Intelligent power distribution room operation and maintenance method and system based on multi-source data fusion

Abstract

The invention relates to an intelligent power distribution room operation and maintenance method and system based on multi-source data fusion, and the method comprises the steps: collecting the environment and equipment data of a power distribution room through multi-source sensing equipment, and obtaining original multi-source data; performing standardization processing on the original multi-source data to obtain standardized heterogeneous data; performing space-time correlation modeling on the equipment vibration characteristics and the equipment power parameters through a space-time diagram convolutional network, and fusing the cross-modal depth characteristics extracted by the standardized heterogeneous data to obtain panoramic perception characteristics; performing anomaly detection through an isolated forest-long and short-term memory hybrid model according to the panoramic perception features, and deploying a causal inference engine to analyze a causal relationship among multiple variables to obtain a health state assessment result of the power distribution room; and performing operation and maintenance decision according to the health state evaluation result of the power distribution room, and performing operation and maintenance on the power distribution room. According to the invention, deep fusion of multi-source data is realized, equipment abnormity can be accurately detected, root causes can be analyzed, and the operation and maintenance efficiency and reliability of a power distribution room are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid management, and in particular to an intelligent distribution room operation and maintenance method and system based on multi-source data fusion. Background Art

[0002] With the continuous improvement of the intelligence level of the power system, the distribution room, as a key node for power transmission and distribution, its safe and stable operation is crucial to the entire power system. Traditional distribution room operation and maintenance mainly rely on manual regular inspections and single-sensor monitoring, which have problems such as single data collection, insufficient data utilization, lack of equipment correlation analysis, limited abnormal detection ability, and lack of root cause analysis ability. Although significant progress has been made in intelligent sensing technology and data analysis methods in recent years, there are still many technical challenges in the field of distribution room operation and maintenance.

[0003] Existing distribution room data fusion technologies mainly process multi-source heterogeneous data by simple splicing or weighted average, etc., which are difficult to accurately reflect the true operating state of equipment. In addition, existing technologies are mostly based on correlation analysis rather than causal reasoning, lacking automated root cause analysis ability, unable to determine the source and propagation path of anomalies, increasing the difficulty and time cost of fault troubleshooting. Summary of the Invention

[0004] The present invention provides an intelligent distribution room operation and maintenance method and system based on multi-source data fusion to solve the defects of the existing technology.

[0005] The present invention provides an intelligent distribution room operation and maintenance method based on multi-source data fusion, including: S1: Collect the distribution room environment and equipment data through multi-source sensing devices to obtain original multi-source data; S2: Perform standardization processing on the original multi-source data to obtain standardized heterogeneous data; S3: Perform spatio-temporal correlation modeling on the equipment vibration characteristics and equipment power parameters through a spatio-temporal graph convolutional network, and fuse the cross-modal deep features extracted from the standardized heterogeneous data to obtain panoramic perception features; S4: According to the panoramic perception features, perform anomaly detection through an isolation forest-long short-term memory hybrid model, and deploy a causal reasoning engine to analyze the causal relationship between multiple variables to obtain the distribution room health status evaluation result; S5: Make operation and maintenance decisions according to the distribution room health status evaluation result, and perform operation and maintenance on the distribution room through the operation and maintenance decisions.

[0006] According to the intelligent distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, the original multi-source data in step S1 includes: Environmental temperature and humidity data, which are collected by temperature and humidity sensors; Gas concentration data, which is collected by a gas concentration sensor; Power operation parameter data, which is collected by a power parameter collector; Equipment vibration characteristic data, which is collected by a vibration sensor; Thermal imaging data, which is collected by an infrared thermal imaging device.

[0007] For a smart distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, step S2 specifically includes: Removing noise and detecting outliers from the original multi-source data through a data cleaning algorithm to obtain cleaned data; Performing maximum-minimum normalization on the numerical data in the cleaned data through normalization processing to obtain normalized numerical data; Performing transformation processing on the skewed distribution data in the cleaned data through logarithmic transformation to obtain evenly distributed data; Aligning timestamps of data with different sampling frequencies through a time synchronization mechanism to obtain data with consistent time series; Completing the missing parts in the data with consistent time series through a missing value filling algorithm to obtain data with enhanced integrity; Performing unified format conversion on the data with enhanced integrity through a data format converter to obtain the standardized heterogeneous data.

[0008] For a smart distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, step S3 further includes: S31: Based on the equipment connection relationship in the distribution room, establish an equipment topology graph, and adjust the edge weights of the equipment topology graph according to the operation status data of the equipment to obtain a dynamic adjacency matrix; S32: Input the dynamic adjacency matrix into a spatio-temporal graph convolutional network to obtain a spatio-temporal joint feature vector; S33: Perform attention interaction calculation on the equipment power parameters and the spatio-temporal joint feature vector to obtain a coupling coefficient matrix, and perform abnormal region detection on the coupling coefficient matrix based on a threshold segmentation method to obtain coupling features; S34: Extract features from the standardized heterogeneous data to obtain cross-modal deep features; S35: Fuse the coupling features and the cross-modal deep features to obtain panoramic perception features.

[0009] For a smart distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, step S32 further includes: S321: Perform spectral graph convolution on the dynamic adjacency matrix through Chebyshev polynomial approximation to extract the spatial correlation features of the device vibration spectrum; S322: Input the spatial correlation features into a gated temporal convolutional network to extract the temporal correlation features of the vibration signal time series; S323: Concatenate the spatial correlation features and the temporal correlation features to obtain fused features; S324: Through residual connection, add the fused features to the original vibration signal in the standardized heterogeneous data to obtain a spatio-temporal joint feature vector.

[0010] According to a smart distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, step S4 further includes: S41: Input the panoramic perception features into an isolation forest-long short-term memory hybrid model to obtain a device anomaly determination result including temporal anomalies and sudden anomalies; S42: Through a structural equation model, perform root cause localization on the abnormal devices in the device anomaly determination result to obtain a distribution room health status evaluation result including the device anomaly determination result and the root cause analysis result.

