Charging pile fault diagnosis method based on network model

Abstract

The invention discloses a charging pile fault diagnosis method based on a network model, and relates to the technical field of fault diagnosis, and the method comprises the steps: carrying out the single-mode feature extraction based on the preprocessed operation data, and carrying out the splicing and fusion of the single-mode features; constructing a space-time diagram neural network model, and generating node-level features after space-time feature extraction based on the fused single-mode features; fault classification is carried out on the node-level features through group collaborative fault diagnosis; based on the fault classification result, diagnosis result feedback and optimization are carried out, and accurate characterization of the complex operation state of the charging pile is achieved; and the accuracy and global applicability of fault classification are improved. According to the overall scheme, a complete, accurate and efficient charging pile fault diagnosis process is formed from multi-modal data processing to group collaborative analysis and then to dynamic optimization, and the accuracy, robustness and applicability are all remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a charging pile fault diagnosis method based on a network model. Background Art

[0002] With the rapid popularization of new energy vehicles, charging piles, as key infrastructure, are increasing in number and frequency of use. However, charging piles may fail due to various factors during long-term operation, such as hardware aging, software anomalies, or external environmental influences. These failures not only affect the service life of the charging pile itself, but may also reduce charging efficiency and even cause safety hazards. Therefore, the diagnosis and maintenance of charging pile failures are of great practical significance. At present, charging pile fault diagnosis mainly relies on traditional methods, including fixed rule-based diagnosis methods and statistical analysis-based prediction models. Fixed rule-based methods achieve diagnosis by manually formulating fault feature thresholds or logical rules, but this method relies too much on human experience and is difficult to deal with the implicit relationship between features when facing complex multimodal data, resulting in low diagnostic accuracy and robustness. On the other hand, although prediction models based on statistical analysis, such as support vector machines (SVM) and random forests (RF), can predict faults through data training, they are still insufficient in dealing with the high dimensionality, temporal sequence, and complex spatial correlation of charging pile operation data. In addition, these methods usually ignore the collaborative relationship between charging piles and fail to make full use of group collaborative information, so they are limited in global fault analysis and accurate classification.

[0003] In view of the shortcomings of existing technologies, in recent years, diagnostic methods based on deep learning have gradually attracted attention, especially graph neural networks (GNNs), which have become a research hotspot due to their advantages in processing complex topological relationships and spatiotemporal data. However, existing fault diagnosis methods based on graph neural networks still have shortcomings. First, traditional graph neural networks are mostly used for static graph structures, which are difficult to adapt to the timing characteristics and dynamic changes of charging pile operation data, resulting in the inability to accurately capture spatiotemporal dependencies. Second, most of the current research focuses only on single-modal data (such as voltage, current, etc.), fails to effectively integrate the characteristics of multi-modal operation data, and cannot fully reflect the operating status of the charging pile. Therefore, in the field of fault diagnosis of charging piles, there is an urgent need for a diagnostic method that can comprehensively consider multi-modal characteristics, spatiotemporal correlations, and group collaborative relationships to improve the accuracy of fault classification and the reliability of diagnostic results. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a charging pile fault diagnosis method based on a network model to solve the problems in the prior art of being difficult to adapt to the timing characteristics of charging pile operation data, being unable to effectively integrate multimodal data, and ignoring the temporal and spatial correlation.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a charging pile fault diagnosis method based on a network model, which includes collecting operating data of the charging pile and preprocessing the collected operating data; extracting single-modal features based on the preprocessed operating data, and splicing and fusing the single-modal features; constructing a spatiotemporal graph neural network model, and generating node-level features after spatiotemporal feature extraction based on the fused single-modal features; classifying faults of the node-level features through group collaborative fault diagnosis; and providing feedback and optimization of the diagnosis results based on the fault classification results.

[0008] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, wherein: the operation data of the charging pile is collected, and the specific steps are as follows:

[0009] Electrical signals, temperature signals, vibration signals, sound signals and image data are collected through the internal device sensors and external data acquisition devices of the charging pile.

[0010] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, the preprocessing of the collected operation data is carried out in the following specific steps:

[0011] The collected operation data is processed through data cleaning, data synchronization and alignment, and data normalization.

