the fault diagnosis of the charging pile, a matrix can be constructed to represent the connection between the various components. For example, assuming that the charging pile has 5 main components, a 5x5 matrix can be constructed, where the (i, j)th element of the matrix represents the connection relationship between component i and component j. If component i and component j are directly connected, the element value is 1, otherwise it is 0. In this way, the complex component connection relationship can be simplified into a matrix form, which is convenient for subsequent analysis and processing. Then, based on the component topology relationship matrix, the interaction mode between the components of the target charging pile is captured to obtain the interaction mode between the components. The purpose of interactive mode capture is to reveal the dynamic interaction relationship between components, not just the static connection relationship. For example, by analyzing sensor data such as current, voltage, and temperature, it can be found that an increase in the temperature of the charging interface may cause an increase in the load on the cooling system, which in turn affects the temperature of the main control board.These dynamic interaction patterns can be captured by methods such as timing analysis and correlation analysis. The captured interaction patterns include not only direct physical connections, but also causal relationships and synergies between components. Based on the interaction patterns between the components, the potential fault components are modeled for influence relationships to obtain the fault relationships of the potential fault components. The purpose of influence relationship modeling is to describe the interaction between potential fault components through a mathematical model. For example, a directed graph can be used to represent the causal relationship between components, where nodes represent components and edges represent the direction and intensity of fault propagation. By analyzing the interaction patterns between components, a fault propagation graph can be constructed, in which each edge has a weight, indicating the probability or intensity of fault propagation from one component to another. This fault relationship model provides a basis for subsequent fault propagation path analysis. Through the preset graph attention network, based on the fault relationship of the potential fault component, the potential fault component is weighted in the component graph of the target charging pile to obtain the weight coefficient of the fault component. Graph Attention Network (GAT) is a graph neural network model based on deep learning that can automatically learn the weight relationship between nodes in a graph. In the fault diagnosis of charging piles, the fault relationship graph of potential fault components can be input into GAT, and the attention weights between components can be learned by training the model. For example, if the fault of the charging interface has a greater impact on the cooling system, GAT will assign a higher weight to this edge. In this way, the weight coefficients of each potential fault component can be obtained, which reflect the importance of each component in fault propagation. Finally, the corresponding potential fault components are connected based on the weight coefficients of the fault components to obtain the potential fault component association map. The potential fault component association map is a visualization tool for displaying the connection relationship and influence relationship between potential fault components. By mapping the weight coefficients of the fault components to the edge weights in the graph, a weighted fault propagation graph can be generated. In this graph, edges with higher weights indicate a greater possibility of fault propagation, and edges with lower weights indicate a lower possibility of fault propagation. By analyzing this association map, you can intuitively see which components are the key nodes of fault propagation, thereby providing a scientific basis for locating the source of the fault and formulating maintenance strategies. Through this series of steps, not only can the structural relationship and interaction mode between the components be extracted from the component map of the charging pile, but the importance of each component in fault propagation can also be automatically learned through the graph attention network, and finally a potential fault component association map is generated. This process not only improves the accuracy and efficiency of fault diagnosis, but also provides a scientific basis for the maintenance and management of charging piles, ensures the stable operation of charging facilities, and improves user experience.

[0101] In a specific embodiment, the analyzing the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path includes:

[0102] Based on the potential fault component association graph, the node centrality of each potential fault component is calculated to obtain the centrality index of each potential fault component; wherein the node centrality includes degree centrality, closeness centrality and betweenness centrality;

[0103] Analyze the influence relationship between potential fault components based on the centrality index;

[0104] Based on the influence relationship between the potential fault components, the potential fault components in the fault component association map are divided into communities to identify potential fault components with similar fault propagation characteristics, and obtain a dynamic community structure of potential fault components;

[0105] A causal relationship analysis is performed on potential fault components in the dynamic community structure through a preset causal reasoning model to obtain a fault propagation causal path, and the fault propagation causal path is used as the fault propagation path.