[0011] According to a smart distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, step S41 specifically includes: S411: Calculate the prediction residual through the LSTM branch of the isolation forest-long short-term memory hybrid model, and mark the devices with the prediction residual greater than the first preset threshold as abnormal to obtain a first marking result; The expression of the first preset threshold is: , where is the prediction residual, is the standard deviation of the residual sequence; S412: Calculate the path length score through the isolation forest branch of the isolation forest-long short-term memory hybrid model, and mark the devices with the path length score greater than the second preset threshold as abnormal to obtain a second marking result; The expression of the second threshold is: , where is the path length score; S413: When any one of the first marking result and the second marking result is marked as abnormal, output the device anomaly determination result as abnormal; when any one of the first marking result and the second marking result is marked as normal, output the device anomaly determination result as normal.

[0012] According to a smart distribution room operation and maintenance method based on multi-source data fusion provided by the present invention, step S42 further includes: S421: Define latent variables and observed variables, describe the variable relationships between the latent variables and the observed variables, and construct a structural equation model; S422: Calculate the path coefficients between the variables of the structural equation model through partial least squares path analysis; S423: According to the structural equation model and the path coefficients between the variables, perform root cause verification through the counterfactual intervention method to obtain the root cause analysis result; S424: Output the equipment anomaly determination result and the root cause analysis result to obtain the health status assessment result of the distribution substation.

[0013] According to a smart distribution substation operation and maintenance method based on multi-source data fusion provided by the present invention, the expression of the variable relationship in step S421 is:

[0014] where, represents the vibration RMS value, represents mechanical wear, represents the SF6 gas concentration, represents insulation deterioration, represents the partial discharge quantity, represents environmental corrosion, represents the temperature rise rate, represents the th factor loading coefficient, represents the th measurement error term; The expression of the structural equation model in step S421 is:

[0015] where, represents the th path coefficient, represents the th structural residual term; The expression of the path coefficients between the variables in step S422 is: where, represents the estimated path coefficients between the variables, represents the total number of paths, represents the insulation deterioration caused by the th path, represents the mechanical wear caused by the th path, represents the environmental corrosion caused by the th path.

[0016] The present invention also provides a smart distribution substation operation and maintenance system based on multi-source data fusion, including: Data acquisition module: used to acquire the environment and equipment data of the distribution room based on multi-source sensing devices to obtain original multi-source data; Standardization module: used to perform standardization processing on the original multi-source data to obtain standardized heterogeneous data; Feature extraction module: used to perform spatio-temporal correlation modeling on the vibration characteristics and power parameters of equipment through a spatio-temporal graph convolutional network, and fuse the cross-modal deep features extracted from the standardized heterogeneous data to obtain panoramic perception features; Detection module: used to perform anomaly detection through an isolation forest-long short-term memory hybrid model according to the panoramic perception features, and deploy a causal inference engine to analyze the causal relationship between multi-variables to obtain the evaluation result of the health status of the distribution room; Decision output module: used to make operation and maintenance decisions according to the evaluation result of the health status of the distribution room, and perform operation and maintenance on the distribution room through the operation and maintenance decisions.

[0017] A method and system for intelligent operation and maintenance of intelligent distribution rooms based on multi-source data fusion. Through a method combining a spatio-temporal graph convolutional network and a multi-modal fusion algorithm, deep fusion of multi-source heterogeneous data in the distribution room is achieved. It can not only process different types of data such as temperature and humidity, gas concentration, power parameters, equipment vibration, and thermal imaging, but also capture the spatial topological relationship between equipment and the temporal evolution characteristics of parameters at the same time. In particular, the present invention uses Chebyshev polynomials for spectral graph convolution, combined with a gated time convolutional network and residual connections, effectively extracts the complex spatio-temporal coupling relationship between equipment vibration and power parameters, greatly improves the comprehensive perception ability of the operation state of the distribution room, avoids the data island problem in traditional methods, and provides a more comprehensive and accurate data basis for anomaly detection and root cause analysis. The present invention also innovatively proposes an isolation forest-long short-term memory hybrid model, which organically combines the distance-based isolation forest algorithm and the prediction-based LSTM model, constructs a dual detection mechanism that can detect both sudden anomalies and temporal anomalies at the same time, identifies temporal anomalies by calculating the prediction residuals through the LSTM branch, and identifies sudden anomalies by calculating the path length score through the isolation forest branch. The two complement each other's advantages, greatly improving the accuracy and coverage of anomaly detection. Especially for the two types of typical anomalies common in power equipment - temporal anomalies caused by slow deterioration and sudden anomalies caused by sudden failures, the hybrid model of the present invention shows excellent detection performance, greatly reducing the false alarm rate and missed alarm rate, providing a reliable basis for operation and maintenance decisions; in addition, the present invention uses a structural equation model and a counterfactual intervention method for root cause analysis, breaking through the limitations of traditional correlation analysis. By defining latent variables and observed variables and describing the relationships between them, a causal network of the equipment state in the distribution room is constructed, which can accurately identify the root cause and propagation path of anomalies. In practical applications, this root cause analysis method based on causal reasoning can quickly locate the fault source, greatly improving the repair efficiency, reducing the downtime, and lowering the maintenance cost; the method of the present invention can not only effectively process static data, but also establishes a dynamic adjacency matrix update mechanism based on the operation state of the equipment, which can adjust the edge weights of the equipment topology map in real time, enabling the model to adaptively reflect the changes in the equipment state. The dynamic adaptive mechanism enables the system to have good adaptability to long-term evolution factors such as equipment aging and environmental changes, maintaining long-term stable performance. At the same time, the coupling coefficient matrix obtained through attention interaction calculation provides accurate positioning of the abnormal area, further enhancing the adaptability and accuracy of the system.

[0018] Based on the health status assessment results, the present invention realizes intelligent operation and maintenance decisions. By combining the equipment anomaly determination results with the root cause analysis results, a complete information chain is provided for operation and maintenance personnel, greatly reducing the cognitive burden of operation and maintenance personnel, improving the decision-making efficiency and accuracy, realizing the proactive prevention work of distribution room operation and maintenance, effectively extending the service life of equipment, and improving the overall reliability of the power distribution system. Brief Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 It is a schematic flow chart of a smart distribution room operation and maintenance method based on multi-source data fusion provided by an embodiment of the present invention; Figure 2 It is a schematic structural diagram of a smart distribution room operation and maintenance system based on multi-source data fusion provided by an embodiment of the present invention.