[0012] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, wherein: the single-mode feature extraction is performed based on the pre-processed operation data, and the specific steps are as follows:

[0013] Based on multi-scale entropy, the signal complexity features are extracted through signal segmentation, multi-scale interpretation, entropy value calculation and feature aggregation. The expression is:

[0014]

[0015] Among them, F elec is the multi-scale complexity characteristic of the electrical signal, S l is the sample entropy of the lth scale, x l is the signal at the lth scale, β is the nonlinear adjustment parameter, T lis the time range corresponding to the lth scale, L is the number of layers of multi-scale decomposition, t represents time, and l represents the scale number in the multi-scale analysis;

[0016] Based on variational mode decomposition, the modal energy features are extracted through signal decomposition, modal energy calculation and modal energy distribution feature extraction. The expression is:

[0017]

[0018] Among them, F vib is the modal energy characteristic of the vibration signal, E k is the energy of the kth IMF, PSD k is the power spectral density of the kth IMF, k represents the kth mode generated by the vibration signal in the variational mode decomposition, K represents the total number of modes, and α is the smoothing parameter;

[0019] Based on Choi-Williams distribution, the time-frequency distribution features are extracted by signal framing, calculation of autocorrelation function, time-frequency distribution calculation and feature aggregation. The expression is:

[0020]

[0021] Among them, F sound is the energy concentration characteristic of the sound signal, S(t,f) is the time-frequency distribution of the sound signal, f is the frequency, σ f is the frequency smoothing parameter, f max and f min is the frequency range;

[0022] Based on the convolutional neural network, high-dimensional image features are extracted, and the expression is:

[0023] F image =ReLU(W*I+b);

[0024] Among them, F image is the image feature vector, W is the convolution kernel matrix, I is the high-dimensional image, b is the bias term of the image feature vector, and * is the convolution operation.

[0025] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, the specific steps of splicing and fusing the single-modal features are as follows:

[0026] The extracted electrical signal features, vibration signal features, sound signal features and image features are concatenated into a joint feature vector F in element order;

[0027] By constructing a dynamic attention map and nonlinear tensor decoupling, the fusion feature is generated, and the expression is:

[0028]

[0029] Among them, F ′ i represents the ith feature after fusion, T ′ ijk represents the interaction value between the i, j, and k features, where i, j, and k represent the indexes of the features.

[0030] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, wherein: the spatiotemporal graph neural network model is constructed, and the node-level features after spatiotemporal feature extraction are generated based on the fused single-modal features. The specific steps are as follows:

[0031] Nodes are defined based on charging pile components. The feature vectors of nodes are represented by fused single-modal features. Edges are constructed based on the physical connection relationship and spatial proximity relationship between components and the dynamic evolution of the time dimension.

[0032] The features of the node itself and its neighboring nodes are aggregated through graph convolution operations to extract local spatial dependencies. The spatial representation of node features is achieved by aggregating the features of adjacent nodes and using the dynamic normalization adjustment of edge weights and features.

[0033] Combine the information of historical time steps to capture the time dependency, extract the time feature through time convolution, and after the spatial feature extraction and time feature extraction, integrate the two to generate the node-level feature after spatiotemporal feature extraction. The expression is:

[0034]

[0035] in, is the node-level feature after spatiotemporal feature extraction, For node v i The time characteristics of is the spatial feature aggregation of neighbor nodes, To integrate the learnable weight matrix of spatial and temporal features, is the feature concatenation operation, is the bias term, and g is the hyperbolic tangent activation function.

[0036] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, wherein: the fault classification and analysis of node-level features are performed through group collaborative fault diagnosis, and the specific steps are as follows:

[0037] The spatiotemporal graph neural network extracts the spatiotemporal features of nodes and calculates the mutual influence weights between nodes by combining the feature similarity and topological relationship between nodes through a dynamic neighborhood collaborative weight mechanism.

[0038] Based on the label propagation mechanism, the fault information of the initial labeled nodes is gradually propagated to the unlabeled nodes. At the same time, the feature guidance factor is introduced to directly participate in the classification optimization, and the fault category probability distribution of each node is obtained through iterative calculation;

[0039] According to the actual application scenarios, the fault categories are divided into normal state, minor fault, moderate fault and severe fault.

[0040] As a preferred solution of the charging pile fault diagnosis method based on the network model of the present invention, wherein: the diagnosis result feedback and optimization are performed based on the fault classification and analysis results, and the specific steps are as follows:

[0041] Perform error evaluation on the fault classification results, calculate the weighted cross entropy loss to quantify the classification error, screen the set of misclassified nodes and their distribution characteristics, and analyze the error concentration categories and graph topology positions;

[0042] Based on the gradient-guided feature importance evaluation method, the feature matrix is ​​optimized and the quality of effective features is enhanced by eliminating low-contribution feature dimensions and neighborhood weighted smoothing;

[0043] The model parameters are adjusted in combination with error feedback, the smoothing parameters and feature guidance factors of the collaborative weights are optimized, and regularization terms are added to control the model complexity. The objective optimization function is updated to iteratively improve the classification performance.

