Power transmission and transformation equipment maintenance and overhaul decision generation system, method, product and equipment

Abstract

The invention discloses a power transmission and transformation equipment maintenance and overhaul decision generation system, method, product and equipment, and relates to the technical field of power equipment state monitoring and operation and maintenance management, the system comprises a sensing layer, an intelligent brain layer and an application layer, the sensing layer collects multi-source heterogeneous monitoring data; the intelligent brain layer extracts feature indexes through data fusion, constructs a digital twin model fused with a multi-physical field mechanism, adopts an AI algorithm to realize fault prediction and health assessment, and generates a maintenance strategy based on multi-objective optimization; and the application layer provides a panoramic cockpit for visual interaction. Through a technical chain of multi-source data fusion, digital twinning, AI prediction and multi-objective optimization, the limitation of data islands, early warning lag and decision isolation in traditional operation and maintenance is broken through, normal form transformation from passive response to active prediction and from local optimization to global optimization is realized, and the method has the advantages of being simple in structure, convenient to operate and high in practicability. The accuracy, the safety and the economical efficiency of operation and maintenance of the power transmission and transformation equipment are remarkably improved, and the method has wide engineering application value.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring and operation and maintenance management technology, and in particular to a system, method, product and equipment for generating maintenance and repair decisions for power transmission and transformation equipment. Background Technology

[0002] With the continuous expansion of the power grid and the ongoing improvement of its intelligence level, higher demands are being placed on the reliability and efficiency of the operation and maintenance of key power transmission and transformation equipment such as transformers, GIS, and circuit breakers. Currently, power system operation and maintenance technology has transitioned from reactive maintenance and periodic inspections to more advanced condition-based maintenance. Especially in critical substations and lines, the application rate of various online monitoring devices (such as online transformer oil chromatography monitoring, partial discharge monitoring, and infrared thermal imaging monitoring) has significantly increased. However, in the process of building smart grids and digital substations, existing condition monitoring solutions based on various online monitoring systems and inspection robots still face a series of technical bottlenecks that urgently need to be addressed in practical applications.

[0003] First, there is a serious problem of data silos, forming "information islands." Different systems, such as transformer oil chromatography, partial discharge, infrared thermography, and high-frequency sensing, typically operate independently, with varying data formats, sampling frequencies, and communication protocols. This leads to isolated monitoring data, making effective fusion and correlation analysis impossible. Early signs of a complex equipment failure (such as transformer inter-turn overheating) may be scattered across different monitoring systems (e.g., abnormal gas in the oil, weak changes in partial discharge signals, abnormal temperature gradients, etc.). Due to the lack of collaborative analysis of multi-source data, these early, cross-system "weak alarms" cannot be effectively captured and correlated, making it difficult for the system to form a unified and comprehensive understanding and assessment of the overall health status of the equipment, often missing the optimal opportunity for early warning.

[0004] Secondly, the early warning mechanism is lagging and reactive, with limited risk prevention capabilities. Existing technologies mostly use simple fixed thresholds for over-limit alarms, a mode that is essentially "after-the-fact" or "during-the-fact"—triggered only when equipment status parameters have significantly deteriorated. This results in severely delayed early warning information for maintenance personnel, a short window for emergency response, and makes it difficult to truly "prevent problems before they occur," thus failing to effectively reduce the risks to power grid operation caused by sudden equipment failures.

[0005] Secondly, there is a lack of accurate prediction and lifespan assessment capabilities for the specific fault modes of power transmission and transformation equipment. Existing technologies primarily focus on monitoring and assessing the "current" or "historical" state of equipment, failing to accurately predict the performance degradation trajectory and remaining service life of equipment based on the unique multi-physics coupling mechanisms (such as electro-thermal-mechanical-chemical fields) and historical trends of multi-dimensional data. Therefore, maintenance strategies lack sufficient foresight, making it difficult to truly move from "periodic maintenance" or "condition-based maintenance" to "predictive maintenance." This hinders the ability to scientifically extend equipment maintenance cycles and optimize resource allocation while ensuring grid safety, thereby achieving optimal maintenance costs.