[0106] Specifically, the influence relationship between each potential fault component is analyzed based on the potential fault component association map to obtain the fault propagation path, including calculating the node centrality of each potential fault component based on the potential fault component association map to obtain the centrality index of each potential fault component; wherein the node centrality includes degree centrality, closeness centrality and betweenness centrality; analyzing the influence relationship between each potential fault component based on the centrality index; dividing the potential fault components in the fault component association map into communities based on the influence relationship between each potential fault component to identify potential fault components with similar fault propagation characteristics and obtain the dynamic community structure of the potential fault component; performing causal relationship analysis on the potential fault components in the dynamic community structure through a preset causal reasoning model to obtain the fault propagation causal path, and using the fault propagation causal path as the fault propagation path. This process is a key step in the charging pile fault diagnosis method. By combining graph theory and causal reasoning technology, the modeling and analysis of the complex influence relationship between potential fault components is realized. First, the node centrality of each potential fault component is calculated based on the potential fault component association map to obtain the centrality index of each potential fault component. Node centrality is an indicator used in graph theory to measure the importance of a node in a network. Common node centralities include degree centrality, closeness centrality, and betweenness centrality. Degree centrality indicates the number of direct connections of a node, closeness centrality indicates the inverse of the shortest path length from a node to all other nodes, and betweenness centrality indicates the frequency with which a node transmits information between other nodes. In charging pile fault diagnosis, the importance of each potential fault component in the network can be evaluated by calculating these centrality indicators. For example, suppose we have built a potential fault component association graph, in which components such as the charging interface, cooling system, and main control board are nodes, and the connection relationship between them is an edge. By calculating the degree centrality, closeness centrality, and betweenness centrality of these components, the importance index of each component in the network can be obtained. For example, the degree centrality of the charging interface is high, indicating that it is directly connected to multiple other components and may be a key node for fault propagation. Next, the influence relationship between each potential fault component is analyzed based on the centrality indicators. By analyzing the centrality indicators, it is possible to preliminarily determine which components have a high influence in the network. For example, if a component has a high betweenness centrality, it means that it plays a bridging role in fault propagation, and the fault is likely to propagate to other components through it. Based on these centrality indicators, the mutual influence relationship between potential fault components can be further analyzed. For example, if the temperature increase of the charging interface causes the load of the cooling system to increase, which in turn affects the temperature of the main control board, then it can be considered that the charging interface has a strong influence on the cooling system and the main control board. By comprehensively considering the centrality indicators and the influence relationship, the influence of each potential fault component can be ranked to determine which components play a key role in fault propagation.Then, based on the influence relationship between the potential fault components, the potential fault components in the fault component association map are divided into communities to identify potential fault components with similar fault propagation characteristics, and obtain the dynamic community structure of the potential fault components. Community division is a technique in graph theory, which is used to divide nodes in a network into several closely connected subgroups. In charging pile fault diagnosis, potential fault components with similar fault propagation characteristics can be identified through community division. For example, the Louvain method or the Girvan-Newman algorithm can be used to divide the potential fault component association map into communities. Assume that through community division, we find that the charging interface, cooling system and main control board belong to the same community, which indicates that these three components have similar characteristics in fault propagation and may be affected by certain factors in common. By identifying these dynamic community structures, the propagation mode of faults between different components can be better understood. Finally, the potential fault components in the dynamic community structure are analyzed for causal relationships through a preset causal reasoning model to obtain a fault propagation causal path, and the fault propagation causal path is used as the fault propagation path. A causal reasoning model is a statistical method for analyzing causal relationships between variables, such as Bayesian networks, structural equation models, etc. In the fault diagnosis of charging piles, causal reasoning models can be used to analyze the causal relationship between potential fault components in the dynamic community structure. For example, assuming that we have determined that the charging interface, cooling system, and main control board belong to the same community, the causal relationship between these components can be analyzed through the causal reasoning model. If the model finds that the temperature increase of the charging interface causes the load of the cooling system to increase, which in turn affects the temperature of the main control board, then this causal path can be used as the fault propagation path. In this way, the specific path of the fault propagation from one component to another can be more accurately identified, providing a scientific basis for locating the source of the fault and formulating maintenance strategies. Through this series of steps, not only can the complex influence relationship between the components be extracted from the association map of potential fault components, but also the key path of fault propagation can be identified through community division and causal reasoning models. This process not only improves the accuracy and efficiency of fault diagnosis, but also provides a scientific basis for the maintenance and management of charging piles, ensures the stable operation of charging facilities, and improves user experience. For example, in the actual application scenario of electric vehicle charging stations, this method can quickly and accurately identify the fault propagation path caused by overheating of the charging interface, thereby guiding maintenance personnel to conduct targeted inspections and repairs, reducing the downtime of charging piles, and ensuring the continuity and stability of charging services.