[0021] Reference Numerals: 100, acquisition module; 200, standardization module; 300, feature extraction module; 400, detection module; 500, decision output module. Detailed Embodiments

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are some, but not all, embodiments of the present invention. They should not be construed as limiting the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for the purpose of description and cannot be construed as indicating or implying relative importance.

[0023] The following describes the embodiments of the present invention with reference to the drawings.

[0024] As Figure 1 shown, the present invention provides a smart distribution room operation and maintenance method based on multi-source data fusion, including: S1: Collect the distribution room environment and equipment data through multi-source sensing devices to obtain original multi-source data.

[0025] Among them, the original multi-source data in step S1 includes: ambient temperature and humidity data, which is collected by a temperature and humidity sensor; gas concentration data, which is collected by a gas concentration sensor; power operation parameter data, which is collected by a power parameter collector; equipment vibration characteristic data, which is collected by a vibration sensor; and thermal imaging data, which is collected by an infrared thermal imaging device.

[0026] Furthermore, the ambient temperature and humidity data are collected by temperature and humidity sensors, which are arranged in key areas of the distribution room, such as around transformers, inside switchgear cabinets, and near ventilation openings, to monitor the temperature and humidity changes in different areas in real time. The gas concentration data are collected by gas concentration sensors, which mainly monitor gas components such as SF6 gas, ozone, and carbon monoxide in the distribution room. The gas concentration sensors use electrochemical sensing technology or infrared absorption spectrometry to selectively detect multiple gases, with a sensitivity up to the ppm level. The power operation parameter data are collected by a power parameter collector, which is directly connected to the key nodes of the power distribution system to monitor core electrical parameters such as voltage, current, power factor, and harmonic content. The equipment vibration characteristic data are collected by vibration sensors, which are installed on key equipment such as transformers, circuit breakers, and switchgear cabinets to monitor parameters such as the vibration amplitude, frequency, and acceleration of the equipment. The thermal imaging data are collected by an infrared thermal imaging device, which performs a non-contact scan of the surface temperature distribution of the equipment in the distribution room to generate a thermal image.

[0027] S2: Standardize the original multi-source data to obtain standardized heterogeneous data.

[0028] Among them, step S2 specifically includes: removing noise and detecting outliers from the original multi-source data through a data cleaning algorithm to obtain cleaned data; performing maximum-minimum normalization on the numerical data in the cleaned data through normalization processing to obtain normalized numerical data; performing transformation processing on the skewed distribution data in the cleaned data through logarithmic transformation to obtain evenly distributed data; aligning the timestamps of data with different sampling frequencies through a time synchronization mechanism to obtain data with consistent time series; filling in the missing parts of the data with consistent time series through a missing value filling algorithm to obtain data with enhanced integrity; and performing unified format conversion on the data with enhanced integrity through a data format converter to obtain the standardized heterogeneous data.

[0029] Furthermore, the standard isomerized data obtained through preprocessing significantly improves the data quality and usability. First, median filtering and the Z-score method are used to remove noise and detect outliers from the original data, effectively eliminating data contamination caused by factors such as electromagnetic interference and sensor failures. Second, different types of data are standardized through min-max normalization and logarithmic transformation, solving the problems of inconsistent dimensions and uneven distributions of multi-source heterogeneous data. Third, a time synchronization mechanism based on UTC timestamps and a multi-strategy missing value filling algorithm are adopted to overcome technical problems such as large differences in sampling frequencies of different sensors and data missing. Finally, a unified data format and descriptor system are constructed, laying a solid foundation for subsequent spatio-temporal correlation modeling and multi-modal fusion. Compared with traditional simple data cleaning methods, this preprocessing method not only retains the key features of the original data but also enhances the comparability and relevance between data through standardized transformation, thus providing high-quality data support for anomaly detection and fault diagnosis in intelligent distribution rooms.

[0030] S3: Perform spatio-temporal correlation modeling on the vibration characteristics and power parameters of the equipment through a spatio-temporal graph convolutional network, and fuse the cross-modal deep features extracted from the standardized heterogeneous data to obtain panoramic perception features.

[0031] Among them, step S3 further includes: S31: Based on the connection relationships of the equipment in the distribution room, establish an equipment topology graph, and adjust the edge weights of the equipment topology graph according to the operating state data of the equipment to obtain a dynamic adjacency matrix.

[0032] In step S31, first, based on the connection relationships of the equipment in the distribution room, establish an equipment topology graph, and adjust the edge weights of the equipment topology graph according to the operating state data of the equipment to obtain a dynamic adjacency matrix. The equipment topology graph is a graph-theoretic representation of the physical connections and electrical associations of the equipment in the distribution room, where each node represents an equipment (such as a transformer, switchgear, circuit breaker, etc.), and the edges between the nodes represent the physical or electrical connection relationships between the equipment.

[0033] The dynamic adjacency matrix is a mathematical representation form of the equipment topology graph. The matrix elements represent the connection relationship strength between two pieces of equipment. The initial adjacency matrix is determined according to the physical connection relationships of the equipment, and then the edge weights are dynamically adjusted through the equipment operating state data. The weight adjustment uses a state vector composed of the vibration characteristics and power parameters of the equipment. When the states of two pieces of equipment are similar, the edge weight between them increases; otherwise, it decreases. The dynamically adjusted adjacency matrix can reflect the real-time association strength between equipment not only based on physical connections but also based on operating states.

[0034] S32: Input the dynamic adjacency matrix into the spatio-temporal graph convolutional network to obtain a spatio-temporal joint feature vector.

[0035] Further, in step S32, the dynamic adjacency matrix is input into the spatio-temporal graph convolutional network, the purpose of which is to obtain the spatio-temporal joint feature vector. The spatio-temporal graph convolutional network is a deep learning architecture that combines the graph convolutional network (GCN) and the temporal convolutional network (TCN), and can capture the spatial dependence and temporal dynamics of data simultaneously.

