Cable branch box comprehensive online monitoring system

Abstract

The present invention discloses a comprehensive online monitoring system for cable branch boxes, comprising: a data acquisition module for collecting operation monitoring data and historical operation data of the cable branch box; a data transmission module for transmitting the collected data to a central monitoring platform; a data preprocessing module for preprocessing the data; an analysis module for applying an improved transformation convolutional neural network to analyze the data and identify faults and abnormal conditions of the cable branch box; a prediction module for combining historical operation data and current monitoring data and applying an improved graph neural network combined with an attention mechanism to perform fault prediction; a central monitoring platform for generating fault warning information; and a maintenance optimization module for proposing corresponding preventive measures and maintenance recommendations. The present invention combines improved convolutional neural networks and graph neural networks to intelligently analyze cable branch box data, achieve fault prediction and warning, and has the advantages of high precision, strong real-time performance, and optimized maintenance.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring of electric power equipment, and in particular to a comprehensive online monitoring system for a cable branch box. Background Art

[0002] As a crucial component of the power system, cable branch boxes are responsible for distributing electrical energy from the main cables to various electrical devices, and therefore play a vital role in power transmission and distribution. With the increase in electricity demand, the scope of use of cable branch boxes has gradually expanded, and their position in power transmission and distribution networks has become increasingly important. However, traditional cable branch box monitoring methods mostly rely on manual inspections or static detection methods, which suffer from poor real-time performance, low accuracy, and high maintenance costs. These traditional monitoring methods not only have difficulty in promptly detecting faults or abnormal conditions in cable branch boxes, but also fail to effectively predict faults and provide early warnings, leading to unstable power system operations and the risk of unexpected equipment downtime.

[0003] Currently, monitoring of cable branch boxes primarily relies on manual inspections and periodic checks. Manual inspections are not only inefficient but also prone to missed faults due to human interference. Periodic inspections rely on set time periods, but this approach doesn't provide real-time insights into the equipment's operating status. This is especially true when the equipment has been operating for extended periods, which increases the probability of failure. Furthermore, traditional monitoring methods often rely on single-sensor data, failing to fully consider the equipment's multi-dimensional operating parameters, such as current, voltage, temperature, and humidity. This reduces predictability and response speed.

[0004] To improve the monitoring efficiency and fault prevention capabilities of cable branch boxes, a growing number of studies are applying intelligent monitoring methods, particularly leveraging modern deep learning techniques for fault detection and prediction. For example, deep convolutional neural networks (DCNNs) are being applied to image or signal data that indicate power equipment faults. However, these methods are primarily used for pattern recognition in signal data and are limited in their ability to analyze time series data. At the same time, graph neural networks (GNNs) are also being gradually introduced into power equipment monitoring and fault prediction, particularly for modeling internode relationships in power networks. Despite their strong structured data processing capabilities, GNNs still face numerous challenges in practical applications, particularly when processing large amounts of data, where computational complexity can be prohibitive.

[0005] Current intelligent monitoring systems typically transmit sensor data to a monitoring platform and use data-based models to perform fault detection and analysis. However, these approaches still have some shortcomings. For example, existing fault prediction models mostly rely on static training datasets and fixed model parameters, and are unable to adaptively adjust their prediction capabilities based on the actual operating status and real-time data of the equipment. Furthermore, existing systems often ignore the temporal correlation between data and the dynamic changes in equipment status during the analysis process, resulting in the inability to timely update fault prediction results and reducing the accuracy of fault warnings.

[0006] Further analysis shows that the common deep learning-based monitoring methods in existing technologies rely on traditional convolutional neural networks or simple graph neural network models. The limitations of these methods are mainly reflected in the processing of complex time series data. Although traditional convolutional neural networks perform well in the fields of image and signal processing, their modeling capabilities for time series data are relatively weak, making it difficult to accurately capture the time-varying characteristics of cable branch boxes. Although graph neural networks can process graph-structured data and establish relationships between nodes, most traditional graph neural network models fail to effectively consider the dynamic changes between nodes and the weight differences between different nodes, resulting in limited accuracy of the prediction results. Therefore, most fault prediction models in existing technologies cannot take into account the temporal nature of equipment status and the complexity of the relationships between equipment, and there is a large room for optimization.

[0007] While existing cable branch box monitoring methods have incorporated some intelligent technologies, such as neural network-based predictive models, they still face several key flaws. First, traditional monitoring methods often fail to provide real-time fault warnings, and the equipment's operating status information is not updated in a timely manner, making it impossible to predict and prevent potential faults in advance. Second, existing methods often rely on data from a single sensor and fail to effectively integrate multi-dimensional sensor data, which limits the accuracy and reliability of fault detection. Finally, the models of existing fault prediction systems are generally relatively fixed and lack adaptive capabilities, making it impossible to optimize prediction strategies in real time based on the different operating environments and status changes of the cable branch box.

[0008] Therefore, how to combine deep learning technology, especially convolutional neural networks and graph neural networks, to establish a more accurate, efficient and adaptive cable branch box monitoring system has become the research focus of current technology. Summary of the Invention

[0009] One purpose of the present invention is to propose a comprehensive online monitoring system for cable branch boxes. The present invention combines an improved transformation convolutional neural network and an improved graph neural network, and uses an attention mechanism to perform in-depth analysis and fault prediction on data. This method collects multi-dimensional monitoring data of cable branch boxes in real time, and through intelligent analysis and prediction, realizes early warning of faults and optimization of maintenance strategies, thereby improving the safety, reliability and economy of the power system.