[0107] In a specific embodiment, the calculating the failure probability of the potential fault component based on the fault propagation path by using a preset Bayesian network model includes:

[0108] Structural modeling of the fault propagation path is performed using a preset Bayesian network model to obtain a Bayesian network structure of the fault propagation;

[0109] Based on the Bayesian network structure, fault parameter estimation is performed on each potential fault component in the fault propagation path to obtain a conditional probability distribution of each potential fault component;

[0110] Performing a causal effect analysis on the potential fault components based on the conditional probability distribution to obtain a causal effect analysis result; wherein the causal effect analysis result is a causal influence result of each potential fault component;

[0111] Based on the causal effect analysis result, the fault expansion of the potential fault component in the time dimension is performed to obtain a dynamic fault propagation probability;

[0112] A robustness calculation is performed on the potential fault component based on the dynamic fault propagation probability to obtain the failure probability of the potential fault component.

[0113] Specifically, the fault probability of the potential fault component is calculated based on the fault propagation path through a preset Bayesian network model, including structural modeling of the fault propagation path through a preset Bayesian network model to obtain a Bayesian network structure of fault propagation; based on the Bayesian network structure, the fault parameters of each potential fault component in the fault propagation path are estimated to obtain the conditional probability distribution of each potential fault component; based on the conditional probability distribution, the potential fault component is subjected to causal effect analysis to obtain a causal effect analysis result; wherein the causal effect analysis result is the causal influence result of each potential fault component; based on the causal effect analysis result, the potential fault component is subjected to fault extension in the time dimension to obtain a dynamic fault propagation probability; based on the dynamic fault propagation probability, the potential fault component is subjected to robustness calculation to obtain the fault probability of the potential fault component. This process is an important part of the charging pile fault diagnosis method. By combining the Bayesian network model and causal reasoning technology, the accurate calculation of the fault probability of the potential fault component is realized. First, the fault propagation path is structurally modeled through a preset Bayesian network model to obtain a Bayesian network structure of fault propagation. Bayesian network is a probabilistic graphical model that can represent the conditional dependency between variables and is very suitable for handling fault diagnosis problems with uncertainty. In the fault diagnosis of charging piles, each potential fault component can be regarded as a node in the Bayesian network, and the edges between the nodes represent the causal relationship or influence relationship between the components. For example, suppose that we have identified potential fault components such as overheating of the charging interface, reduced efficiency of the cooling system, and increased temperature of the main control board through the previous steps, and have clarified the fault propagation path between them. Then, in the Bayesian network, overheating of the charging interface can be used as a starting node, connected to reduced efficiency of the cooling system through an edge, which in turn is connected to increased temperature of the main control board. In this way, a complete Bayesian network structure of the fault propagation path can be constructed. Next, based on the Bayesian network structure, the fault parameters of each potential fault component in the fault propagation path are estimated to obtain the conditional probability distribution of each potential fault component. The purpose of fault parameter estimation is to determine the conditional probability distribution of each node given the state of its parent node. These parameters can be learned from historical data or manually set based on the knowledge of domain experts. For example, by analyzing historical fault records, we can find out the probability of cooling system efficiency degradation when the charging interface is overheated; similarly, we can also calculate the probability of the main control board temperature rising when the cooling system efficiency decreases. These conditional probability distributions form the core of the Bayesian network, enabling the model to infer the probability of unknown events based on known information. For example, if an abnormal increase in the charging interface