Intelligent power distribution operation and maintenance management system based on 5G transmission

Abstract

The invention relates to the technical field of power distribution operation and maintenance management, and discloses an intelligent power distribution operation and maintenance management system based on 5G transmission. The system comprises a 5G real-time acquisition module, a multi-dimensional feature fusion module, a dynamic topology generation module, an anomaly propagation analysis module and a strategy optimization feedback module. The 5G real-time acquisition module acquires operation state data streams such as current and voltage waveforms, an equipment temperature sequence and environment monitoring indexes of the power distribution equipment through a 5G network; the multi-dimensional feature fusion module is used for separating equipment state features, calculating mutual information amount and generating equipment state feature tensors; the dynamic topology generation module constructs an association intensity matrix according to the feature tensor, and generates a hierarchical connection path and a dynamic equipment topological graph; the abnormal propagation analysis module extracts a state fluctuation sequence, identifies an abnormal transmission path and marks a core propagation node; and a strategy optimization feedback module generates a maintenance strategy priority queue according to the dynamic topology map, and feeds back an execution result to update the dynamic topology map, so that the intelligence and accuracy of power distribution operation and maintenance management are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution operation and maintenance management, in particular to an intelligent power distribution operation and maintenance management system based on 5G transmission. BACKGROUND

[0002] In the operation process of the power system, as the key link connecting the power generation side and the user side, the stable operation of the power distribution network is directly related to the reliability and safety of power supply. With the continuous growth of social electricity demand and the increasing proportion of new energy access, the structure of the power distribution network is becoming increasingly complex, the number of equipment is increasing significantly, and the traditional power distribution operation and maintenance management mode has gradually exposed many problems. Under the traditional mode, the collection of the operation state of the power distribution equipment mainly depends on wired transmission mode or low-rate wireless transmission technology. The wired transmission mode not only has high construction cost and long construction period, but also is difficult to achieve full coverage in some complex terrains or old urban areas, resulting in that the equipment state data in some areas cannot be obtained in time. The low-rate wireless transmission technology is limited by bandwidth and cannot transmit large-capacity and high-real-time data such as current and voltage waveforms, so that the operation and maintenance personnel cannot accurately master the real-time operation state of the equipment. In addition, the traditional state monitoring usually only focuses on single-dimensional indicators, such as only monitoring the temperature or current of the equipment, ignoring the correlation between different dimensional data, and cannot fully reflect the overall operation state of the equipment, which is easy to cause misjudgment or omission. In terms of power distribution network topology management, the traditional method usually adopts a fixed topology structure for management, which cannot be adjusted in real time according to the changes of the equipment operation state. When the equipment in the power distribution network has state fluctuations or faults, the fixed topology cannot quickly and accurately present the correlation between the equipment, so that the operation and maintenance personnel cannot timely identify the propagation path and core influence node of the fault, prolonging the time of fault troubleshooting and processing. At the same time, when formulating the equipment maintenance strategy, the traditional mode is mostly based on experience or fixed period, lacking dynamic evaluation of the real-time state and fault risk of the equipment, which is easy to cause problems of over-maintenance or insufficient maintenance. Over-maintenance will increase the operation and maintenance cost, while insufficient maintenance may cause frequent equipment faults, affecting the stable operation of the power distribution network. With the development of the power system towards intelligence and digitization, the traditional power distribution operation and maintenance management mode cannot meet the needs of the new era, and an intelligent operation and maintenance management system capable of realizing high real-time data collection, multi-dimensional data fusion analysis, dynamic topology management and precise maintenance strategy formulation is urgently needed. SUMMARY

[0003] The purpose of the present application is to provide an intelligent power distribution operation and maintenance management system based on 5G transmission to solve the problems raised in the background.

[0004] In order to achieve the above object, the application provides an intelligent power distribution operation and maintenance management system based on 5G transmission, which comprises: The 5G real-time acquisition module obtains the running state data stream of the power distribution equipment through the 5G network, and the running state data stream comprises current and voltage waveforms, equipment temperature sequences and environmental monitoring indexes; The multi-dimensional feature fusion module separates the equipment state features of different dimensions based on the running state data stream, calculates the mutual information between the features of each dimension, and integrates to generate an equipment state feature tensor; The dynamic topology generation module constructs the correlation strength matrix between the nodes of the power distribution equipment according to the equipment state feature tensor, generates the equipment hierarchical connection path according to the correlation strength threshold, and forms a dynamic equipment topology map; The abnormal propagation analysis module extracts the state fluctuation sequence between the correlated equipment nodes based on the dynamic equipment topology map, identifies the abnormal fluctuation transmission path and marks the core propagation node; The strategy optimization feedback module generates an equipment maintenance strategy priority queue according to the core propagation node and its abnormal fluctuation transmission path, and updates the dynamic equipment topology map with the execution result as feedback input.