[0010] A comprehensive online monitoring system for a cable branch box according to an embodiment of the present invention includes:

[0011] Data acquisition module, used to collect operation monitoring data and historical operation data of the cable branch box;

[0012] Data transmission module, used to transmit the collected operation monitoring data and historical operation data to the central monitoring platform through the Internet of Things technology;

[0013] Data preprocessing module, used to preprocess the real-time transmitted operation monitoring data;

[0014] An analysis module is used to analyze the pre-processed data using an improved transform convolutional neural network to identify faults and abnormal conditions of the cable branch box;

[0015] The prediction module combines historical operation data with current monitoring data and uses an improved graph neural network combined with an attention mechanism to predict faults.

[0016] The central monitoring platform is used to receive the fault prediction results output by the prediction module, generate fault warning information, and display the operating status of the cable branch box through a visual interface;

[0017] The maintenance optimization module is used to optimize maintenance strategies based on fault prediction results and equipment maintenance plans, and to propose corresponding preventive measures and maintenance recommendations.

[0018] Optionally, modules can be connected using the following methods:

[0019] S1. Collecting operation monitoring data and historical operation data of the cable branch box through sensors;

[0020] S2. Transmit the collected operation monitoring data and historical operation data to the central monitoring platform through the Internet of Things technology;

[0021] S3. Preprocess the real-time transmitted operation monitoring data to generate preprocessed data;

[0022] S4. Construct an improved transform convolutional neural network, apply the improved transform convolutional neural network to analyze the preprocessed data, use transform convolution operations to extract features from the preprocessed data, introduce residual connections into the improved transform convolutional neural network, and use the fully connected layer of the improved transform convolutional neural network to identify faults and abnormal conditions of the cable branch box;

[0023] S5. Based on historical operation data and current operation monitoring data, combined with the attention mechanism and the improved graph neural network, through graph convolution operations and calculation of attention weights, the node features of the improved graph neural network are updated, the information transmission between nodes is adjusted, and the pooling layer and fully connected layer of the improved graph neural network are used to perform fault prediction and generate prediction results;

[0024] S6. Generate fault warning information based on the analysis and prediction results, and display the operating status of the cable branch box through the monitoring platform;

[0025] S7. Combine the prediction results with the equipment maintenance plan, optimize the maintenance strategy, and propose corresponding preventive measures and maintenance recommendations.

[0026] Optionally, the operation monitoring data includes current, voltage, temperature and humidity.

[0027] Optionally, S3 includes the following specific steps:

[0028] S31. Perform a preliminary check on the received operation monitoring data to ensure the integrity and validity of the data. For values ​​that do not meet the format requirements or are abnormal, mark them according to the preset threshold and process or eliminate missing values.

[0029] S32. Use Kalman filter to filter and denoise the operation monitoring data to remove the existing random noise:

[0030]

[0031] in, represents the current estimated value, Represents the estimated value of the previous moment, y k Represents the current observation value, H k represents the state transition matrix, H k represents the Kalman gain;

[0032] S33. Normalize the denoised data to eliminate differences in data dimensions and bring the range of operation monitoring data within a unified standard.

[0033]

[0034] Among them, x' represents the normalized data, x represents the original data value, min( x ) and max ( x ) Represent the minimum and maximum values ​​in the original data values ​​respectively;

[0035] S34. Smoothing the normalized data using a sliding window method, setting a window size N, and calculating the average value of the N data points before and after each data point:

[0036]

[0037] Among them, y i represents the smoothed data point, x j represents the original data point, and P represents the window size;

[0038] S35. Perform time series processing on the smoothed data, sort the data in chronological order and construct a timestamp;

[0039] S36. Perform differential processing on the time series data using the first difference formula:

[0040] Δx t =x t -x t-1 ;

[0041] Where Δx t Represents the data after difference, x t Represents the data at the current moment, x t-1 Represents the data of the previous moment;

[0042] S37. Perform outlier detection on the processed data and use the Z-score method to measure the deviation of each data point. If the Z-score is greater than a preset threshold, the data point is considered an outlier:

[0043]

[0044] Among them, Z c represents the value of the Z score, x represents the data point, μ represents the mean of the data set, and σ represents the standard deviation of the data set.

[0045] Optionally, S4 includes the following specific steps:

[0046] S41. Constructing an improved transformation convolutional neural network, wherein the improved transformation convolutional neural network comprises:

[0047] Input layer, inputs preprocessed data and enters the network for processing;

[0048] Transform the convolution layer, extract local features, and dynamically adjust the convolution kernel;

[0049] Temporal convolution layer, extracts time-dependent features and captures the changing pattern of time series;

[0050] Pooling layer, downsampling the features;

[0051] Multi-scale convolutional layer, which captures features at different scales and processes global and local features;

[0052] The fully connected output layer integrates the extracted features and uses the ReLU activation function to output the prediction results;

[0053] S42. Process the preprocessed data using a transform convolution operation. In the transform convolution operation, the convolution kernel is dynamically adjusted:

[0054]

[0055] in, Represents the feature map extracted by the convolution layer, M represents the height of the convolution kernel, and N represents the width of the convolution kernel. Represents the preprocessed input data, Represents the convolution kernel, which is used to perform convolution operations with the input data to extract local features, b (1) Represents the bias term, which is added to the convolution result to adjust the output. m and n represent the position index of the element in the convolution kernel, and i and j represent the position index of the element in the output feature map.