temperature is currently detected, this information can be input into the Bayesian network model, and the model will calculate the probability of cooling system efficiency degradation and main control board temperature increase based on the known fault propagation path and conditional probability distribution.Then, based on the conditional probability distribution, a causal effect analysis is performed on the potential fault component to obtain a causal effect analysis result. The purpose of the causal effect analysis is to evaluate the causal influence between the potential fault components through the Bayesian network model. For example, assuming that the probability of overheating of the charging interface is 70%, the probability of reduced cooling system efficiency is 60%, and the probability of increased temperature of the main control board is 50%. Through the causal effect analysis, the degree of influence of overheating of the charging interface on the reduced cooling system efficiency and the degree of influence of reduced cooling system efficiency on the increased temperature of the main control board can be further determined. These causal influence results not only reflect the dependency between the components, but also provide a basis for subsequent fault extension and robustness calculations. For example, if the analysis results show that the impact of overheating of the charging interface on the reduced cooling system efficiency is very significant, then this relationship can be taken as a key factor to consider. Based on the causal effect analysis results, the potential fault component is extended in the time dimension to obtain a dynamic fault propagation probability. The purpose of fault extension is to consider the evolution of the fault in the time dimension and evaluate the possibility of fault propagation from one component to another. For example, assuming that the charging interface is currently detected to be overheated, the Bayesian network model can be used to calculate the probability of cooling system efficiency degradation and the probability of main control board temperature increase in the future. Through dynamic fault expansion, a time-varying fault propagation probability curve can be obtained. For example, as time goes by, the probability of cooling system efficiency degradation gradually increases, which may eventually lead to an increase in the temperature of the main control board. In this way, the dynamic process of fault propagation can be more comprehensively understood, providing a basis for fault prevention and emergency handling. Finally, based on the dynamic fault propagation probability, robustness calculations are performed on potential fault components to obtain the failure probability of potential fault components. The purpose of robustness calculations is to evaluate the failure probability of each potential fault component under different conditions to ensure the reliability and accuracy of the calculation results. For example, the Monte Carlo simulation method can be used to run the Bayesian network model multiple times, each time inputting different initial conditions and parameters to calculate the failure probability of each potential fault component. Through multiple simulations, a probability distribution can be obtained to evaluate the confidence interval of the failure probability of each component. For example, suppose that after 100 simulations, the probability distribution of charging port overheating is between 65% and 75%, the probability distribution of cooling system efficiency degradation is between 55% and 65%, and the probability distribution of main control board temperature increase is between 45% and 55%. These probability distributions not only reflect the failure probability of each component, but also provide confidence intervals, providing more reference information for decision-making. Through this series of steps, not only can the conditional probability distribution of each potential fault component be extracted from the Bayesian network model, but also the propagation process of the fault in the time dimension can be evaluated through causal effect analysis and dynamic fault expansion, and finally the failure probability of each potential fault component can be obtained.This process not only improves the accuracy and reliability of fault diagnosis, but also provides a scientific basis for the maintenance and management of charging piles, ensures the stable operation of charging facilities, and improves user experience. For example, in the actual application scenario of electric vehicle charging stations, this method can quickly and accurately calculate the probability of fault propagation caused by overheating of the charging interface, thereby guiding maintenance personnel to conduct targeted inspections and repairs, reduce the downtime of charging piles, and ensure the continuity and stability of charging services.