[0005] Preferably, the multi-dimensional feature fusion module comprises: The spectral feature analysis submodule performs multi-scale decomposition on the current and voltage waveforms, extracts the energy distribution ratio of the fundamental frequency component and the harmonic component, and generates a current and voltage spectral feature vector; The temperature trend modeling submodule divides the equipment temperature sequence into time window segments, calculates the temperature change rate and extreme point position in each window, and outputs an equipment temperature trend feature vector; The environmental correlation quantization submodule calculates the contribution weight of environmental factors to the equipment state based on the time shift correlation between the environmental monitoring indexes and the equipment temperature sequence; The feature tensor integration submodule concatenates the current and voltage spectral feature vector, the equipment temperature trend feature vector and the environmental factor contribution weight, and outputs the equipment state feature tensor after normalization.

[0006] Preferably, the dynamic topology generation module comprises: The correlation strength calculation submodule calculates the cosine similarity and dynamic time warping distance of the feature vectors of any two power distribution equipment based on the equipment state feature tensor, and generates the equipment correlation strength value by weighting; The path generation submodule traverses all the equipment correlation strength values, filters the correlation strength values exceeding the preset threshold, and constructs the directed connection link between the equipment nodes in descending order of strength value; The topology iteration submodule performs closed loop detection on the directed connection link, generates a hierarchical device connection path after eliminating redundant links, and outputs a node relationship matrix of a dynamic device topology graph.

[0007] Preferably, the anomaly propagation analysis module comprises: The fluctuation pattern extraction submodule collects the state feature variation of each device node in a continuous time slice based on the node relationship matrix of the dynamic device topology graph, and marks the feature offset that exceeds the normal fluctuation range; The propagation path tracking submodule traces back the abnormal fluctuation source node and records the sequence of the transmission path nodes according to the occurrence time sequence of the feature offset and the direction of the device connection path; The core node identification submodule counts the triggered frequency and path coverage of each node in the transmission path node sequence, and screens the nodes with high frequency and coverage as core propagation nodes.

[0008] Preferably, the strategy optimization feedback module comprises: The strategy priority generation submodule calculates the device maintenance urgency score according to the device type of the core propagation node and the length of the abnormal fluctuation transmission path, and generates a maintenance strategy priority queue in order of score; The feedback mapping submodule collects the device state feature tensor variation after the execution of the maintenance strategy, quantifies the abnormal fluctuation elimination rate and maps it to the corresponding device node; The topology update submodule adjusts the correlation strength value between device nodes based on the abnormal fluctuation elimination rate, and updates the dynamic device topology graph after reconstructing the device connection path.

[0009] Preferably, the system further comprises a fault prediction module: The fault prediction module extracts the state feature attenuation trend of the key device nodes based on the updated dynamic device topology graph, predicts the remaining effective operation time of the device, and outputs a life warning list.

[0010] Preferably, the system further comprises a multi-stage optimization module: The multi-stage optimization module cross- validates the life warning list and the maintenance strategy priority queue, and marks a set of device nodes that need to be processed in priority; The multi-stage optimization module corrects the attenuation rate calculation model of the device state feature tensor based on the maintenance records of the set of device nodes.

[0011] Preferably, the system further comprises a resource scheduling module: The resource scheduling module generates an optimal path scheduling scheme based on the geographical location distribution of the devices in the maintenance strategy priority queue, combined with the location information of the maintenance personnel.

[0012] Preferably, the system further comprises a digital twin mapping module: The digital twin mapping module maps the dynamic device topology atlas and the real-time device state feature tensor to a three-dimensional power grid model, generating a power distribution device operation state heat map.

[0013] Preferably, the system further comprises a decision verification module: The decision verification module compares the power distribution device operation state heat map with historical operation and maintenance records, verifies the execution effectiveness of the maintenance strategy priority queue, and outputs a deviation report.