[0036] Among them, step S32 further includes: S321: Perform spectral graph convolution on the dynamic adjacency matrix through Chebyshev polynomial approximation to extract the spatial correlation features of the device vibration spectrum.

[0037] First, in step S321, perform spectral graph convolution on the dynamic adjacency matrix through Chebyshev polynomial approximation to extract the spatial correlation features of the device vibration spectrum. Chebyshev polynomial approximation is a method for quickly calculating spectral graph convolution, which avoids the complex operation of directly calculating the eigen-decomposition of the graph Laplacian matrix. Specifically, the K-order Chebyshev polynomial approximation formula is: , where, is the extracted spatial correlation feature, is the polynomial order, is the learnable coefficient, is the Chebyshev polynomial, is the time series of vibration signals. In practical applications, taking can obtain a good approximation effect. Through the approximation operation, the vibration spectrum features of each device node can be weighted and combined with the features of its adjacent devices to extract the spatial correlation features, which reflect the mutual influence and correlation pattern between devices.

[0038] S322: Input the spatial correlation features into the gated temporal convolutional network to extract the temporal correlation features of the vibration signal time series.

[0039] In step S33, the spatial correlation features in step S321 are input into the gated temporal convolutional network to extract the temporal correlation features of the vibration signal time series. The gated temporal convolutional network is an improved one-dimensional convolutional neural network that introduces a gating mechanism to control the information flow. The gating mechanism is similar to the gating unit in the long short-term memory network (LSTM), and can selectively retain or ignore the temporal information, effectively dealing with the long-term dependencies in the long sequence data. In the present invention, the gated temporal convolutional network processes the temporal vibration data of each device node, applies a multi-layer convolutional structure, and each layer contains convolutional kernels of different sizes to capture the pattern changes at different time scales. Residual connections and batch normalization layers are also set between the convolutional layers to accelerate the training process and avoid the problem of gradient disappearance. Through the gated temporal convolutional network, the dynamic features of the device vibration signal changing with time are extracted, such as the periodicity, trend, and mutation characteristics of the vibration mode.

[0040] S323: Concatenate the spatial correlation feature and the temporal correlation feature to obtain a fused feature.

[0041] Next, concatenate the spatial correlation feature and the temporal correlation feature to obtain a fused feature. The feature concatenation operation connects the spatial correlation feature and the temporal correlation feature in the feature dimension, generating a higher-dimensional feature vector. The concatenation operation preserves the complete information of the spatial feature and the temporal feature. Specifically, during concatenation, the output feature of the spatial graph convolution and the output feature of the temporal convolution are concatenated along the channel dimension and input into the spatio-temporal feature fusion layer. Cross-modal feature fusion is achieved through 1×1 convolution. Through such transformation, the network can learn the complex relationship between the spatial feature and the temporal feature and generate a more expressive fused feature.

[0042] S324: Add the fused feature to the original vibration signal in the normalized heterogeneous data through a residual connection to obtain a spatio-temporal joint feature vector.

[0043] After obtaining the fused feature, a residual connection mechanism is also introduced in step S324. The original vibration signal is added to the fused feature to alleviate the problem of gradient disappearance and generate a spatio-temporal joint feature vector. Residual connection is a commonly used technique in deep neural networks. By adding cross-layer connections in the forward propagation, it effectively solves the problem of gradient disappearance in deep networks. The residual connection directly transmits the original signal information to the deep network, ensuring that important basic features will not be lost during multi-layer processing. At the same time, it allows the network to learn the residual mapping between the original signal and the processed feature, making it easier to optimize and train. Through the residual connection, the spatio-temporal joint feature vector contains both the basic features of the original vibration signal and the high-order features extracted through the deep network, forming a more comprehensive and robust feature representation.

[0044] S33: Perform an attention interaction calculation between the device power parameters and the spatio-temporal joint feature vector to obtain a coupling coefficient matrix, and perform abnormal region detection on the coupling coefficient matrix using a threshold segmentation method to obtain a coupling feature.

[0045] In step S33, a cross-modal cross-attention calculation is performed between the spatio-temporal joint feature vector and the power parameters (current, voltage harmonics) to generate a coupling coefficient matrix between the mechanical vibration and electrical parameters of the device; subsequently, an abnormal region detection is performed on the coupling coefficient matrix through an adaptive threshold segmentation algorithm to obtain a mechanical-electrical coupling abnormal mode identifier.

[0046] Specifically, first, attention interaction calculation is performed on the device power parameters and the spatio-temporal joint feature vector. Attention interaction calculation is a feature interaction method based on the attention mechanism, which is used to capture the mutual relationship between different feature domains. In the present invention, attention interaction calculation is used to model the association between device power parameters and vibration characteristics. The specific calculation process is as follows: First, the power parameter vector and the spatio-temporal joint feature vector are respectively mapped to the query space, the key space, and the value space through linear transformation; then, the attention weights are calculated, and the attention weights represent the attention degree of each power parameter to each vibration feature; finally, the value vectors are weighted and summed through the attention weights to obtain the attention output of the power parameters to the vibration features, and the coupling coefficient matrix is obtained by calculating the correlation between the power parameters and the attention output again, which characterizes the coupling strength between the i-th power parameter and the j-th vibration feature.

[0047] Subsequently, anomaly region detection based on the threshold segmentation method is performed on the coupling coefficient matrix, that is, the region where the coupling coefficient significantly deviates from the normal mode is found. The threshold segmentation method uses an adaptive threshold: , where is the obtained threshold, is the mean value of the coupling coefficient matrix, is the standard deviation of the coupling coefficient matrix, is an adjustable parameter, usually taken as 2. When the coupling coefficient is greater than , it is determined as an anomaly region, indicating that there is an abnormally strong coupling relationship between the corresponding power parameter and the vibration feature, which may indicate a device failure. After threshold segmentation, the coupling coefficients of the anomaly region are retained and other regions are set to zero to form a sparse coupling feature matrix, highlighting the potential abnormal patterns.

[0048] S34: Extract features from the standardized heterogeneous data to obtain cross-modal deep features.