[0114] In a specific embodiment, analyzing the fault type of the fault source by a preset classification algorithm and formulating a maintenance strategy for the target charging pile based on the fault type includes:

[0115] Performing multi-dimensional feature extraction on the fault source corresponding to the maximum fault probability to obtain a multi-modal fault feature;

[0116] Classifying the multi-modal fault features into fault types using a preset classification algorithm to obtain a preliminary fault type;

[0117] Performing knowledge reasoning on the preliminary fault type to obtain a fault type report;

[0118] Preliminarily designing a maintenance strategy for the target charging pile based on the fault type report to obtain a preliminary maintenance plan;

[0119] Performing simulation verification on the fault source based on the preliminary maintenance plan to obtain a simulation verification result;

[0120] The simulation verification result is evaluated to obtain a simulation verification evaluation value. If the simulation verification evaluation value is less than a preset simulation verification evaluation value, the preliminary maintenance plan is optimized and adjusted based on the simulation verification result to obtain a maintenance strategy for the target charging pile.

[0121] Specifically, first, multi-dimensional feature extraction is performed on the fault source corresponding to the maximum fault probability to obtain multi-modal fault features. In the previous steps, we have calculated the fault probability of each potential fault component through the Bayesian network model, and determined the component with the highest fault probability as the fault source. Next, it is necessary to perform detailed multi-dimensional feature extraction on this fault source. Multi-modal fault features include but are not limited to current features, voltage features, temperature features, vibration features, etc. For example, assuming that the overheating of the charging interface is determined as the fault source through the Bayesian network model, the current fluctuation, temperature change curve, vibration frequency and other features of the charging interface can be extracted. These multi-modal fault features can reflect the state of the fault source from multiple angles and provide rich information for subsequent fault type classification. Next, the multi-modal fault features are classified into fault types through a preset classification algorithm to obtain a preliminary fault type. The purpose of the classification algorithm is to map multi-modal fault features to specific fault types through machine learning or pattern recognition technology. For example, classifiers such as support vector machines (SVM), random forests (Random Forest) or neural networks can be used to train and classify the extracted multi-modal fault features. Assume that we have trained a classifier that can classify the multimodal fault features of the charging interface into several common fault types such as poor contact, aging of internal wires, and loose plugs. Through classification, the specific fault type of the fault source can be preliminarily determined. For example, the classifier may output that the fault type of the charging interface is poor contact. Then, the preliminary fault type is subjected to knowledge reasoning to obtain a fault type report. The purpose of knowledge reasoning is to further verify and refine the preliminary fault type by combining domain knowledge and expert experience. For example, assuming that the preliminary classification result shows that the fault type of the charging interface is poor contact, the cause of the poor contact can be analyzed through knowledge reasoning, such as loose plugs, worn jacks, and oxidized contact surfaces. Knowledge reasoning can be implemented through technologies such as rule engines, expert systems, or knowledge graphs. Through knowledge reasoning, a detailed fault type report can be generated, which includes not only the preliminary fault type, but also the specific cause and possible impact of the fault. For example, the report may point out that the cause of the poor contact is a loose plug, which may lead to reduced charging efficiency and safety hazards. Based on the fault type report, the maintenance