[0056] S43. Use the maximum pooling method to select the maximum value from each pooling window to generate the pooled feature map:

[0057]

[0058] in, Represents the output feature map after the pooling operation, which reduces the spatial dimension. Represents the output feature map of the convolution operation, pool window represents the pooling window, max represents the maximum value, and ∈ represents the set operation "belongs";

[0059] S44. Use convolution kernel Processing the feature map after pooling Capturing temporal dependencies through temporal convolution operations:

[0060]

[0061] in, Represents the output of the temporal convolution, which is the captured time-dependent feature, and T represents the length of the temporal convolution kernel. represents the weight of the temporal convolution kernel, represents the feature map after pooling, b(3) Represents the bias term of temporal convolution;

[0062] S45. Extract features simultaneously at different scales through multiple convolution kernels:

[0063]

[0064] in, Represents the output of multi-scale convolution, which contains feature information of multiple scales. Represents the output of the temporal convolution operation, which is used as the input of the multi-scale convolution. m represents the length of the m-th scale convolution kernel, b (4) represents the bias term of multi-scale convolution, Represents the weight of the convolution kernel of the mth scale, and m represents the number of scale convolution kernels;

[0065] S46, introduce residual connection, the features processed by residual connection are passed to the fully connected layer for processing:

[0066]

[0067] Among them, Y (5) Represents the output of the fully connected layer, the feature representation after residual connection processing, ReLU represents the nonlinear activation function, W (5) represents the weight matrix of the fully connected layer, Y (4) represents the output of multi-scale convolution, b (5 represents the bias term of the fully connected layer, Represents the input of the residual connection, which comes from the input features of the multi-scale convolutional layer;

[0068] S47. According to the output of the fully connected layer, a threshold is set. When the probability of a certain type of fault exceeds the set threshold, it is considered that the cable branch box has this type of fault.

[0069] Optionally, S5 includes the following specific steps:

[0070] S51. Generate an input data matrix by splicing the historical operation data and the current operation monitoring data together along the time axis. The input data matrix represents the historical status of the cable branch box and the monitoring data at the current moment. Each row represents the data of one time step, and each column corresponds to one monitoring data:

[0071]

[0072] Where X represents the input data matrix, X history Indicates historical operation data, X current Indicates the current operation monitoring data;

[0073] S52. Input the input data matrix X into the improved graph convolutional network and map it to the node feature space of the graph structure through a mapping function. The mapping function linearly maps the input data matrix X through a weight matrix and a bias term, and strengthens the relationship between nodes in combination with the adaptively adjusted adjacency matrix:

[0074] H=f(X)=XW node +b node +αA adj ·X;

[0075] Among them, H represents the node feature matrix of the graph neural network, which is the state information of the node, and W node Represents the mapping weight matrix, which is responsible for mapping the input data to the node space, b node represents the bias term, which adjusts the node features, α represents the adaptive coefficient, which determines the influence of the adjacency matrix on the node features, and A adj Represents the adjacency matrix of the graph, describing the connection relationship between nodes in the graph, f represents the mapping function, αA adj Represents the interaction between the adjacency matrix and the input feature matrix, which is used to dynamically adjust the features of the nodes according to the neighbor relationships of the nodes;

[0076] S53. Introduce the attention mechanism so that each node adaptively focuses on the most relevant neighbor node features when updating. By calculating the attention weights between neighbor nodes, the information transfer between nodes is adjusted and the node features are updated:

[0077]

[0078] Among them, H k+1 represents the feature matrix of node k+1, represents the normalized adjacency matrix, N ( k ) represents the set of neighbor nodes of node k, ReLU represents the nonlinear activation function, W k represents the weight matrix of node k, ε represents the learning parameter of the attention mechanism, which is used to adjust the attention weight and describe the influence of node i on node k, H k and H i Represents the feature matrix of node k and node i respectively, Attention ( H k ,H i) represents the attention score between node k and its neighbor node i:

[0079]

[0080] Where exp represents the natural exponential function, LeakyReLU represents the nonlinear activation function, a represents the parameter for calculating the attention weight, T represents the transpose operation, || represents the operation of splicing two vectors together, and W represents the weight matrix;

[0081] S54. After several layers of graph convolution and attention mechanisms, the node features are updated, the graph size is reduced through graph pooling, and the node information is aggregated:

[0082] H pooled =Pooling ( ∑ i∈V α i ·H k ) ;

[0083] Among them, H pooled Represents the node feature matrix after graph pooling, which is the graph information compressed after pooling. Pooling represents the graph pooling operation, V represents the node set in the graph, and α i represents the weight applied to node i during the pooling process, H k Represents the feature matrix of node k;

[0084] S55. The pooled feature matrix is ​​passed through the fully connected layer for final fault prediction:

[0085] Y output =Softmax ( W out H pooled +b out ) ;

[0086] Among them, Y output Represents the predicted probability of the fault category, which is used to predict the fault type of the cable branch box. out Represents the weight matrix of the output layer, mapping the pooled node features to the fault category space, Softmax represents normalization, b out Represents the bias term of the output layer.