[0049] In step S34, features are extracted from the standardized heterogeneous data to obtain cross-modal deep features. Cross-modal deep feature extraction is to extract semantically consistent and complementary feature representations from multiple different types of data. In the distribution room operation and maintenance scenario, in addition to device vibration and power parameters, it also includes multi-modal information such as temperature and humidity data, gas concentration data, and thermal imaging data.

[0050] The present invention uses a method that combines a modality-specific encoder and a shared encoder to extract features from these heterogeneous data. The modality-specific encoder designs different feature extraction networks for each data type: a multi-layer perceptron (MLP) is used to extract environmental features from temperature and humidity data; a one-dimensional convolutional neural network (1D CNN) is used to extract gas features from gas concentration data; and a two-dimensional convolutional neural network (2D CNN) is used to extract thermal distribution features from thermal imaging data. Each specific encoder maps the original data to a latent space, and the shared encoder, through techniques such as adversarial training and maximizing mutual information, maps the features of different modalities to the same semantic space. Subsequently, through adversarial training, the features of different modalities are mapped to the same distribution space. Through the combination of the modality-specific encoder and the shared encoder, the present invention extracts cross-modal deep features with consistent semantics from multi-modal data, and these features contain descriptive information about the state of the distribution room from different perspectives.

[0051] S35: Fuse the coupled features and the cross-modal deep features to obtain panoramic perception features.

[0052] In step S35, the present invention performs feature fusion on the coupled features and the cross-modal deep features. The purpose is to integrate features from different sources and different representation methods into a unified and more expressive feature representation. The feature fusion in the present invention adopts a multi-level attention fusion mechanism, including feature-level attention and decision-level attention. Feature-level attention calculates the importance weights of each feature. Through feature-level attention, different weights are assigned to each element in the coupled features and the cross-modal deep features to highlight important information; decision-level attention calculates the overall weights of the coupled features and the cross-modal deep features, and the final panoramic perception features are obtained through weighted fusion. S4: According to the panoramic perception features, perform anomaly detection through an isolation forest-long short-term memory hybrid model, and deploy a causal inference engine to analyze the causal relationship between multi-variables to obtain the evaluation result of the health state of the distribution room.

[0053] Among them, step S4 further includes: S41: Input the panoramic perception features into the isolation forest-long short-term memory hybrid model to obtain a device anomaly determination result including temporal anomalies and sudden anomalies.

[0054] In step S41, first, the panoramic perception features are input into the Isolation Forest-Long Short-Term Memory hybrid model to obtain a device anomaly determination result including temporal anomalies and sudden anomalies. The Isolation Forest-Long Short-Term Memory hybrid model provided by the present invention is an innovative anomaly detection architecture that combines the isolation-based Isolation Forest algorithm and the prediction-based Long Short-Term Memory (LSTM) network, capable of simultaneously identifying sudden anomalies and temporal anomalies in data. The hybrid model contains two parallel branches: the LSTM branch is responsible for detecting temporal anomalies, and the Isolation Forest branch is responsible for detecting sudden anomalies.

[0055] Among them, step S41 specifically includes: S411: Calculate the prediction residuals through the LSTM branch of the Isolation Forest-Long Short-Term Memory hybrid model, and mark the devices with the prediction residuals greater than the first preset threshold as anomalies to obtain a first marking result; the expression of the first preset threshold is: , where is the prediction residual, is the standard deviation of the residual sequence.

[0056] In the LSTM branch, the present invention takes the panoramic perception features of the past N time steps as input to predict the feature value of the next time step, and presets the panoramic perception features at time. For the prediction target of the LSTM network, the network is trained by minimizing the mean square error between the predicted value and the actual value. Therefore, after training is completed, the present invention makes predictions on new data and calculates the prediction residuals, that is, the prediction errors, which means the Euclidean distance between the actual observed value and the predicted value. The prediction residuals reflect the degree to which the current observed value deviates from the model prediction. Larger residual values usually indicate that the data contains abnormal patterns.

[0057] Then, by marking the devices with the prediction residuals greater than the first preset threshold as anomalies, a first marking result is obtained. The method for setting the threshold is the 3σ principle. Under the assumption of a normal distribution, the probability that the data falls outside the range of μ±3σ is only 0.3%. Therefore, the data points outside this range are regarded as anomalies. So, the present invention marks the abnormal time points as anomalies by comparing the prediction residuals at each time point with the threshold size, forming a preliminary temporal anomaly detection result, that is, the first marking result.

[0058] S412: Calculate the path length score through the Isolation Forest branch of the Isolation Forest-Long Short-Term Memory hybrid model, and mark the devices with the path length score greater than the second preset threshold as anomalies to obtain a second marking result; the expression of the second threshold is: where is the path length score.

[0059] Further, the Isolation Forest is an anomaly detection algorithm based on a tree structure. The Isolation Forest algorithm constructs a forest composed of multiple isolation trees (iTrees) and calculates the difficulty of isolating data points. The process of constructing an isolation tree is to randomly select a dimension in the feature space, randomly select a splitting point on this dimension, divide the data into two parts, and recursively execute this operation on each part after splitting until the preset tree height limit is reached or there is only one data point in the node. For a given data point, calculate the path length from the root node to the leaf node in each isolation tree, and then take the average value of all trees as the anomaly score of this point.

[0060] In the Isolation Forest algorithm, the path length score has a value range of (0, 1). The closer it is to 1, the more anomalous the data point is. Usually, data points with a path length score greater than 0.8 in the Isolation Forest algorithm are considered anomalous. By comparing the path length score of each data point with the threshold, the anomalous data points are marked as anomalous, forming the sudden anomaly detection result, that is, the second marking result.

[0061] S413: When any one of the first marking result and the second marking result marks an anomaly, the device anomaly determination result is output as anomalous. When both of the first marking result and the second marking result mark no anomaly, the device anomaly determination result is output as normal.

[0062] In step S413, as long as a data point is identified as anomalous by any branch, the final determination result is anomalous. The anomaly reporting strategy of the present invention aims to have high requirements for the sensitivity of anomaly detection. For example, for early fault warning of power distribution equipment, it is better to give false alarms than to miss potential risks.