strategy of the target charging pile is preliminarily designed to obtain a preliminary maintenance plan. The purpose of the preliminary design is to develop a preliminary repair plan based on the information in the fault type report. For example, if the fault type report indicates that the fault type of the charging interface is poor contact and the cause is a loose plug, the preliminary repair plan may include measures such as tightening the plug, cleaning the contact surface, and checking the wear of the socket. The preliminary repair plan needs to take into account factors such as the severity of the fault, the difficulty of repair, and the possible repair cost to ensure the feasibility and economy of the plan.For example, the preliminary maintenance plan may include the following steps: 1. Disconnect the power supply to ensure safety; 2. Check the wear of the plug and the socket; 3. Clean the contact surface and remove the oxide; 4. Tighten the plug to ensure good contact. Next, the source of the fault is simulated and verified based on the preliminary maintenance plan to obtain a simulation verification result. The purpose of simulation verification is to evaluate the effect of the preliminary maintenance plan in actual application through simulation technology. For example, simulation software can be used to simulate the operating state of the charging interface after tightening the plug and cleaning the contact surface, and observe the changes in parameters such as current, voltage and temperature. Through simulation verification, potential problems in the preliminary maintenance plan can be discovered in advance to ensure the effectiveness of the plan. Assume that the simulation results show that the temperature of the charging interface drops significantly after tightening the plug, and the current and voltage also return to the normal range, indicating that the preliminary maintenance plan is effective. Finally, the simulation verification results are evaluated to obtain a simulation verification evaluation value. If the simulation verification evaluation value is less than the preset simulation verification evaluation value, the preliminary maintenance plan is optimized and adjusted based on the simulation verification results to obtain the maintenance strategy of the target charging pile. The purpose of the evaluation is to quantitatively evaluate the simulation verification results by setting a set of evaluation indicators. For example, evaluation indicators such as temperature drop, current fluctuation range, and maintenance cost can be set to calculate the simulation verification evaluation value. Assuming that the preset simulation verification evaluation value is 0.8, and the actual simulation verification evaluation value is 0.9, it means that the effect of the preliminary maintenance plan is very good and can be implemented directly. If the simulation verification evaluation value is less than the preset value, such as 0.7, it means that the preliminary maintenance plan is insufficient and needs to be optimized and adjusted. The optimization adjustment can include adding additional inspection steps, replacing higher quality parts, improving maintenance processes, etc. Through optimization and adjustment, it can be ensured that the final maintenance strategy is both effective and reliable. Through this series of steps, not only can the specific fault type of the fault source be accurately identified from multi-dimensional feature extraction and fault type classification, but also the effectiveness and reliability of the maintenance plan can be ensured through knowledge reasoning and simulation verification. In the end, this method not only improves the accuracy and efficiency of charging pile fault diagnosis and maintenance, but also provides a scientific basis for the maintenance and management of charging piles, ensures the stable operation of charging facilities, and improves user experience. For example, in the actual application scenario of an electric vehicle charging station, this method can quickly and accurately identify the specific cause of overheating of the charging interface and develop an effective maintenance plan, thereby reducing the downtime of the charging pile and ensuring the continuity and stability of the charging service.