[0087] The beneficial effects of the present invention are:

[0088] First, the intelligent monitoring method of the present invention effectively solves the deficiencies of traditional methods in processing multi-dimensional data of cable branch boxes by improving the transformation convolutional neural network and the improved graph neural network. The improved transformation convolutional neural network can dynamically adjust the convolution kernel to adapt to changes in different types of data, thereby extracting more accurate features. At the same time, the improved graph neural network combined with the attention mechanism can adaptively consider the complex relationships between nodes and the importance of different nodes, thereby optimizing the accuracy of fault prediction. By combining the improved transformation convolutional neural network and the improved graph neural network, the faults and abnormal conditions of the cable branch box can be identified more comprehensively and accurately, avoiding the accuracy and real-time problems brought about by the traditional monitoring method relying only on a single data source or static model.

[0089] Secondly, the present invention combines historical operating data with current monitoring data to predict faults. This allows fault warnings to be based not only on the current operating status but also on the equipment's historical performance and environmental changes. This comprehensive data analysis approach significantly improves the accuracy of fault predictions, enabling power systems to issue early warnings before equipment failures occur, thereby enabling early intervention and reducing unexpected downtime.

[0090] Finally, by employing an improved graph neural network and attention mechanism, fault prediction not only identifies potential problems but also performs adaptive optimization based on the operating mode of the cable branch box equipment. The system can dynamically adjust prediction strategies based on real-time data to adapt to changes in different equipment and environments, improving the robustness and flexibility of the prediction system. Furthermore, combined with a maintenance optimization module, it can optimize equipment maintenance strategies based on prediction results, rationally scheduling inspection cycles and maintenance plans, thereby improving operational efficiency and resource utilization, and reducing equipment maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0092] Figure 1 This is a method flow chart of a comprehensive online monitoring system for cable branch boxes proposed by the present invention;

[0093] Figure 2 This is a fault prediction flow chart of a comprehensive online monitoring system for cable branch boxes proposed by the present invention. DETAILED DESCRIPTION

[0094] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0095] refer to Figure 1 and Figure 2 , a comprehensive online monitoring system for cable branch boxes, comprising:

[0096] Data acquisition module, used to collect operation monitoring data and historical operation data of the cable branch box;

[0097] Data transmission module, used to transmit the collected operation monitoring data and historical operation data to the central monitoring platform through the Internet of Things technology;

[0098] Data preprocessing module, used to preprocess the real-time transmitted operation monitoring data;

[0099] An analysis module is used to analyze the pre-processed data using an improved transform convolutional neural network to identify faults and abnormal conditions of the cable branch box;

[0100] The prediction module combines historical operation data with current monitoring data and uses an improved graph neural network combined with an attention mechanism to predict faults.

[0101] The central monitoring platform is used to receive the fault prediction results output by the prediction module, generate fault warning information, and display the operating status of the cable branch box through a visual interface;

[0102] The maintenance optimization module is used to optimize maintenance strategies based on fault prediction results and equipment maintenance plans, and to propose corresponding preventive measures and maintenance recommendations.

[0103] In this embodiment, the modules are connected through the following methods:

[0104] S1. Collecting operation monitoring data and historical operation data of the cable branch box through sensors;

[0105] S2. Transmit the collected operation monitoring data and historical operation data to the central monitoring platform through the Internet of Things technology;

[0106] S3. Preprocess the real-time transmitted operation monitoring data to generate preprocessed data;

[0107] S4. Construct an improved transform convolutional neural network, apply the improved transform convolutional neural network to analyze the preprocessed data, use transform convolution operations to extract features from the preprocessed data, introduce residual connections into the improved transform convolutional neural network, and use the fully connected layer of the improved transform convolutional neural network to identify faults and abnormal conditions of the cable branch box;

[0108] S5. Based on historical operation data and current operation monitoring data, combined with the attention mechanism and the improved graph neural network, through graph convolution operations and calculation of attention weights, the node features of the improved graph neural network are updated, the information transmission between nodes is adjusted, and the pooling layer and fully connected layer of the improved graph neural network are used to perform fault prediction and generate prediction results;

[0109] S6. Generate fault warning information based on the analysis and prediction results, and display the operating status of the cable branch box through the monitoring platform;

[0110] S7. Combine the prediction results with the equipment maintenance plan, optimize the maintenance strategy, and propose corresponding preventive measures and maintenance recommendations.

[0111] In this embodiment, the operation monitoring data includes current, voltage, temperature and humidity.

[0112] In this embodiment, S3 includes the following specific steps:

[0113] S31. Perform a preliminary check on the received operation monitoring data to ensure the integrity and validity of the data. For values ​​that do not meet the format requirements or are abnormal, mark them according to the preset threshold and process or eliminate missing values.