[0063] S42: Through the Structural Equation Model, perform root cause location on the anomalous devices in the device anomaly determination result to obtain the power distribution room health status evaluation result including the device anomaly determination result and the root cause analysis result.

[0064] In step S42, the present invention performs root cause location on the anomalous devices in the device anomaly determination result through the Structural Equation Model to obtain the power distribution room health status evaluation result including the device anomaly determination result and the root cause analysis result. The Structural Equation Model (SEM) is a statistical method that can analyze the causal relationship between multiple variables, combining factor analysis and path analysis, and is suitable for modeling the complex relationship between latent variables (variables that cannot be directly observed) and observed variables.

[0065] Among them, step S42 further includes: S421: Define latent variables and observed variables, and describe the variable relationship between the latent variables and the observed variables to construct a structural equation model.

[0066] Among them, the expression of the variable relationship in step S421 is: Among them, represents the vibration RMS value, represents mechanical wear, represents the SF6 gas concentration, represents insulation deterioration, represents the partial discharge quantity, represents environmental corrosion, represents the temperature rise rate, represents the th factor loading coefficient, represents the th measurement error term.

[0067] The above-mentioned factor loading coefficient is used to quantify the explanatory power of the latent variable for the observed variable. For example, =0.85 indicates that mechanical wear explains 85% of the vibration RMS variation, and the measurement error term follows a normal distribution, representing sensor noise or unmodeled factors.

[0068] The expression of the structural equation model in step S421 is: Among them, represents the th path coefficient, represents the th structural residual term.

[0069] In step S421 of the present invention, latent variables and observed variables are first defined, the variable relationship between the latent variables and the observed variables is described, and a structural equation model is constructed. In the analysis of the equipment status in the distribution room, the latent variables are usually the root causes of equipment failures. In the present invention, mechanical wear, insulation deterioration, and environmental corrosion are taken as the causes. These latent variables are all abstract concepts that cannot be directly measured, while the observed variables are specific indicators measured by various sensors. In the present invention, the vibration RMS value, SF6 gas concentration, partial discharge quantity, and temperature rise rate are taken as specific indicators, which are physical quantities that can be directly measured.

[0070] S422: Calculate the path coefficient between variables of the structural equation model through partial least squares path analysis.

[0071] Among them, the expression of the path coefficient between variables in step S422 is:

[0072] Among them, represents the estimated path coefficient between variables, represents the total number of paths, represents the Insulation degradation caused by the th mechanical wear caused by the

[0073] The path coefficients between the variables of the structural equation model are calculated by partial least squares path analysis. Partial least squares path analysis (PLS-PA) is an estimation method that does not rely on the assumption of normal distribution and is suitable for cases with a small sample size and multicollinearity between variables. PLS-PA estimates the model parameters by maximizing the covariance between latent variables. The iterative process includes external model estimation, internal model estimation, and weight update, and finally outputs the path coefficients between variables. For example, , if

[0074] S423: According to the structural equation model and the path coefficients between the variables, perform root cause verification through the counterfactual intervention method to obtain the root cause analysis result.

[0075] The counterfactual intervention method mentioned above is a root cause analysis method based on causal reasoning. By simulating counterfactual scenarios of how the result would change if a certain factor were changed, it evaluates the causal effect of each factor on the result.

[0076] Under the framework of the structural equation model, counterfactual intervention can be achieved by adjusting specific model parameters or exogenous variables and then recalculating the model prediction values. For example, to verify whether mechanical wear is the root cause of abnormal vibration of a certain device, mechanical wear can be set to 0 in the model to represent no wear, and then the vibration RMS value is predicted. If the predicted value is significantly lower than the actual observed value, it supports that mechanical wear is the root cause of abnormal vibration. Similarly, by systematically conducting intervention experiments on each potential root cause and comparing the magnitudes of the intervention effects, the root cause factor with the greatest impact on the abnormality can be identified.

[0077] S424: Output the device abnormality determination result and the root cause analysis result to obtain the evaluation result of the health status of the distribution room.

[0078] Finally, output the device anomaly determination result and the root cause analysis result to obtain the power distribution room health status assessment result. The health status assessment result includes information such as the current device status (normal / anomalous), anomaly type (sequential anomaly / sudden anomaly), and root cause analysis result, providing a comprehensive basis for subsequent operation and maintenance decisions. For example, the final power distribution room health status assessment result is: An anomaly was detected in Circuit Breaker No. 1 (meeting the conditions of both sequential anomaly and sudden anomaly), the anomaly indicators are the vibration RMS value and the temperature rise rate, and the root cause analysis shows that mechanical wear is the main reason. Based on the main reason, it is recommended to inspect the circuit breaker drive mechanism and consider replacing key components according to the service life.

[0079] S5: Make operation and maintenance decisions based on the power distribution room health status assessment result, and perform operation and maintenance on the power distribution room through the operation and maintenance decisions.

[0080] In step S5, making operation and maintenance decisions based on the power distribution room health status assessment result and performing operation and maintenance on the power distribution room through the operation and maintenance decisions means that based on the power distribution room health status assessment result obtained from the previous steps, generate specific operation and maintenance strategies through an operation and maintenance decision algorithm, and then execute the corresponding operation and maintenance operations, which can achieve a closed loop from state perception to action execution, and transform the results of anomaly detection and root cause analysis into actionable operation and maintenance actions.

[0081] The generation of operation and maintenance decisions first assesses the risk level of the anomalous devices in the health status assessment result. According to the type of anomaly (sequential anomaly or sudden anomaly), the degree of anomaly, and the root cause analysis result, the anomalies are classified into three risk levels: emergency, important, and general. Emergency risk refers to anomalies that may cause immediate device failure or endanger the safety of the power distribution room, such as mechanical jamming of circuit breakers, short circuits in transformer windings, etc.; important risk refers to anomalies that may develop into serious faults in the short term, such as poor contact in switchgear, continuous increase in transformer oil temperature, etc.; general risk refers to anomalies that need attention but do not immediately affect device operation, such as minor vibration anomalies, slow temperature fluctuations, etc.