[0122] The above describes the charging pile fault diagnosis method in the embodiment of the present invention. The following describes the charging pile fault diagnosis device in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, a device for diagnosing faults in a charging pile includes:

[0123] The acquisition module 21 is used to acquire the fault status data of each component of the target charging pile, and perform feature extraction on the fault status data to obtain fault feature data;

[0124] An identification module 22, configured to identify faults of components of the target charging pile based on the fault characteristic data to obtain potential faulty components;

[0125] A construction module 23 is used to obtain a component map of a target charging pile from a database, and construct a map of the potential faulty components based on the component map to obtain a potential faulty component association map;

[0126] An analysis module 24 is used to analyze the influence relationship between potential fault components based on the potential fault component association map to obtain a fault propagation path;

[0127] A calculation module 25, configured to calculate the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model;

[0128] The formulation module 26 is used to take the potential fault component corresponding to the greatest fault probability as the fault source, analyze the fault type of the fault source based on the fault source, and formulate the maintenance strategy of the target charging pile based on the fault type.

[0129] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0130] Reference Figure 3 The present invention also provides a computer device in an embodiment, wherein the internal structure of the computer device can be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0131] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0132] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0133] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0134] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0135] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A charging pile fault diagnosis method, characterized in that: The following steps are involved: Acquire fault status data of each component of the target charging pile, and perform feature extraction on the fault status data to obtain fault feature data; Performing fault identification on components of the target charging pile based on the fault characteristic data to obtain potential fault components; Acquire a component map of the target charging pile from a database, and construct a map of the potential faulty components based on the component map to obtain a potential faulty component association map; Analyze the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path; Calculating the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model; The potential fault component corresponding to the greatest fault probability is taken as the fault source, and the fault type of the fault source is analyzed by a preset classification algorithm, and a maintenance strategy for the target charging pile is formulated based on the fault type; The step of constructing a graph of the potential faulty components based on the component graph to obtain a potential faulty component association graph includes: Performing structural analysis on the component atlas of the target charging pile to obtain structural relationship analysis results between the components; Constructing a component topology relationship matrix based on the structural relationship analysis results; Based on the component topology relationship matrix, the interaction mode between the components of the target charging pile is captured to obtain the interaction mode between the components; wherein the interaction mode between the components is the action relationship and connection mode between the components; Modeling the influence relationship of the potential fault components based on the interaction mode between the components to obtain the fault relationship of the potential fault components; By using a preset graph attention network, based on the fault relationship of the potential fault component, the potential fault component is weighted in the component graph of the target charging pile to obtain a weight coefficient of the fault component; Connecting corresponding potential faulty components based on the faulty component weight coefficients to obtain a potential faulty component association map; The analyzing the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path includes: Based on the potential fault component association graph, the node centrality of each potential fault component is calculated to obtain the centrality index of each potential fault component; wherein the node centrality includes degree centrality, closeness centrality and betweenness centrality; Analyze the influence relationship between potential fault components based on the centrality index; Based on the influence relationship between the potential fault components, the potential fault components in the fault component association map are divided into communities to identify potential fault components with similar fault propagation characteristics, and obtain a dynamic community structure of potential fault components; A causal relationship analysis is performed on potential fault components in the dynamic community structure through a preset causal reasoning model to obtain a fault propagation causal path, and the fault propagation causal path is used as the fault propagation path.

2. The charging pile fault diagnosis method according to claim 1, characterized in that: The acquiring of the fault status data of each component of the target charging pile and performing feature extraction on the fault status data to obtain fault feature data includes: The fault parameters of each component in the target charging pile are collected by using multiple sensors arranged inside the target charging pile to obtain original fault state parameters; Performing parameter clustering on the original fault state parameters to obtain clustered fault parameters; wherein the clustered fault parameters include voltage parameters, current parameters and temperature parameters; Performing time-frequency analysis on the clustered fault parameters to obtain fault time-frequency characteristic data; Performing dimension reduction processing on the fault time-frequency characteristic data to obtain dimension-reduced fault parameters; Feature extraction is performed on the dimension-reduced fault parameters to obtain fault feature data.

3. The charging pile fault diagnosis method according to claim 1, characterized in that: The performing fault identification on the components of the target charging pile based on the fault characteristic data to obtain potential fault components includes: Acquire historical charging pile fault feature data, perform tensor decomposition on the historical charging pile fault feature data to obtain a three-dimensional tensor feature matrix, and perform dimensionality reduction processing on the three-dimensional tensor feature matrix to obtain a core tensor and a factor matrix; Performing non-negative matrix decomposition on the core tensor and factor matrix to obtain a historical fault feature subspace; wherein the historical fault feature subspace includes typical fault mode information of various components of the historical charging pile; Performing fault identification learning on the historical fault feature subspace through a preset initial fault identification algorithm to obtain a fault identification algorithm; Performing time-frequency fault identification on the fault feature data based on the fault identification algorithm to obtain time-frequency fault feature data; Based on the time-frequency fault characteristic data, fault identification is performed on components of the target charging pile to obtain potential fault components.