[0044] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the charging pile fault diagnosis method based on the network model as described in the first aspect of the present invention is implemented.

[0045] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the charging pile fault diagnosis method based on a network model as described in the first aspect of the present invention is implemented.

[0046] The beneficial effects of the present invention are as follows: the present invention realizes accurate characterization of the complex operating status of charging piles through feature extraction and fusion of multimodal data; captures the spatiotemporal correlation characteristics of equipment operation through the construction of spatiotemporal graph neural network and node-level feature generation; improves the accuracy and global applicability of fault classification through group collaborative diagnosis and dynamic optimization. The overall invention scheme forms a complete, accurate and efficient charging pile fault diagnosis process from multimodal data processing to group collaborative analysis and then to dynamic optimization, which has significant improvements in accuracy, robustness and applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0048] Figure 1 This is a flow chart of the charging pile fault diagnosis method based on the network model in Example 1.

[0049] Figure 2 This is a schematic diagram of generating node-level features after spatiotemporal feature extraction in Example 1. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0051] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0052] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0053] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a charging pile fault diagnosis method based on a network model, comprising the following steps:

[0054] S1: Collecting the operation data of the charging pile and preprocessing the collected operation data;

[0055] Furthermore, electrical signals, temperature signals, vibration signals, sound signals and image data are collected through internal device sensors of the charging pile and external data acquisition devices;

[0056] It should be noted that electrical signals include current, voltage and power, with a sampling frequency of 1kHz to ensure high-precision capture of dynamic changes during operation; the temperature signal sampling frequency is 1Hz, which is suitable for slowly changing temperature monitoring; the vibration signal sampling frequency is 10kHz, which is used to capture subtle mechanical vibrations; the sound signal sampling frequency is 44.1kHz; the image data acquisition frequency is 1 frame per second, which is suitable for monitoring the appearance and environment of the equipment;

[0057] Process the collected operation data through data cleaning, data synchronization and alignment, and data normalization;

[0058] It should be noted that data cleaning includes removing noise from the collection of electrical signals, vibration signals and sound signals, using linear interpolation to fill missing values ​​for data lost during the collection process, and performing outlier detection and elimination; data synchronization and alignment includes timestamp standardization, interpolation alignment and data alignment.

[0059] S2: Extract unimodal features based on the preprocessed running data, and concatenate and fuse the unimodal features;

[0060] Based on multi-scale entropy, the signal complexity features are extracted through signal segmentation, multi-scale interpretation, entropy value calculation and feature aggregation. The expression is:

[0061]

[0062] Among them, F elec is the multi-scale complexity characteristic of the electrical signal, S l is the sample entropy of the lth scale, x l is the signal at the lth scale, β is the nonlinear adjustment parameter, T l is the time range corresponding to the lth scale, L is the number of layers of multi-scale decomposition, t represents time, and l represents the scale number in the multi-scale analysis;

[0063] It should be noted that signal segmentation is to segment the electrical signal by time to form multiple fixed length interfaces; multi-scale interpretation is to use a method based on the segment mean to process the signal at multiple scales, and the signal at each scale is generated by downsampling; at each scale, the sample entropy of the signal is calculated to measure the complexity of the signal, and the expression is:

[0064]

[0065] Among them, S l is the sample entropy of the lth scale, p l,j is the probability of the jth state of the signal at the lth scale, M lis the number of states at the lth scale, l is the scale number in the multiscale entropy analysis, and j represents the state number at the lth scale;

[0066] The entropy values ​​at all scales are weighted averaged to obtain the final multi-scale entropy feature;

[0067] Based on variational mode decomposition, the modal energy features are extracted through signal decomposition, modal energy calculation and modal energy distribution feature extraction. The expression is:

[0068]

[0069] Among them, F vib is the modal energy characteristic of the vibration signal, E k is the energy of the kth IMF, PSD k is the power spectral density of the kth IMF, k represents the kth mode generated by the vibration signal in the variational mode decomposition, K represents the total number of modes, and α is the smoothing parameter;

[0070] It should be noted that signal decomposition is to input the vibration signal into the VKM algorithm and decompose it into K intrinsic mode functions. IMF is the vibration mode of the signal. The energy characteristics of IMF in the time domain are calculated. The expression is:

[0071]

[0072] Among them, E k is the energy characteristic of IMF in time domain, u k (t) is the kth IMF, t0 and t1 are the time horizons;

[0073] Based on Choi-Williams distribution, the time-frequency distribution features are extracted by signal framing, calculation of autocorrelation function, time-frequency distribution calculation and feature aggregation. The expression is:

[0074]

[0075] Among them, F sound is the energy concentration characteristic of the sound signal, S(t,f) is the time-frequency distribution of the sound signal, f is the frequency, σ f is the frequency smoothing parameter, f max and f min is the frequency range;

[0076] It should be noted that signal framing is to divide the sound signal into several frames, and calculate the autocorrelation function of each frame signal. The expression is:

[0077]

[0078] Among them, R(τ) is the autocorrelation function, A(t) is the sound signal, and τ is the time delay;

[0079] The time-frequency distribution is calculated using the Choi-Williams distribution formula, which is expressed as:

[0080]

[0081] Among them, σ is the time smoothing parameter;

[0082] Feature aggregation is to calculate the energy concentration characteristics of the time-frequency distribution to obtain the final sound signal characteristics;

[0083] Based on the convolutional neural network, high-dimensional image features are extracted, and the expression is:

[0084] F image =ReLU(W*I+b);

[0085] Among them, F image is the image feature vector, W is the convolution kernel matrix, I is the high-dimensional image, b is the bias term of the image feature vector, and * is the convolution operation.

[0086] The extracted electrical signal features, vibration signal features, sound signal features and image features are concatenated into a joint feature vector F in element order;

[0087] By constructing a dynamic attention map and nonlinear tensor decoupling, the fusion feature is generated, and the expression is:

[0088]

[0089] Among them, F ′ i represents the ith feature after fusion, T ′ ijk represents the interaction value between the i, j, and k features, where i, j, and k represent the indexes of the features.

[0090] It should be noted that based on the concatenated joint features, a dynamic attention graph between modalities is constructed, and the mutual influence between modalities is constructed through a nonlinear function, which is expressed as:

[0091]

[0092] Among them, A ij represents the dynamic attention weight between the i-th feature and the j-th feature, |F i -F j | is the absolute difference between the two features, σ(F) is the normalization factor;

[0093] Based on the dynamic attention map, a joint feature tensor is constructed, and the tensor is nonlinearly decoupled to generate a decoupled feature tensor. The final fusion feature is generated by weighting the elements of the tensor.

[0094] S3: Build a spatiotemporal graph neural network model to generate node-level features after spatiotemporal feature extraction based on the fused single-modal features;

[0095] Nodes are defined based on charging pile components. The feature vectors of nodes are represented by fused single-modal features. Edges are constructed based on the physical connection relationship and spatial proximity relationship between components and the dynamic evolution of the time dimension.

[0096] The features of the node itself and its neighboring nodes are aggregated through graph convolution operations to extract local spatial dependencies. The spatial representation of node features is achieved by aggregating the features of adjacent nodes and using the dynamic normalization adjustment of edge weights and features. The expression is:

[0097]

[0098] in, For node v i At the l+1th layer and time t, the spatial feature w ij (t) is time t moment v i and v j The edge weight, F j (t) is the node v j The characteristics at time t, is the learnable weight matrix in the spatial direction, ∥F j (t)∥2 is the node v j The L2 norm of the eigenvector, is the bias term, g is the hyperbolic tangent activation function;

[0099] Combine the information of historical time steps to capture the time dependency, and extract the time feature through time convolution. The expression is:

[0100]

[0101] in, is the node v at time t i In the temporal features of the l+1th layer, is the node v at time q i Features, is the learnable weight matrix in the time direction, Δt is the time window length, For node v i The L1 norm of the feature vector at time τ, η is the adjustment parameter that controls the smoothness of the temporal feature, is the bias term, h is the rectified linear unit, and q is the time step from time t-Δt to t;

[0102] After spatial feature extraction and temporal feature extraction, the two are integrated to generate node-level features after spatiotemporal feature extraction. The expression is:

[0103]

[0104] in, is the node-level feature after spatiotemporal feature extraction, For node v i The time characteristics of is the spatial feature aggregation of neighbor nodes, To integrate the learnable weight matrix of spatial and temporal features, is the feature concatenation operation, is the bias term, and g is the hyperbolic tangent activation function.