[0014] Compared with the prior art, the present application has the following advantages: The intelligent power distribution operation and maintenance management system based on 5G transmission acquires the operation state data stream of the power distribution device through the 5G real-time acquisition module using the 5G network, including current and voltage waveforms, device temperature sequences, and environmental monitoring indicators. The 5G network has the characteristics of high bandwidth, low latency, and wide connectivity, which can break through the limitations of traditional transmission methods and realize high-speed and real-time acquisition of power distribution device operation state data, ensuring that operation and maintenance personnel can timely and accurately grasp the real-time operation of the device, avoiding the deviation in judging the device state caused by incomplete or delayed data transmission. The multi-dimensional feature fusion module separates the device state features of different dimensions based on the acquired operation state data stream and calculates the mutual information between the features of each dimension, and integrates to generate a device state feature tensor. This module breaks through the limitations of traditional single-dimensional monitoring, fully considers the correlation between different dimensional device state data, and through deep fusion analysis of multi-dimensional data, can more comprehensively and objectively reflect the overall operation state of the device, allowing operation and maintenance personnel to no longer be limited to single-index judgment, but to understand the device operation condition from the overall level, reducing the misjudgment or omission caused by one-sided data, and providing a more comprehensive basis for subsequent device state evaluation and fault analysis. The dynamic topology generation module constructs the correlation strength matrix between power distribution device nodes according to the device state feature tensor, generates device hierarchical connection paths according to the correlation strength threshold, and forms a dynamic device topology atlas. Unlike traditional fixed topology management methods, this module can adjust the topology structure in real time according to the changes in device operation state, clearly presenting the correlation between devices as the state changes. When the device state fluctuates, the dynamic topology atlas can quickly reflect the changes in the correlation strength between device nodes, helping operation and maintenance personnel to intuitively understand the mutual influence between devices and laying a foundation for subsequent abnormal propagation analysis. The abnormal propagation analysis module extracts the state fluctuation sequence between the associated device nodes based on the dynamic device topology map, identifies the abnormal fluctuation transmission path and marks the core propagation node. With the help of the device association relationship presented by the dynamic topology map, this module can accurately track the propagation trajectory of abnormal fluctuations in the power distribution network and accurately lock the core node that plays a key role in abnormal propagation. Through the information obtained by this module, the operation and maintenance personnel can quickly locate the abnormal source and the scope of influence, avoid blind searching when troubleshooting, and greatly shorten the preparation time for abnormal processing. The strategy optimization feedback module generates a device maintenance strategy priority queue according to the core propagation node and its abnormal fluctuation transmission path, and updates the dynamic device topology map with the execution result as feedback input. This module makes the formulation of the device maintenance strategy no longer dependent on experience or fixed period, but based on the real-time abnormal situation and association relationship of the device, realizing the dynamic optimization of the maintenance strategy. By generating a priority queue, the operation and maintenance personnel can clearly determine the focus and order of the maintenance work, reasonably allocate operation and maintenance resources, avoid wasting resources on non-critical devices, and also ensure that critical devices are maintained in a timely manner. In addition, the maintenance execution result is fed back to update the dynamic topology map, which can keep the topology map consistent with the actual running state and maintenance situation of the device at all times, further improving the accuracy of subsequent topology analysis, abnormal identification and maintenance strategy formulation, forming a closed-loop operation and maintenance management process, continuously improving the operation efficiency and reliability of the entire power distribution operation and maintenance management system, and ensuring the stable operation of the power distribution network. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 a timing diagram of the intelligent power distribution operation and maintenance management system based on 5G transmission according to the present application; Figure 2 a flowchart of the workflow of the multi-dimensional feature fusion module; Figure 3 a flowchart of the workflow of the abnormal propagation analysis module; Figure 4 a flowchart of the workflow of the fault prediction module. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0017] Please refer to Figure 1 The present application provides an intelligent power distribution operation and maintenance management system based on 5G transmission, which comprises: The running state data stream of the power distribution equipment is acquired by the 5G real-time acquisition module, including current and voltage waveforms, equipment temperature sequences, and environmental monitoring indicators. The multi-dimensional feature fusion module separates the device state features of different dimensions based on the running state data stream, calculates the mutual information between the dimensions, and integrates to generate a device state feature tensor. The dynamic topology generation module constructs the correlation strength matrix between the nodes of the power distribution equipment according to the device state feature tensor, generates the device hierarchical connection path according to the correlation strength threshold, and forms a dynamic device topology map. The abnormal propagation analysis module extracts the state fluctuation sequence between the associated device nodes based on the dynamic device topology map, identifies the abnormal fluctuation transmission path and marks the core propagation nodes. The strategy optimization feedback module generates a device maintenance strategy priority queue according to the core propagation nodes and their abnormal fluctuation transmission path, and updates the dynamic device topology map with the execution results as feedback input.

[0018] Embodiment 1: refer to Figure 2 The multi-dimensional feature fusion module converts the raw data stream from the 5G real-time acquisition module into normalized tensor data that can fully represent the device state. The current and voltage waveforms received by the system are high-frequency sampled time series data, continuously transmitted at millisecond intervals. These waveform data are processed by multi-scale decomposition, and the wavelet transform method is used to decompose the signal into different frequency subbands. Each subband corresponds to a frequency range, and the energy proportion of each subband is calculated to form a vector representing the fundamental frequency and harmonic distribution characteristics. This vector can reflect the power quality status and harmonic pollution level, for example, a decrease in fundamental frequency energy proportion may indicate the presence of a nonlinear load, and an increase in a specific harmonic energy may indicate a device fault.

[0019] The processing of the equipment temperature sequence uses a sliding time window mechanism, and the window length is dynamically adjusted according to the device type and running history. For devices prone to overheating, a shorter window is used to capture rapid changes, and for devices with large thermal inertia, a longer analysis window is used. Within each window, the temperature change rate is calculated by difference operation between adjacent sampling points, and local extreme point positions are detected. These features together form a temperature trend feature vector that can reflect the heating law and heat dissipation characteristics of the device. For example, a continuously rising temperature change rate may indicate insulation aging, and an abnormal temperature extreme value distribution may indicate a local overheating problem.