[0082] Next, for anomalies of different risk levels, generate corresponding operation and maintenance strategies based on a preset decision rule library. The decision rule library includes the mapping relationship between device type, anomaly mode, root cause type, and recommended operation and maintenance measures. Specific operation and maintenance measures include: immediate shutdown for maintenance, live detection, planned maintenance, enhanced status monitoring, and routine inspection, etc. For emergency risks, usually adopt the strategy of immediate shutdown for maintenance; for important risks, you can choose live detection or arrange for recent maintenance; for general risks, adopt planned maintenance or enhanced monitoring.

[0083] Then, prioritize the generated operation and maintenance strategies and allocate resources. The prioritization takes into account equipment importance, risk level, and resource availability. Equipment importance is determined based on its location and role in the power distribution system. For example, main transformers usually have the highest importance. Resource allocation includes the scheduling and allocation of maintenance personnel, spare parts, and maintenance tools to ensure the smooth execution of operation and maintenance actions.

[0084] Finally, perform specific operation and maintenance operations according to the operation and maintenance strategies, including on-site inspections, equipment testing, component replacement, parameter adjustment, etc. After the operation and maintenance are completed, it is also necessary to verify the effectiveness, check whether the abnormal state has been improved, and ensure the effectiveness of the operation and maintenance actions.

[0085] In a specific embodiment, for example, in the case of a circuit breaker mechanical failure, during the early data collection, the vibration sensor on the No. 1 circuit breaker in a 10 kV switchgear room collected that the vibration RMS value increased sharply from the normal level of 1.0 to 2.8 and lasted for 48 hours. At the same time, the temperature sensor recorded that the temperature rise rate in the operating mechanism area increased from the normal value of 1.0 to 1.3, and the current transformer detected that the tripping time extended from the standard 80 ms to 120 ms.

[0086] By using the isolation forest-long short-term memory hybrid model detection of the present invention for time series anomalies and sudden anomalies, the anomaly scores are 0.85 and 0.82 respectively, both higher than the threshold. Through the structural equation model for root cause analysis, it is determined that the main root cause is the mechanical wear of the operating mechanism, and the attribution confidence is 0.92. Specifically, it is caused by the increased friction due to the aging of the transmission bearing. The evaluation result is finally an important risk and needs to be dealt with in a timely manner, otherwise it may develop into a circuit breaker refusal to operate fault.

[0087] Based on the health status evaluation result, the operation and maintenance decision-making system outputs the following operation and maintenance strategies: arrange maintenance personnel to perform live infrared detection and acoustic detection on the No. 1 circuit breaker within 72 hours to confirm the fault location; plan to carry out a power outage repair next Tuesday (during the low load period), replace the bearings of the operating mechanism and perform mechanical lubrication; increase the monitoring frequency before the repair, and record the change trend of the tripping time every 4 hours. This strategy takes into account equipment importance, fault risk, and system operating conditions, and balances equipment safety and power supply reliability.

[0088] As Figure 2 shown, the present invention also provides an intelligent switchgear operation and maintenance system based on multi-source data fusion, including: Acquisition module 100: used to collect switchgear environment and equipment data based on multi-source sensing devices to obtain original multi-source data; Standardization module 200: used to perform standardization processing on the original multi-source data to obtain standardized heterogeneous data; Feature extraction module 300: It is used to perform spatio-temporal correlation modeling on the device vibration characteristics and device power parameters through a spatio-temporal graph convolutional network, and fuse the cross-modal deep features extracted from the standardized heterogeneous data to obtain panoramic perception features; Detection module 400: It is used to perform anomaly detection through an isolation forest-long short-term memory hybrid model according to the panoramic perception features, and deploy a causal inference engine to analyze the causal relationship between multiple variables to obtain the evaluation result of the health status of the distribution room; Decision output module 500: It is used to make operation and maintenance decisions according to the evaluation result of the health status of the distribution room, and perform operation and maintenance on the distribution room through the operation and maintenance decisions.

[0089] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0090] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0091] A smart distribution room operation and maintenance method and system based on multi-source data fusion provided by the present invention models the spatio-temporal correlation relationship between device vibration characteristics and power parameters through a spatio-temporal graph convolutional network, combines a hybrid anomaly detection model and a causal inference engine, realizes accurate identification and root cause location of distribution equipment anomalies, and has significant technical effects. The present invention can not only detect temporal anomalies and sudden anomalies at the same time, but also accurately trace the root cause of the anomalies, provides a comprehensive and accurate decision-making basis for the operation and maintenance of the distribution room, significantly improves the pertinence and effectiveness of preventive maintenance, reduces unnecessary power outage maintenance, extends the service life of the equipment, reduces the operation and maintenance cost, and improves the overall reliability and safety of the distribution system.

[0092] Finally, 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 foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart distribution room operation and maintenance method based on multi-source data fusion, characterized in that: include: S1: Collect the distribution room environment and equipment data through multi-source sensing equipment to obtain original multi-source data; S2: performing standardization processing on the original multi-source data to obtain standardized heterogeneous data; S3: The spatiotemporal correlation modeling of the equipment vibration characteristics and the equipment power parameters is performed through the spatiotemporal graph convolutional network, and the cross-modal deep features extracted from the standardized heterogeneous data are integrated to obtain the panoramic perception features; S4: Based on the panoramic perception features, anomaly detection is performed through an isolation forest-long short-term memory hybrid model, and a causal reasoning engine is deployed to analyze the causal relationship between multiple variables to obtain a health status assessment result of the distribution room; S5: Make an operation and maintenance decision according to the health status assessment result of the power distribution room, and operate and maintain the power distribution room according to the operation and maintenance decision.

2. According to claim 1, a smart distribution room operation and maintenance method based on multi-source data fusion is characterized in that: The original multi-source data in step S1 includes: Ambient temperature and humidity data, the ambient temperature and humidity data is collected by a temperature and humidity sensor; Gas concentration data, the gas concentration data is collected by a gas concentration sensor; Electric power operation parameter data, the electric power operation parameter data is collected by an electric power parameter collector; Equipment vibration characteristic data, the equipment vibration characteristic data is collected by a vibration sensor; Thermal imaging data, wherein the thermal imaging data is collected by infrared thermal imaging equipment.