[0105] S4: Fault classification based on node-level features through group collaborative fault diagnosis;

[0106] The spatiotemporal graph neural network extracts the spatiotemporal features of nodes and calculates the mutual influence weights between nodes by combining the feature similarity and topological relationship between nodes through a dynamic neighborhood collaborative weight mechanism.

[0107] Based on the label propagation mechanism, the fault information of the initial labeled nodes is gradually propagated to the unlabeled nodes. At the same time, the feature guidance factor is introduced to directly participate in the classification optimization, and the fault category probability distribution of each node is obtained through iterative calculation;

[0108] According to the actual application scenarios, the fault categories are divided into normal state, minor fault, moderate fault and severe fault.

[0109] It should be noted that fault classification is defined based on the degree of change in the characteristics of the equipment's operating status, and is determined through feature extraction and engineering experience combined with machine learning algorithms. Each type of fault corresponds to a specific set of characteristic patterns and probability distributions.

[0110] S5: Provide diagnostic result feedback and optimization based on fault classification results

[0111] Perform error evaluation on the fault classification results, calculate the weighted cross entropy loss to quantify the classification error, screen the set of misclassified nodes and their distribution characteristics, and analyze the error concentration categories and graph topology positions;

[0112] Based on the gradient-guided feature importance evaluation method, the feature matrix is ​​optimized and the quality of effective features is enhanced by eliminating low-contribution feature dimensions and neighborhood weighted smoothing;

[0113] Adjust model parameters in combination with error feedback, including optimizing the smoothing parameters and feature guidance factors of collaborative weights, adding regularization terms to control model complexity, and updating the target optimization function to iteratively improve classification performance;

[0114] It should be noted that the optimized classification results are re-evaluated and a diagnostic report is generated, including the overall accuracy, category distribution, error node analysis and optimization suggestions, so as to achieve precise diagnostic results and continuous optimization of the model, and improve the reliability and robustness of fault classification.

[0115] This embodiment also provides a computer device, which is suitable for the case of a charging pile fault diagnosis method based on a network model, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the charging pile fault diagnosis method based on a network model as proposed in the above embodiment.

[0116] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0117] The present embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the charging pile fault diagnosis method based on a network model as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic storage device, a flash memory, a disk or an optical disk.

[0118] In summary, the present invention achieves accurate characterization of the complex operating status of charging piles through: feature extraction and fusion of multimodal data; captures the spatiotemporal correlation characteristics of equipment operation through the construction of spatiotemporal graph neural network and node-level feature generation; and improves the accuracy and global applicability of fault classification through group collaborative diagnosis and dynamic optimization. The overall invention scheme, from multimodal data processing to group collaborative analysis, and then to dynamic optimization, forms a complete, accurate and efficient charging pile fault diagnosis process, which has significant improvements in accuracy, robustness and applicability.

[0119] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A charging pile fault diagnosis method based on a network model, characterized in that: include, Collect the operation data of the charging pile and pre-process the collected operation data; Based on the preprocessed running data, single-modal features are extracted and then the single-modal features are spliced ​​and fused; Construct a spatiotemporal graph neural network model to generate node-level features after spatiotemporal feature extraction based on the fused single-modal features; Through group collaborative fault diagnosis, fault classification is performed based on node-level features; Based on the fault classification results, the diagnostic results are fed back and optimized.

2. The charging pile fault diagnosis method based on the network model according to claim 1, characterized in that: The specific steps of collecting the operation data of the charging pile are as follows: Electrical signals, temperature signals, vibration signals, sound signals and image data are collected through the internal device sensors and external data acquisition devices of the charging pile.

3. The charging pile fault diagnosis method based on the network model as claimed in claim 2, characterized in that: The specific steps of preprocessing the collected operation data are as follows: The collected operation data is processed through data cleaning, data synchronization and alignment, and data normalization.