[0020] The environmental monitoring indicators include humidity, dust concentration, and wind speed, and other multi-dimensional parameters. There is a complex correlation between these parameters and the temperature sequence of the equipment. By calculating the time-shift mutual information of the environmental indicators and the temperature data, the influence degree and the lag effect of different environmental factors on the equipment temperature are quantified. For example, the change in humidity may have a significant impact on the equipment temperature after several hours, and this time-shift characteristic is captured through mutual information calculation. The contribution weight calculation adopts a multivariate regression analysis method to determine the influence weight of each environmental factor on the equipment state, forming an environmental correlation feature vector.

[0021] The feature tensor integration process multi-dimensionally splices the above three feature vectors. The current and voltage spectrum feature vector contains the energy distribution information of multiple frequency components, the equipment temperature trend feature vector contains the change rate and extreme point features, and the environmental factor contribution weight vector represents the influence degree of the external environment. The multi-dimensional array after splicing is normalized by using the minimum-maximum scaling method to map each feature dimension to a unified numerical range, eliminating the magnitude difference caused by different physical dimensions. The final output of the equipment state feature tensor is a structured multi-dimensional data object that can reflect the electrical characteristics, thermodynamic characteristics, and environmental correlation characteristics of the equipment at the same time. This tensor serves as a unified equipment state representation, providing a standardized input for subsequent topology construction and anomaly analysis.

[0022] The entire processing process adopts a pipeline architecture, and each sub-module is executed sequentially but supports parallel processing. The spectrum feature analysis submodule deploys digital signal processing dedicated hardware acceleration wavelet transform calculation, the temperature trend modeling submodule uses a sliding window buffer mechanism to realize real-time temperature feature extraction, and the environmental correlation quantization submodule maintains a time series database to store historical environmental data for time-shift correlation calculation. The feature tensor integration submodule realizes the alignment and splicing of feature vectors, ensuring that the features from different sources are strictly synchronized in time stamp.

[0023] When the system is running, the multi-dimensional feature fusion module continuously receives real-time data streams transmitted by the 5G network, and outputs the equipment state feature tensor at fixed time intervals. Each output tensor is attached with complete timestamp information and equipment identification information, ensuring that the subsequent processing steps can accurately trace the data source. The tensor data structure design supports flexible expansion, and can accommodate new feature dimensions without changing the overall framework. This implementation enables the system to adapt to different scales and types of power distribution networks, providing a unified and comprehensive equipment state representation basis for intelligent operation and maintenance management.

[0024] Embodiment 2: see Figure 3, the implementation process of the dynamic topology generation module starts with the input of the device state feature tensor. The module processes the structured tensor data output from the multi-dimensional feature fusion module, which contains the standardized state features of all devices in the power distribution network. The correlation strength calculation submodule extracts the corresponding feature vectors for each pair of power distribution device nodes for similarity analysis. Cosine similarity calculation is used to evaluate the similarity of feature vectors in spatial distribution, reflecting the static correlation characteristics of device state. At the same time, dynamic time warping distance calculation is used to measure the similarity of device state change patterns in the time dimension. This calculation can handle the time offset problem between different devices. The two measurement results are fused through a pre-set weighting coefficient to generate a comprehensive device correlation strength value. For example, when two transformers show harmonic feature growth patterns in the same time interval and the temperature change curves are highly consistent, a higher correlation strength score will be obtained.

[0025] The path generation submodule receives the correlation strength calculation results of all device pairs, and the system pre-sets a dynamically adjusted strength threshold for filtering significant correlation relationships, which is determined based on historical operation data statistical analysis. Device correlation relationships exceeding the threshold are retained and sorted from high to low according to the strength value. Each valid correlation relationship is converted into a directed connection link, and the link direction is determined by the time sequence of state changes. For example, when the voltage fluctuation of substation A always occurs before the current anomaly of distribution cabinet B, the system establishes a directed link from A to B. These links form the initial framework of the device correlation network.

[0026] The topology iteration submodule processes the initial link set, which first detects closed-loop structures in the network, such as when a ring path A→B→C→A appears, the system identifies the closed loop and removes the redundant link with the lowest strength. Then, the minimum spanning tree algorithm is applied to delete secondary connections while ensuring network connectivity, forming a tree-like structure without loops. Finally, the device nodes are hierarchically divided according to the correlation strength values, with high strength values placed in the upper layer nodes. For example, regional substations are placed at the top of the topology due to their wide range of state change influence, while end distribution boxes are placed at the bottom. After processing, the node relationship matrix is output, which clearly records the hierarchical position of each device node and its upstream and downstream connection relationships.

[0027] The abnormal propagation analysis module works based on the dynamic device topology map, and the fluctuation pattern extraction submodule continuously monitors the state feature changes of each device node. The system sets a normal fluctuation range for each feature dimension, which is obtained by analyzing historical device operation data. When a feature value of a device node exceeds the normal range in consecutive time slices, it is marked as a feature offset. For example, if the temperature feature of a switchgear exceeds the historical mean value by three standard deviations for three consecutive 5-minute time slices, it is identified as an effective offset event.