[0114] S32. Use Kalman filter to filter and denoise the operation monitoring data to remove the existing random noise:

[0115]

[0116] in, represents the current estimated value, Represents the estimated value of the previous moment, y k Represents the current observation value, H k represents the state transition matrix, H k represents the Kalman gain;

[0117] S33. Normalize the denoised data to eliminate differences in data dimensions and bring the range of operation monitoring data within a unified standard.

[0118]

[0119] Among them, x' represents the normalized data, x represents the original data value, min ( x ) and max ( x ) Represent the minimum and maximum values ​​in the original data values ​​respectively;

[0120] S34. Smoothing the normalized data using a sliding window method, setting a window size N, and calculating the average value of the N data points before and after each data point:

[0121]

[0122] Among them, y i represents the smoothed data point, x j represents the original data point, and P represents the window size;

[0123] S35. Perform time series processing on the smoothed data, sort the data in chronological order and construct a timestamp;

[0124] S36. Perform differential processing on the time series data using the first difference formula:

[0125] Δx t =x t -x t-1 ;

[0126] Where Δx t Represents the data after difference, x t Represents the data at the current moment, x t-1 Represents the data of the previous moment;

[0127] S37. Perform outlier detection on the processed data and use the Z-score method to measure the deviation of each data point. If the Z-score is greater than a preset threshold, the data point is considered an outlier:

[0128]

[0129] Among them, Z c represents the value of the Z score, x represents the data point, μ represents the mean of the data set, and σ represents the standard deviation of the data set.

[0130] In this embodiment, S4 includes the following specific steps:

[0131] S41. Constructing an improved transformation convolutional neural network, wherein the improved transformation convolutional neural network comprises:

[0132] Input layer, inputs preprocessed data and enters the network for processing;

[0133] Transform the convolution layer, extract local features, and dynamically adjust the convolution kernel;

[0134] Temporal convolution layer, extracts time-dependent features and captures the changing pattern of time series;

[0135] Pooling layer, downsampling the features;

[0136] Multi-scale convolutional layer, which captures features at different scales and processes global and local features;

[0137] The fully connected output layer integrates the extracted features and uses the ReLU activation function to output the prediction results;

[0138] S42. Process the preprocessed data using a transform convolution operation. In the transform convolution operation, the convolution kernel is dynamically adjusted:

[0139]

[0140] in, Represents the feature map extracted by the convolution layer, M represents the height of the convolution kernel, and N represents the width of the convolution kernel. Represents the preprocessed input data, Represents the convolution kernel, which is used to perform convolution operations with the input data to extract local features, b (1) Represents the bias term, which is added to the convolution result to adjust the output. m and n represent the position index of the element in the convolution kernel, and i and j represent the position index of the element in the output feature map.

[0141] S43. Use the maximum pooling method to select the maximum value from each pooling window to generate the pooled feature map:

[0142]

[0143] in, Represents the output feature map after the pooling operation, which reduces the spatial dimension. Represents the output feature map of the convolution operation, pool window represents the pooling window, max represents the maximum value, and ∈ represents the set operation "belongs";

[0144] S44. Use convolution kernel Processing the feature map after pooling Capturing temporal dependencies through temporal convolution operations:

[0145]

[0146] in, Represents the output of the temporal convolution, which is the captured time-dependent feature, and T represents the length of the temporal convolution kernel. represents the weight of the temporal convolution kernel, represents the feature map after pooling, b (3) Represents the bias term of temporal convolution;

[0147] S45. Extract features simultaneously at different scales through multiple convolution kernels:

[0148]

[0149] in, Represents the output of multi-scale convolution, which contains feature information of multiple scales. Represents the output of the temporal convolution operation, which is used as the input of the multi-scale convolution. m represents the length of the m-th scale convolution kernel, b (4) represents the bias term of multi-scale convolution, Represents the weight of the convolution kernel of the mth scale, and m represents the number of scale convolution kernels;

[0150] S46, introduce residual connection, the features processed by residual connection are passed to the fully connected layer for processing:

[0151]

[0152] Among them, Y (5) Represents the output of the fully connected layer, the feature representation after residual connection processing, ReLU represents the nonlinear activation function, W (5) represents the weight matrix of the fully connected layer, Y (4) represents the output of multi-scale convolution, b (5 represents the bias term of the fully connected layer, Represents the input of the residual connection, which comes from the input features of the multi-scale convolutional layer;

[0153] S47. According to the output of the fully connected layer, a threshold is set. When the probability of a certain type of fault exceeds the set threshold, it is considered that the cable branch box has this type of fault.