3. According to claim 1, a smart distribution room operation and maintenance method based on multi-source data fusion is characterized in that: Step S2 specifically includes: Performing noise removal and outlier detection on the original multi-source data by using a data cleaning algorithm to obtain cleaned data; Normalizing the maximum and minimum values ​​of the numerical data in the cleaned data by normalization processing to obtain normalized numerical data; Transforming the skewed distribution data in the cleaned data by logarithmic transformation to obtain balanced distribution data; The time synchronization mechanism is used to align the timestamps of data with different sampling frequencies to obtain data with consistent timing. Completing the missing parts in the time-series consistent data by using a missing value filling algorithm to obtain integrity-enhanced data; The integrity enhanced data is converted into a unified format through a data format converter to obtain the standardized heterogeneous data.

4. According to claim 1, a smart distribution room operation and maintenance method based on multi-source data fusion is characterized in that: Step S3 further comprises: S31: establishing a device topology map based on the device connection relationship in the power distribution room, adjusting the edge weights of the device topology map according to the device operation status data, and obtaining a dynamic adjacency matrix; S32: Input the dynamic adjacency matrix into the spatiotemporal graph convolutional network to obtain a spatiotemporal joint feature vector; S33: performing attention interaction calculation on the device power parameter and the spatiotemporal joint feature vector to obtain a coupling coefficient matrix, and performing abnormal area detection based on a threshold segmentation method on the coupling coefficient matrix to obtain coupling features; S34: extracting features from the standardized heterogeneous data to obtain cross-modal deep features; S35: Fusing the coupling feature and the cross-modal depth feature to obtain a panoramic perception feature.

5. According to claim 4, a smart distribution room operation and maintenance method based on multi-source data fusion is characterized in that: Step S32 further includes: S321: performing spectral graph convolution on the dynamic adjacency matrix through Chebyshev polynomial approximation to extract spatial correlation features of the equipment vibration spectrum; S322: Inputting the spatial correlation features into a gated temporal convolutional network to extract the temporal correlation features of the vibration signal time series; S323: combining the spatial correlation features and the temporal correlation features to obtain fusion features; S324: adding the fused features to the original vibration signal in the standardized heterogeneous data through residual connection to obtain a spatiotemporal joint feature vector.

6. The intelligent distribution room operation and maintenance method based on multi-source data fusion according to claim 1 is characterized in that: Step S4 further comprises: S41: inputting the panoramic perception features into an isolation forest-long short-term memory hybrid model to obtain a device abnormality determination result including a timing abnormality and a sudden abnormality; S42: using a structural equation model, root cause location is performed on the abnormal equipment in the equipment abnormality determination result, and a distribution room health status assessment result including the equipment abnormality determination result and the root cause analysis result is obtained.

7. The intelligent distribution room operation and maintenance method based on multi-source data fusion according to claim 6 is characterized in that: Step S41 specifically includes: S411: Calculate the prediction residual by using the LSTM branch of the isolation forest-long short-term memory hybrid model, and mark the device whose prediction residual is greater than a first preset threshold as abnormal, to obtain a first marking result; The expression of the first preset threshold is: ,in is the prediction residual, is the standard deviation of the residual series; S412: Calculate the path length score by using the isolation forest branch of the isolation forest-long short-term memory hybrid model, and mark the device whose path length score is greater than the second preset threshold as abnormal to obtain a second marking result; The expression of the second threshold is: ,in Score the path length; S413: When any one of the first marking result and the second marking result is marked as abnormal, the device abnormality determination result is output as abnormal; when any one of the first marking result and the second marking result is marked as no abnormality, the device abnormality determination result is output as normal.

8. The intelligent distribution room operation and maintenance method based on multi-source data fusion according to claim 6 is characterized in that: Step S42 further includes: S421: Define latent variables and observed variables, describe the variable relationship between latent variables and observed variables, and construct a structural equation model; S422: calculating the path coefficients between the variables of the structural equation model by partial least squares path analysis; S423: According to the structural equation model and the path coefficients between the variables, root cause verification is performed through a counterfactual intervention method to obtain root cause analysis results; S424: Output the equipment abnormality determination result and the root cause analysis result to obtain a health status assessment result of the power distribution room.

9. The intelligent distribution room operation and maintenance method based on multi-source data fusion according to claim 8 is characterized in that: The expression of the variable relationship in step S421 is: ； in, Indicates the vibration RMS value, Indicates mechanical wear, Indicates the SF6 gas concentration, Indicates insulation degradation, Indicates the amount of partial discharge, Indicates environmental corrosion, represents the rate of temperature rise, Indicates Factor loading coefficients, Indicates measurement error terms; The expression of the structural equation model in step S421 is: ; in, Indicates The path coefficients, Indicates structural residual terms; The expression of the path coefficient between the variables in step S422 is: ;in, represents the estimated path coefficient between variables, Represents the total number of paths, Indicates Insulation degradation caused by the Indicates Mechanical wear caused by the following paths, Indicates Environmental corrosion caused by a path.

10. A smart distribution room operation and maintenance system based on multi-source data fusion, characterized in that: include: Acquisition module: used to collect distribution room environment and equipment data based on multi-source sensor equipment to obtain original multi-source data; Standardization module: used to perform standardization processing on the original multi-source data to obtain standardized heterogeneous data; Feature extraction module: used to perform spatiotemporal correlation modeling of equipment vibration characteristics and equipment power parameters through a spatiotemporal graph convolutional network, and to fuse the cross-modal deep features extracted from the standardized heterogeneous data to obtain panoramic perception features; Detection module: used to perform anomaly detection through an isolation forest-long short-term memory hybrid model based on the panoramic perception features, and deploy a causal reasoning engine to analyze the causal relationship between multiple variables to obtain a health status assessment result of the distribution room; Decision output module: used to make operation and maintenance decisions according to the health status assessment results of the distribution room, and operate and maintain the distribution room according to the operation and maintenance decisions.

Images