[0028] The propagation path tracking submodule is started immediately after detecting a feature deviation event. The submodule first determines the time point of the deviation event, and then reverses the direction of the directed connection in the topology graph. The system traces back to the upstream nodes along the device connection path, checks the state records of each upstream node at an earlier time slice. When a certain upstream node is found to have the same type of feature deviation and the occurrence time is earlier than the current node, it is marked as a potential source. For example, when the power distribution transformer B has a current harmonic deviation, the system traces back and finds that the upper-level substation A has a similar deviation 15 minutes ago, so A is determined to be the abnormal source, and the transmission path is recorded as A→B.

[0029] The core node identification submodule statistically analyzes the historical propagation path data. The system maintains all identified abnormal propagation path records and counts the frequency of each node appearing in the propagation path. At the same time, the path coverage index of each node is calculated, which reflects the degree of the node as a hub in the abnormal propagation network. For example, a regional tie switch appears in 32 abnormal propagation paths in three months, and its downstream influence range covers 15% of the device nodes in the network. The system sets a frequency threshold and a coverage threshold. When a node exceeds both thresholds, it is marked as a core propagation node. These core nodes usually correspond to key connection points or vulnerable devices in the power distribution network.

[0030] The dynamic topology generation task is executed by a dedicated graph computing engine, and the abnormal propagation analysis is implemented in real time using a stream processing framework. The system maintains a state feature sliding window memory for each device node to ensure fast access to historical state data. The node relationship matrix and abnormal propagation records are stored in a time series database, supporting efficient time range queries and pattern analysis operations. This implementation enables the system to adapt to real-time analysis requirements of large-scale power distribution networks, providing dynamic updated network relationship models and abnormal propagation features for operation and maintenance decisions.

[0031] Embodiment 3: refer to Figure 4 The implementation process of the strategy optimization feedback module is based on the core propagation nodes and their abnormal fluctuation transmission paths output by the abnormal propagation analysis module. The strategy priority generation submodule first classifies each core propagation node by device type, and different types of devices have different fault influence coefficients. At the same time, the length of each abnormal fluctuation transmission path is calculated, i.e. the number of device nodes passed from the source node to the end node. The device maintenance urgency score is calculated by the following formula:

[0032] Wherein: represents the maintenance urgency score of the i-th core propagation node, represents the device type weight coefficient, represents the length of the abnormal fluctuation transmission path, Indicates the abnormal trigger frequency of the node within the statistical period. 、 、 The values ​​are the adjustment coefficients for each parameter. After the scoring is calculated, the system sorts the scores from high to low and generates a maintenance strategy priority queue. This queue not only includes device identification and priority order, but also records the corresponding abnormality type and maintenance recommendations for each device.

[0033] The feedback mapping submodule begins after the maintenance strategy is executed. It continuously collects feature tensors of the device's post-maintenance state and compares them with the baseline tensor before maintenance. The system calculates the change in each feature dimension, paying particular attention to those that previously exhibited abnormal fluctuations. The elimination rate of abnormal fluctuations is calculated as the reduction in feature offsets. This elimination rate is mapped to the corresponding device node, and the timestamp and details of the maintenance operation are recorded.

[0034] The topology update submodule adjusts device associations based on the feedback mapping results. The system proportionally adjusts the association strength between the device node and its adjacent nodes based on the abnormal fluctuation elimination rate. Device nodes with high elimination rates have correspondingly weakened association strengths with upstream nodes, indicating that maintenance has effectively reduced the risk of abnormality propagation. The association strength adjustment formula is: ,in represents the updated association strength, represents the original association strength, represents the abnormal fluctuation elimination rate, After the strength value is updated, the system re-executes the device connection path generation process and reconstructs the entire dynamic device topology map.

[0035] The fault prediction module works based on the updated topology map. The module first identifies the key equipment nodes in the network. The selection criteria include the central position of the node in the topology, the frequency of historical anomalies, and maintenance records. For each key node, the system extracts the time series data of its state feature tensor and analyzes the changing trends of each feature dimension. In particular, for features that are sensitive to equipment life, such as the temperature accumulation effect of insulating materials and the number of operation records of mechanical components, a time series fitting method is used to establish an attenuation model. The prediction of the remaining effective operating time of the equipment is based on the time point when the characteristic value reaches the preset failure threshold, while taking into account the influence of the equipment operating environment and load conditions. The final output life warning list contains equipment identification, predicted failure time, confidence assessment, and a list of key attenuation features.

[0036] During the system implementation, the strategy optimization feedback module and the fault prediction module form a closed-loop cooperation. After each maintenance strategy execution, the feedback data is not only used to update the topology relationship, but also used to calibrate the fault prediction model. The life warning list and the maintenance strategy priority queue are cross-validated, when a certain device appears in both the high-priority maintenance queue and the short-term life warning list, the system will upgrade its maintenance urgency. All maintenance records and prediction results are stored in the historical database, which is used to optimize the device state feature degradation rate calculation model, and continuously improve the prediction accuracy.