[0154] In this embodiment, S5 includes the following specific steps:

[0155] S51. Generate an input data matrix by splicing the historical operation data and the current operation monitoring data together along the time axis. The input data matrix represents the historical status of the cable branch box and the monitoring data at the current moment. Each row represents the data of one time step, and each column corresponds to one monitoring data:

[0156]

[0157] Where X represents the input data matrix, X history Indicates historical operation data, X current Indicates the current operation monitoring data;

[0158] S52. Input the input data matrix X into the improved graph convolutional network and map it to the node feature space of the graph structure through a mapping function. The mapping function linearly maps the input data matrix X through a weight matrix and a bias term, and strengthens the relationship between nodes in combination with the adaptively adjusted adjacency matrix:

[0159] H=f ( X ) =XW node +b node +αA adj ·X;

[0160] Among them, H represents the node feature matrix of the graph neural network, which is the state information of the node, and W node Represents the mapping weight matrix, which is responsible for mapping the input data to the node space, b node represents the bias term, which adjusts the node features, α represents the adaptive coefficient, which determines the influence of the adjacency matrix on the node features, and A adj Represents the adjacency matrix of the graph, describing the connection relationship between nodes in the graph, f represents the mapping function, αA adj Represents the interaction between the adjacency matrix and the input feature matrix, which is used to dynamically adjust the features of the nodes according to the neighbor relationships of the nodes;

[0161] S53. Introduce the attention mechanism so that each node adaptively focuses on the most relevant neighbor node features when updating. By calculating the attention weights between neighbor nodes, the information transfer between nodes is adjusted and the node features are updated:

[0162]

[0163] Among them, H k+1 represents the feature matrix of node k+1, represents the normalized adjacency matrix, N ( k ) represents the set of neighbor nodes of node k, ReLU represents the nonlinear activation function, W k represents the weight matrix of node k, ε represents the learning parameter of the attention mechanism, which is used to adjust the attention weight and describe the influence of node i on node k, H k and H i Represents the feature matrix of node k and node i respectively, Attention ( H k ,H i) represents the attention score between node k and its neighbor node i:

[0164]

[0165] Where exp represents the natural exponential function, LeakyReLU represents the nonlinear activation function, a represents the parameter for calculating the attention weight, T represents the transpose operation, || represents the operation of splicing two vectors together, and W represents the weight matrix;

[0166] S54. After several layers of graph convolution and attention mechanisms, the node features are updated, the graph size is reduced through graph pooling, and the node information is aggregated:

[0167] H pooled =Pooling ( ∑ i∈V α i ·H k ) ;

[0168] Among them, H pooled Represents the node feature matrix after graph pooling, which is the graph information compressed after pooling. Pooling represents the graph pooling operation, V represents the node set in the graph, and α i represents the weight applied to node i during the pooling process, H k Represents the feature matrix of node k;

[0169] S55. The pooled feature matrix is ​​passed through the fully connected layer for final fault prediction:

[0170] Y output =Softmax ( W out H pooled +b out ) ;

[0171] Among them, Y output Represents the predicted probability of the fault category, which is used to predict the fault type of the cable branch box. out Represents the weight matrix of the output layer, mapping the pooled node features to the fault category space, Softmax represents normalization, b out Represents the bias term of the output layer.

[0172] Example 1:

[0173] To verify the feasibility of the present invention, it was applied to a cable branch box monitoring system for a power company in a core urban area. To ensure the stable operation of the power grid, the power company has deployed multiple cable branch boxes, which are responsible for providing power distribution services to residents and commercial users in different areas. Because cable branch boxes are exposed to the elements for long periods of time and operate under complex conditions, the power company requires an intelligent system that can monitor equipment status in real time, identify faults promptly, and make predictions. Traditional manual inspections and regular maintenance methods are unable to detect potential faults in a timely manner, and are prone to false detections and missed detections, which cannot effectively reduce sudden equipment failures.

[0174] Against this backdrop, the power company introduced an online monitoring system for cable branch boxes based on the Internet of Things (IoT) and deep learning. By installing sensors in each cable branch box, the system collects multiple parameters, such as current, voltage, temperature, and humidity, in real time. Leveraging IoT technology, the system transmits this data to a central monitoring platform. Using this platform, the system combines improved transform convolutional neural networks and graph neural networks to analyze the data and predict faults.

[0175] In the initial stage of system deployment, 10 cable branch boxes were selected as monitoring targets, and data was continuously monitored for two months. By comparing the monitoring data with the actual occurrence of equipment failures, the accuracy of the fault prediction model was verified. Table 1 shows the cable branch box monitoring data.

[0176] Table 1 Cable branch box monitoring data

[0177]

[0178]

[0179] Table 2 Comparison between cable branch box fault prediction and actual status

[0180] Equipment Number Actual fault status Fault prediction status Failure prediction probability Is the prediction accurate? 1 normal normal 5 yes 2 normal normal 4 yes 3 normal normal 6 yes 4 abnormal abnormal 85 yes 5 normal normal 7 yes 6 normal normal 6 yes 7 abnormal abnormal 80 yes 8 normal normal 5 yes 9 abnormal abnormal 82 yes 10 normal normal 6 yes

[0181] Table 2 shows the five dimensions of the system: device number, actual fault status, predicted fault status, predicted fault probability, and prediction accuracy. Comparing the actual and predicted fault statuses clearly shows that the predicted results for all devices are consistent with their actual fault status, demonstrating the system's high accuracy in fault identification and prediction. For example, device 4's actual fault status is "abnormal," and the system's predicted result is also "abnormal," with a predicted probability of 85%. This high-probability prediction demonstrates that the system effectively identifies potential equipment failure risks and issues timely warnings, facilitating proactive maintenance and fault prevention.