4. The charging pile fault diagnosis method according to claim 1, characterized in that: The calculating the failure probability of the potential fault component based on the fault propagation path by using a preset Bayesian network model includes: Structural modeling of the fault propagation path is performed using a preset Bayesian network model to obtain a Bayesian network structure of the fault propagation; Based on the Bayesian network structure, fault parameter estimation is performed on each potential fault component in the fault propagation path to obtain a conditional probability distribution of each potential fault component; Performing a causal effect analysis on the potential fault components based on the conditional probability distribution to obtain a causal effect analysis result; wherein the causal effect analysis result is a causal influence result of each potential fault component; Based on the causal effect analysis result, the fault expansion of the potential fault component in the time dimension is performed to obtain a dynamic fault propagation probability; A robustness calculation is performed on the potential fault component based on the dynamic fault propagation probability to obtain the failure probability of the potential fault component.

5. The charging pile fault diagnosis method according to claim 1, characterized in that: The method of analyzing the fault type of the fault source by a preset classification algorithm and formulating a maintenance strategy for the target charging pile based on the fault type includes: Performing multi-dimensional feature extraction on the fault source corresponding to the maximum fault probability to obtain a multi-modal fault feature; Classifying the multi-modal fault features into fault types using a preset classification algorithm to obtain a preliminary fault type; Performing knowledge reasoning on the preliminary fault type to obtain a fault type report; Preliminarily designing a maintenance strategy for the target charging pile based on the fault type report to obtain a preliminary maintenance plan; Performing simulation verification on the fault source based on the preliminary maintenance plan to obtain a simulation verification result; The simulation verification result is evaluated to obtain a simulation verification evaluation value. If the simulation verification evaluation value is less than a preset simulation verification evaluation value, the preliminary maintenance plan is optimized and adjusted based on the simulation verification result to obtain a maintenance strategy for the target charging pile.

6. A charging pile fault diagnosis device, characterized in that: include: An acquisition module is used to acquire the fault status data of each component of the target charging pile, and perform feature extraction on the fault status data to obtain fault feature data; An identification module, used to identify faults of components of the target charging pile based on the fault characteristic data to obtain potential fault components; A construction module, used for obtaining a component map of a target charging pile from a database, and constructing a map of the potential faulty components based on the component map to obtain a potential faulty component association map; An analysis module, configured to analyze the influence relationship between the potential fault components based on the potential fault component association map to obtain a fault propagation path; A calculation module, used to calculate the failure probability of the potential fault component based on the fault propagation path through a preset Bayesian network model; A formulation module, used to take the potential fault component corresponding to the greatest fault probability as the fault source, analyze the fault type of the fault source through a preset classification algorithm, and formulate a maintenance strategy for the target charging pile based on the fault type; The step of constructing a graph of the potential faulty components based on the component graph to obtain a potential faulty component association graph includes: Performing structural analysis on the component atlas of the target charging pile to obtain structural relationship analysis results between the components; Constructing a component topology relationship matrix based on the structural relationship analysis results; Based on the component topology relationship matrix, the interaction mode between the components of the target charging pile is captured to obtain the interaction mode between the components; wherein the interaction mode between the components is the action relationship and connection mode between the components; Modeling the influence relationship of the potential fault components based on the interaction mode between the components to obtain the fault relationship of the potential fault components; By using a preset graph attention network, based on the fault relationship of the potential fault component, the potential fault component is weighted in the component graph of the target charging pile to obtain a weight coefficient of the fault component; Connecting corresponding potential faulty components based on the faulty component weight coefficients to obtain a potential faulty component association map; The analyzing the influence relationship between the potential fault components based on the potential fault component association map to obtain the fault propagation path includes: Based on the potential fault component association graph, the node centrality of each potential fault component is calculated to obtain the centrality index of each potential fault component; wherein the node centrality includes degree centrality, closeness centrality and betweenness centrality; Analyze the influence relationship between potential fault components based on the centrality index; Based on the influence relationship between the potential fault components, the potential fault components in the fault component association map are divided into communities to identify potential fault components with similar fault propagation characteristics, and obtain a dynamic community structure of potential fault components; A causal relationship analysis is performed on potential fault components in the dynamic community structure through a preset causal reasoning model to obtain a fault propagation causal path, and the fault propagation causal path is used as the fault propagation path.

7. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images