[0037] The entire implementation architecture adopts micro-service design, and the strategy optimization feedback module and the fault prediction module run as independent services, and exchange data through the message queue. The maintenance strategy priority queue is pushed to the operation and maintenance management platform, and the life warning list is sent to the scheduling system at the same time. The system establishes a digital file for each device, complete records its state feature history, maintenance records and prediction data, and provides technical support for the whole life cycle management of power distribution equipment.

[0038] Example 4: The multi-stage optimization module receives the life warning list from the fault prediction module and the maintenance strategy priority queue from the strategy optimization feedback module. The module establishes a device cross-analysis matrix, and matches the device identifiers in the two lists. For the device nodes that exist in both lists, the system extracts their maintenance urgency scores and predicted remaining life data, and sets double judgment conditions to mark the urgency. When the device meets both conditions of maintenance score higher than threshold S_th and remaining life lower than threshold T_risk, it is marked as red emergency state; when only one condition is met, it is marked as yellow attention state. For example, a 110kV oil-immersed transformer has a score of 92 points (full score 100) in the priority queue, and the life warning shows that the remaining operation time is only 45 days, the system classifies it as a red emergency device. The final generated device node set that needs to be processed in priority contains device identifier, urgency level and recommended processing time window.

[0039] Table 1: Device node urgency marking logic table.

[0040]

[0041] The module synchronously accesses the equipment maintenance history database to extract the complete maintenance records of the marked equipment. Each record contains fields such as maintenance time, operation type, replaced parts, and comparison of characteristic values before and after maintenance. The system establishes a characteristic degradation model for each type of equipment, such as for the spring fatigue characteristic of a circuit breaker mechanism, the model contains the corresponding relationship between operation times and spring pressure degradation. When a new maintenance record is stored, the system compares the actual maintenance effect of the equipment with the predicted value of the model. The actual improvement rate of the contact wear characteristic value of a certain vacuum circuit breaker after maintenance is 78%, while the model predicts a value of 85%, and the system starts the model parameter calibration program. The incremental learning mechanism is used to modify the attenuation rate, based on the gradient of the characteristic tensor before and after maintenance, to dynamically adjust the characteristic degradation coefficient of this type of equipment. The modified model will be applied to the life prediction calculation of subsequent devices of the same type.

[0042] The resource scheduling module is based on the priority processing device node set. This module connects the device asset management system to obtain device geographic information data, including substation coordinates, distribution room locations, tower GPS points, etc. At the same time, it accesses the real-time positioning data of the mobile terminal of the operation and maintenance personnel and obtains personnel skill qualification information. The system establishes a device-personnel matching matrix, first filters maintenance teams with the qualifications to handle equipment faults, and then calculates the optimal path from the current location of each team to the set of devices to be processed.

[0043] The path planning uses an improved ant colony algorithm, considering the following constraints: device processing urgency level (red devices need to be reached first), device geographic location distribution density (regional clustered devices for batch processing), personnel skill matching degree (specialized devices are preferentially assigned), and estimated single-point operation time. The system optimizes the calculation with the dual goals of shortest total distance and highest coverage rate of urgent devices. For example, a certain dispatch involves 17 devices (including 3 red urgent devices) distributed within an 8-kilometer radius, and the system generates inspection paths for two maintenance teams: Team A path: current location of personnel → red device S-101 (distance 3.2 km) → yellow device D-205 (1.8 km) → red device T-307 (4.1 km) Team B path: current location of personnel → red device R-208 (2.7 km) → clustered device group {D-211, D-215, D-219} (region radius 500 m) → yellow device S-104 (3.3 km) The generated scheduling scheme contains detailed navigation path, estimated arrival time, job time budget and equipment processing sequence. The scheme is pushed to maintenance personnel through mobile APP, and the execution progress is monitored in real time. When the personnel arrive at the equipment location, the system automatically records the actual arrival time; when the equipment state returns to normal, the maintenance personnel feed back the completion information through the APP. These execution data are returned to the multi-stage optimization module for verifying the accuracy of the equipment emergency degree marking and serving as a reference basis for subsequent maintenance strategy adjustment.

[0044] The entire implementation process adopts a distributed architecture, and the multi-stage optimization module runs on a data analysis server, and the resource scheduling module is deployed on a cloud computing platform. The equipment node set update trigger mechanism is set to be executed automatically every hour, and the real-time scheduling is triggered immediately when the emergency red equipment is found. The system maintains the equipment geospatial database, records the equipment position change history; the personnel position data is updated every 30 seconds to ensure the timeliness of the scheduling scheme. The scheduling scheme generation engine sets a timeout control mechanism, which automatically outputs the current optimal solution when the calculation exceeds the preset time, ensuring the timeliness of operation and maintenance response.