[0182] The "Failure Prediction Probability" column in the table displays the predicted probability of failure for each device. This data column demonstrates the consistency between the predicted results and the actual failure status. In cases where the predicted probability of failure was high (e.g., 85% and 80% for Device 4 and Device 7, respectively), the system accurately identified the failure status of these devices and successfully issued an early warning. For other devices with a lower predicted probability of failure (e.g., Devices 1, 2, and 3), the predicted results were normal, which is consistent with the actual operating status of the devices and indicates that these devices did not fail during the monitoring period.

[0183] The "Prediction Accuracy" column confirms the accuracy of the system's predictions. According to the table data, all devices showed a "yes" prediction, indicating that the predicted faults fully corresponded to the actual fault conditions. This demonstrates that the cable branch box online monitoring system demonstrated effective fault identification and prediction capabilities in this test.

[0184] In summary, the comparative analysis of experimental data demonstrates the accuracy and reliability of the proposed cable branch box fault prediction system. This system not only monitors equipment operating status in real time but also accurately predicts equipment failures, improving power system operation and maintenance efficiency, reducing the occurrence of sudden equipment failures, and ensuring the stability and security of the power supply.

[0185] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A comprehensive online monitoring system for cable branch boxes, characterized in that: include: Data acquisition module, used to collect operation monitoring data and historical operation data of the cable branch box; Data transmission module, used to transmit the collected operation monitoring data and historical operation data to the central monitoring platform through the Internet of Things technology; Data preprocessing module, used to preprocess the real-time transmitted operation monitoring data; An analysis module is used to analyze the pre-processed data using an improved transform convolutional neural network to identify faults and abnormal conditions of the cable branch box; The prediction module combines historical operation data with current operation monitoring data, applies an improved graph neural network combined with an attention mechanism, updates the node features of the improved graph neural network through graph convolution operations and calculation of attention weights, adjusts information transmission between nodes, and uses the pooling layer and fully connected layer of the improved graph neural network for fault prediction; The central monitoring platform is used to receive the fault prediction results output by the prediction module, generate fault warning information, and display the operating status of the cable branch box through a visual interface; The maintenance optimization module is used to optimize maintenance strategies based on fault prediction results and equipment maintenance plans, and to propose corresponding preventive measures and maintenance recommendations; The identification of the fault and abnormal state of the cable branch box includes the following specific steps: S41. Constructing an improved transformation convolutional neural network, wherein the improved transformation convolutional neural network comprises: Input layer, inputs preprocessed data and enters the network for processing; Transform the convolution layer, extract local features, and dynamically adjust the convolution kernel; Temporal convolution layer, extracts time-dependent features and captures the changing pattern of time series; Pooling layer, downsampling the features; Multi-scale convolutional layer, which captures features at different scales and processes global and local features; The fully connected output layer integrates the extracted features and uses The activation function outputs the prediction result; S42. Process the preprocessed data using a transform convolution operation. In the transform convolution operation, the convolution kernel is dynamically adjusted: ； in, represents the feature map extracted by the transformed convolutional layer, represents the height of the convolution kernel, represents the width of the convolution kernel, Represents the preprocessed input data, Represents the convolution kernel, which is used to perform transformation convolution operation with the input data to extract local features. Represents the bias term, which is added to the convolution result to adjust the output. and Represents the position index of the element in the convolution kernel, and Represents the position index of the element in the output feature map; S43. Use the maximum pooling method to select the maximum value from each pooling window to generate the pooled feature map: ； in, Represents the output feature map after the pooling operation, which reduces the spatial dimension. represents the output feature map of the transformed convolution operation, represents the pooling window, Indicates the maximum value, Indicates the set operation "belongs to"; S44. Use the weight of the temporal convolution kernel Processing the feature map after pooling , capturing temporal dependencies through temporal convolution operations: ； in, Represents the output of temporal convolution, which is the captured time-dependent features. Represents the length of the temporal convolution kernel, represents the weight of the temporal convolution kernel, represents the feature map after pooling, Represents the bias term of temporal convolution; S45. Extract features simultaneously at different scales through multiple convolution kernels: ； in, Represents the output of multi-scale convolution, which contains feature information of multiple scales. Represents the output of the temporal convolution operation, which serves as the input of the multi-scale convolution. Indicates the The length of the convolution kernel of scale, represents the bias term of multi-scale convolution, Indicates the The weight of the convolution kernel of each scale, Indicates the number of scale convolution kernels; S46, introduce residual connection, the features processed by residual connection are passed to the fully connected layer for processing: ； in, Represents the output of the fully connected layer, which is the feature representation processed by the residual connection. represents a nonlinear activation function, represents the weight matrix of the fully connected layer, represents the output of multi-scale convolution, represents the bias term of the fully connected layer, Represents the input of the residual connection, which comes from the input features of the multi-scale convolutional layer; S47. According to the output of the fully connected layer, a threshold is set. When the probability of a certain type of fault exceeds the set threshold, it is considered that the cable branch box has a fault of this type.