4. The charging pile fault diagnosis method based on the network model as claimed in claim 3, characterized in that: The single-mode feature extraction is performed based on the pre-processed operation data. The specific steps are as follows: Based on multi-scale entropy, the signal complexity features are extracted through signal segmentation, multi-scale interpretation, entropy value calculation and feature aggregation. The expression is: Among them, F elec is the multi-scale complexity characteristic of the electrical signal, S l is the sample entropy of the lth scale, x l is the signal at the lth scale, β is the nonlinear adjustment parameter, T l is the time range corresponding to the lth scale, L is the number of layers of multi-scale decomposition, t represents time, and l represents the scale number in the multi-scale analysis; Based on variational mode decomposition, the modal energy features are extracted through signal decomposition, modal energy calculation and modal energy distribution feature extraction. The expression is: Among them, F vib is the modal energy characteristic of the vibration signal, E k is the energy of the kth IMF, PSD k is the power spectral density of the kth IMF, k represents the kth mode generated by the vibration signal in the variational mode decomposition, K represents the total number of modes, and α is the smoothing parameter; Based on Choi-Williams distribution, the time-frequency distribution features are extracted by signal framing, calculation of autocorrelation function, time-frequency distribution calculation and feature aggregation. The expression is: Among them, F sound is the energy concentration characteristic of the sound signal, S(t,f) is the time-frequency distribution of the sound signal, f is the frequency, σ f is the frequency smoothing parameter, f max and f min is the frequency range; Based on the convolutional neural network, high-dimensional image features are extracted, and the expression is: F image =ReLU(W*I+b); Among them, F image is the image feature vector, W is the convolution kernel matrix, I is the high-dimensional image, b is the bias term of the image feature vector, and * is the convolution operation.

5. The charging pile fault diagnosis method based on the network model as claimed in claim 4, characterized in that: The specific steps of concatenating and fusing the single-modal features are as follows: The extracted electrical signal features, vibration signal features, sound signal features and image features are concatenated into a joint feature vector F in element order; By constructing a dynamic attention map and nonlinear tensor decoupling, the fusion feature is generated, and the expression is: Among them, F ′ i represents the ith feature after fusion, T ′ ijk represents the interaction value between the i, j, and k features, where i, j, and k represent the indexes of the features.

6. The charging pile fault diagnosis method based on the network model as claimed in claim 5, characterized in that: The spatiotemporal graph neural network model is constructed to generate node-level features after spatiotemporal feature extraction based on the fused single-modal features. The specific steps are as follows: Nodes are defined based on charging pile components. The feature vectors of nodes are represented by fused single-modal features. Edges are constructed based on the physical connection relationship and spatial proximity relationship between components and the dynamic evolution of the time dimension. The features of the node itself and its neighboring nodes are aggregated through graph convolution operations to extract local spatial dependencies. The spatial representation of node features is achieved by aggregating the features of adjacent nodes and using the dynamic normalization adjustment of edge weights and features. Combine the information of historical time steps to capture the time dependency, extract the time feature through time convolution, and after the spatial feature extraction and time feature extraction, integrate the two to generate the node-level feature after spatiotemporal feature extraction. The expression is: in, is the node-level feature after spatiotemporal feature extraction, For node v i The time characteristics of is the spatial feature aggregation of neighbor nodes, To integrate the learnable weight matrix of spatial and temporal features, is the feature concatenation operation, is the bias term, and g is the hyperbolic tangent activation function.

7. The charging pile fault diagnosis method based on the network model as claimed in claim 6, characterized in that: The specific steps of group collaborative fault diagnosis and fault classification based on node-level features are as follows: The spatiotemporal graph neural network extracts the spatiotemporal features of nodes and calculates the mutual influence weights between nodes by combining the feature similarity and topological relationship between nodes through a dynamic neighborhood collaborative weight mechanism. Based on the label propagation mechanism, the fault information of the initial labeled nodes is gradually propagated to the unlabeled nodes. At the same time, the feature guidance factor is introduced to directly participate in the classification optimization, and the fault category probability distribution of each node is obtained through iterative calculation; According to the actual application scenarios, the fault categories are divided into normal state, minor fault, moderate fault and severe fault.

8. The charging pile fault diagnosis method based on the network model as claimed in claim 7, characterized in that: The diagnostic result feedback and optimization are performed based on the fault classification result. The specific steps are as follows: Perform error evaluation on the fault classification results, calculate the weighted cross entropy loss to quantify the classification error, screen the set of misclassified nodes and their distribution characteristics, and analyze the error concentration categories and graph topology positions; Based on the gradient-guided feature importance evaluation method, the feature matrix is ​​optimized and the quality of effective features is enhanced by eliminating low-contribution feature dimensions and neighborhood weighted smoothing; The model parameters are adjusted in combination with error feedback, the smoothing parameters and feature guidance factors of the collaborative weights are optimized, and regularization terms are added to control the model complexity. The objective optimization function is updated to iteratively improve the classification performance.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the charging pile fault diagnosis method based on the network model according to any one of claims 1 to 8 are implemented.

10. 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 charging pile fault diagnosis method based on a network model according to any one of claims 1 to 8 are implemented.

Images