[0045] Example 5: The implementation of the digital twin mapping module is based on dynamic equipment topology map and real-time equipment state feature tensor. The module first imports the three-dimensional geographic information system model of the power distribution network, which contains detailed three-dimensional models of substation buildings, power distribution rooms, power transmission line corridors, tower equipment, etc. Each three-dimensional model object has a unique equipment identification code, which is mapped to the equipment information in the real-time database. The system accurately matches the equipment logical node with the three-dimensional space entity through coordinate conversion algorithm, ensuring the position consistency of the virtual model and the physical equipment.

[0046] Each dimension of data in the equipment state feature tensor is mapped to a different visualization channel of the three-dimensional model. The current voltage spectrum features are presented through the color temperature change of the equipment shell, and the normal state displays cold tone, and the abnormal state gradually changes to warm tone according to the severity. The temperature trend feature adopts the heat map rendering technology to generate a temperature distribution gradient on the surface of the equipment, and the high temperature area presents red gradient, and the low temperature area displays blue gradient. The environmental monitoring indicators are visualized through environmental special effects, such as humidity indicators are presented through spatial fog density, and dust concentration is displayed through particle effect. These visualization effects together constitute the power distribution equipment operation state heat map, and the operator can intuitively master the whole network equipment operation condition through the three-dimensional perspective.

[0047] The decision verification module runs continuously in the background, comparing the real-time heat map with a database of historical maintenance records. This database stores a snapshot of all maintenance events, including a complete record of the equipment's status before maintenance, the maintenance measures taken, and the changes in status after maintenance. The system uses an image feature comparison algorithm to extract feature vectors from abnormal areas in the current heat map and compare them with abnormal patterns in historical records. If a current abnormal pattern is found to be highly similar to a historical event, the system automatically retrieves the handling records and effectiveness evaluation data for that event.

[0048] The effectiveness of the maintenance policy priority queue is verified through time series comparison. The system records the policy execution time, captures heat map data for specific time windows before and after execution, and calculates quantitative indicators such as the rate of change in the area of ​​abnormal areas and the change in color intensity. For example, after a maintenance operation on an overloaded transformer, the system detected a 62% reduction in the red area of ​​the device in the heat map, and the color intensity dropped from the highest warning level to the normal range. The system also analyzes the status transmission of the device's associated lines to confirm that the abnormal fluctuation has not spread to upstream and downstream equipment.

[0049] The deviation report generation mechanism is based on multi-dimensional data analysis. The system compares the differences between the expected and actual effects of maintenance strategies, identifying three types of effects: significant, average, and poor. For operations with significant effects, the system records the characteristic patterns of the maintenance measures and stores them in the best practice library. For operations with poor results, the system analyzes possible causes, such as delayed maintenance, incomplete measures, and unsynchronized processing of related equipment. The report output uses a structured format and includes sections such as deviation description, possible cause analysis, and improvement suggestions. Each report is associated with a specific device identifier and maintenance event number, supporting historical tracing and effect review.

[0050] During system implementation, a digital twin model update mechanism was established. Whenever physical equipment changes occur, such as equipment replacement or line modification, the 3D model is updated with version information. Real-time data stream processing utilizes a distributed architecture to ensure highly concurrent visualization of large-scale device status data. Historical data retrieval optimizes the index structure of the time series database, supporting rapid query and comparison. Visualization configuration utilizes a parametric design, allowing users to adjust the display weighting and presentation of different features based on their focus.

[0051] The entire digital twin system supports multi-dimensional viewing, allowing operators to freely switch between viewing modes, including global overview, regional focus, and detailed inspection of individual devices. The system records user operation logs and analysis paths, providing data support for subsequent optimization of the human-computer interaction experience. All visualizations are based on physically accurate rendering technology, ensuring that visual elements such as color, shape, and position truly reflect the actual device status, providing intuitive and reliable visual basis for operation and maintenance decisions.

[0052] It has to be noted that, in the present document, the terms "first", "second", etc. merely serve to identify different entities or actions from each other, without necessarily requiring or implying any actual relationship or order between these entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0053] While embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, changes, and variations can be made in the embodiments without departing from the spirit and scope of the present application, which is defined by the appended claims and their equivalents.

Claims

1. A smart power distribution operation and maintenance management system based on 5G transmission, characterized in that, The system comprises: The 5G real-time acquisition module acquires the running state data stream of the power distribution equipment through the 5G network, and the running state data stream comprises current and voltage waveforms, equipment temperature sequences and environmental monitoring indexes; The multi-dimensional feature fusion module separates the equipment state features of different dimensions based on the running state data stream, calculates the mutual information between the features of each dimension, and integrates to generate an equipment state feature tensor; The dynamic topology generation module constructs the correlation strength matrix between the nodes of the power distribution equipment according to the equipment state feature tensor, generates the equipment hierarchical connection path according to the correlation strength threshold, and forms a dynamic equipment topology map; The abnormal propagation analysis module extracts the state fluctuation sequence between the associated equipment nodes based on the dynamic equipment topology map, identifies the abnormal fluctuation transmission path and marks the core propagation nodes; The strategy optimization feedback module generates an equipment maintenance strategy priority queue according to the core propagation nodes and their abnormal fluctuation transmission paths, and updates the dynamic equipment topology map with the execution results as feedback input.