2. A cable branch box comprehensive online monitoring system according to claim 1, characterized in that: The modules are implemented as follows: S1. Collecting operation monitoring data and historical operation data of the cable branch box through sensors; S2. Transmit the collected operation monitoring data and historical operation data to the central monitoring platform through the Internet of Things technology; S3. Preprocess the real-time transmitted operation monitoring data to generate preprocessed data; S4. Construct an improved transform convolutional neural network, apply the improved transform convolutional neural network to analyze the preprocessed data, use transform convolution operation to extract features from the preprocessed data, introduce residual connection into the improved transform convolutional neural network, and use the fully connected layer of the improved transform convolutional neural network to identify faults and abnormal conditions of the cable branch box; S5. Based on historical operation data and current operation monitoring data, combined with the attention mechanism and the improved graph neural network, through graph convolution operations and calculation of attention weights, the node features of the improved graph neural network are updated, the information transmission between nodes is adjusted, and the pooling layer and fully connected layer of the improved graph neural network are used to perform fault prediction and generate prediction results; S6. Generate fault warning information based on the analysis and prediction results, and display the operating status of the cable branch box through the central monitoring platform; S7. Combine the prediction results with the equipment maintenance plan, optimize the maintenance strategy, and propose corresponding preventive measures and maintenance recommendations.

3. A cable branch box comprehensive online monitoring system according to claim 2, characterized in that: The operation monitoring data includes current, voltage, temperature and humidity.

4. A cable branch box comprehensive online monitoring system according to claim 2, characterized in that: The S3 includes the following specific steps: S31. Perform a preliminary check on the received operation monitoring data to ensure the integrity and validity of the data. Data that does not meet the format requirements or is abnormal will be marked using a preset threshold and missing values ​​will be processed or eliminated. S32. Use Kalman filter to filter and denoise the operation monitoring data to remove the existing random noise: ； in, represents the current estimated value, represents the estimated value at the previous moment, represents the current observation value, represents the state transition matrix, represents the Kalman gain; S33. Normalize the denoised data to eliminate differences in data dimensions and bring the range of operation monitoring data within a unified standard. ； in, represents the normalized data, Represents the original data value, and Represent the minimum and maximum values ​​in the original data respectively; S34, use the sliding window method to smooth the normalized data, set a window size , for each data point before and after Calculate the average value of data points: ； in, represents the smoothed data points, represents the original data points, Indicates the window size; S35. Perform time series processing on the smoothed data, sort the data in chronological order and construct a timestamp; S36. Perform differential processing on the data after time series processing, using the first difference formula: ； in, represents the data after difference, Represents the data at the current moment, Represents the data of the previous moment; S37. Perform outlier detection on the processed data and use the Z-score method to measure the deviation of each data point. If the Z-score is greater than a preset threshold, the data point is considered an outlier: ； in, represents the value of the Z score, represents a data point, represents the mean of the data set, Represents the standard deviation of the data set.

5. A cable branch box comprehensive online monitoring system according to claim 2, characterized in that: The S5 includes the following specific steps: S51. Generate an input data matrix by splicing the historical operation data and the current operation monitoring data together along the time axis. The input data matrix represents the historical status of the cable branch box and the monitoring data at the current moment. Each row represents the data of one time step, and each column corresponds to one monitoring data: ； in, represents the input data matrix, Indicates historical operation data. Indicates the current operation monitoring data; S52, input data matrix Input to the improved graph neural network, mapped to the node feature space of the graph structure through a mapping function, the mapping function is a weight matrix and a bias term on the input data matrix Perform linear mapping and combine it with the adaptively adjusted adjacency matrix to strengthen the relationship between nodes: ； Among them, H represents the node feature matrix of the improved graph neural network, which is the state information of the node. Represents the mapping weight matrix, which is responsible for mapping the input data to the node space. Represents the bias term, adjusts the node characteristics, Represents the adaptive coefficient, which determines the influence of the adjacency matrix on the node characteristics. Represents the adjacency matrix of the graph, describing the connection relationship between nodes in the graph, represents the mapping function, Represents the interaction between the adjacency matrix and the input data matrix, which is used to dynamically adjust the characteristics of the nodes according to the neighbor relationships of the nodes; S53. Introduce the attention mechanism so that each node adaptively focuses on the most relevant neighbor node features when updating. By calculating the attention weights between neighbor nodes, the information transfer between nodes is adjusted and the node features are updated: ； in, Representation node The characteristic matrix of represents the normalized adjacency matrix, Representation node The set of neighbor nodes of represents a nonlinear activation function, Representation node The weight matrix, Represents the learning parameters of the attention mechanism, which are used to adjust the attention weight and describe the node For Node influence, and Represents nodes respectively and nodes The characteristic matrix of Representation node and neighbor nodes Attention score between: ； in, represents the natural exponential function, represents a nonlinear activation function, Represents the parameters for calculating attention weights, represents the transpose operation, represents the operation of concatenating two vectors together, represents the weight matrix; S54. After several layers of graph convolution and attention mechanisms, the node features are updated, the graph size is reduced through graph pooling, and the node information is aggregated: ； in, Represents the node feature matrix after graph pooling, which is the graph information compressed after pooling. represents the graph pooling operation, represents the set of nodes in the graph, Indicates the pooling process applied to the node The weight of Representation node The characteristic matrix of S55. The pooled feature matrix is ​​passed through the fully connected layer for final fault prediction: ； in, Represents the predicted probability of the fault category, which is used to predict the fault category of the cable branch box. Represents the weight matrix of the output layer, mapping the pooled node features to the fault category space, represents normalization, Represents the bias term of the output layer.

Images