2. The intelligent power distribution operation and maintenance management system based on 5G transmission according to claim 1, characterized in that, The multi-dimensional feature fusion module comprises: The spectral feature analysis submodule performs multi-scale decomposition on the current and voltage waveforms, extracts the energy distribution ratio of the fundamental frequency component and the harmonic component, and generates a current and voltage spectrum feature vector; The temperature trend modeling submodule divides the equipment temperature sequence into time windows, calculates the temperature change rate and extreme point position in each window, and outputs an equipment temperature trend feature vector; The environmental correlation quantization submodule calculates the contribution weight of environmental factors to the equipment state based on the time shift correlation between the environmental monitoring indexes and the equipment temperature sequence; The feature tensor integration submodule concatenates the current and voltage spectrum feature vector, the equipment temperature trend feature vector and the environmental factor contribution weight, and outputs the equipment state feature tensor after normalization. 3.The intelligent power distribution operation and maintenance management system based on 5G transmission of claim 2, wherein, The dynamic topology generation module comprises: The correlation strength calculation submodule calculates the cosine similarity and dynamic time warping distance of the feature vectors of any two power distribution equipment based on the equipment state feature tensor, and generates the equipment correlation strength value by weighting; The path generation submodule traverses all equipment correlation strength values, filters the correlation strength values exceeding the preset threshold, and constructs the directed connection link between the equipment nodes in descending order of strength value; The topology iteration submodule performs closed loop detection on the directed connection link, generates a hierarchical equipment connection path after eliminating redundant links, and outputs the node relationship matrix of the dynamic equipment topology map.

4. The intelligent power distribution operation and maintenance management system based on 5G transmission according to claim 3, characterized in that, The abnormal propagation analysis module comprises: The fluctuation mode extraction submodule collects the state feature change amount of each equipment node in the continuous time slice based on the node relationship matrix of the dynamic equipment topology map, and marks the feature offset amount exceeding the normal fluctuation range; The propagation path tracking submodule backtracks the abnormal fluctuation source node and records the transmission path node sequence according to the occurrence time sequence of the feature offset amount and the direction of the equipment connection path; The core node identification submodule counts the triggered frequency and path coverage of each node in the transmission path node sequence, and selects the nodes with high frequency and coverage as the core propagation nodes.

5. The intelligent power distribution operation and maintenance management system based on 5G transmission according to claim 4, characterized in that, The strategy optimization feedback module comprises: The policy priority generation submodule calculates a device maintenance urgency score according to the device type of the core propagation node and the abnormal fluctuation transmission path length, sorts the score to generate a maintenance policy priority queue; The feedback mapping submodule collects the device state feature tensor change after the maintenance policy is executed, quantifies the abnormal fluctuation elimination rate and maps it to the corresponding device node; The topology update submodule adjusts the correlation strength value between device nodes based on the abnormal fluctuation elimination rate, reconstructs the device connection path and updates the dynamic device topology map. 6.The intelligent power distribution operation and maintenance management system based on 5G transmission of claim 5, wherein, The system further comprises a fault prediction module: The fault prediction module extracts the state feature attenuation trend of the key device node based on the updated dynamic device topology map, predicts the remaining effective operation time of the device and outputs a life warning list.

7. The intelligent power distribution operation and maintenance management system based on 5G transmission according to claim 6, characterized in that, The system further comprises a multi-stage optimization module: The multi-stage optimization module cross- validates the life warning list and the maintenance policy priority queue, and marks a set of device nodes that need to be processed in priority; The multi-stage optimization module corrects the attenuation rate calculation model of the device state feature tensor according to the maintenance record of the device node set. 8.The intelligent power distribution operation and maintenance management system based on 5G transmission of claim 7, wherein, The system further comprises a resource scheduling module: The resource scheduling module generates an optimal path scheduling scheme based on the geographical location distribution of the device in the maintenance policy priority queue and in combination with the maintenance personnel location information. 9.The intelligent power distribution operation and maintenance management system based on 5G transmission of claim 8, wherein, The system further comprises a digital twin mapping module: The digital twin mapping module maps the dynamic device topology map and the real-time device state feature tensor to a three-dimensional power grid model to generate a power distribution device operating state heat map.

10. The intelligent power distribution operation and maintenance management system based on 5G transmission of claim 9, characterized in that, The system further comprises a decision verification module: The decision verification module compares the power distribution device operating state heat map with the historical operation and maintenance record, verifies the execution effectiveness of the maintenance policy priority queue and outputs a deviation report.

Images