[0006] Finally, maintenance decisions are severely disconnected from the actual operating conditions of the power grid, failing to achieve multi-objective optimization. Existing maintenance decision-making processes often focus only on the urgency of equipment status, failing to systematically consider and quantify multiple complex power system constraints, such as equipment risk levels, maintenance costs, load losses due to power outages, the current operating mode of the power grid, the N-1 safety criterion, and spare parts inventory. This results in maintenance plans that may not be globally optimal, easily leading to resource waste or uneven control of power grid operational risks.

[0007] Therefore, it is necessary to design a power transmission and transformation equipment maintenance and repair decision generation system to address the problems of weak data fusion capabilities leading to missed early faults and delayed warnings, lack of predictive maintenance models that integrate the multi-physics mechanisms of equipment and the operating characteristics of the power grid, and failure to deeply coordinate and optimize maintenance decisions with power grid safety and economy. Summary of the Invention

[0008] The technical problem this invention aims to solve is to address the shortcomings of the existing technology by providing an operation and maintenance system and method based on multi-source data fusion and digital twins, which deeply integrates the multi-physics mechanism of power transmission and transformation equipment with power grid operation constraints. This system can specifically target key power equipment such as transformers and GIS, enabling early and accurate warnings and health trend predictions of typical fault modes (such as insulation aging, mechanical loosening, and overheating). The warning and prediction process does not rely on the personal experience of operation and maintenance personnel. The entire operation and maintenance decision-making process can automatically generate dynamic maintenance plans that are risk-controllable and economically optimal, based on multiple objectives such as real-time equipment status, power grid risk assessment, maintenance costs, and power outage losses, strictly adhering to power grid safety operation procedures. This effectively avoids over-maintenance and under-maintenance, significantly improving the reliability, safety, and economy of power grid operation.

[0009] The solution adopted by this invention to solve its technical problem is as follows:

[0010] A method for generating maintenance and repair decisions for power transmission and transformation equipment, based on multi-source data fusion technology and digital twin technology, includes:

[0011] The perception layer is used to collect multi-source heterogeneous data from power transmission and transformation equipment. This multi-source heterogeneous data at least includes equipment operating status data, environmental data, power grid disturbance data, and historical operation and maintenance data.

[0012] The intelligent brain layer, deployed in the backend server cluster, is the core processing and decision-making center of the system. It is used to receive and process multi-source heterogeneous data uploaded by the perception layer to generate maintenance decisions.

[0013] The application layer, serving as the output interface for system functions, is used to provide visual interaction for operations and maintenance personnel through a client.

[0014] The perception layer uploads multi-source heterogeneous data to the intelligent brain layer via the network.

[0015] The intelligent brain layer includes:

[0016] The data fusion and feature extraction module receives multi-source heterogeneous data from the perception layer, performs power data-specific cleaning, format alignment, and time synchronization, and extracts key feature indicators related to the health diagnosis of power transmission and transformation equipment.

[0017] The equipment digital twin model module is used to establish a digital twin model of the power transmission and transformation equipment, and to drive the digital twin model based on the data processed by the data fusion and feature extraction module to simulate the internal state of the power transmission and transformation equipment.

[0018] The AI-based fault prediction and health assessment module is used to perform short-term fault prediction and long-term health assessment based on the key feature indicators and simulation data from the digital twin model, generating prediction and assessment results.

[0019] The dynamic maintenance decision optimization module receives the prediction and evaluation results from the AI ​​fault prediction and health assessment module, and, combined with power grid operation constraints, generates the optimal maintenance strategy through multi-objective optimization.

[0020] The visual interaction includes at least a panoramic cockpit page, an early warning and reporting center page, and a maintenance work order management page. The panoramic cockpit page includes at least a main visualization area, an equipment early warning list area, a maintenance strategy recommendation area, and a function entry area.

[0021] As a preferred embodiment of the present invention

[0022] The sensing layer includes:

[0023] Fixed online monitoring units are used to collect oil chromatography data, partial discharge data, winding temperature, oil temperature, and bushing dielectric loss from power transmission and transformation equipment.

[0024] The mobile inspection unit, comprising inspection robots and drones, is used to collect high-definition images, infrared thermal images, and operational noise audio data of power transmission and transformation equipment.

[0025] The environmental information access unit is used to access external environmental data and power grid disturbance data.

[0026] The operation and maintenance history database is used to store ledgers, defect records, maintenance reports, and electrical test data of power transmission and transformation equipment.

[0027] As a preferred embodiment of the present invention

[0028] The dynamic maintenance decision optimization module aims to minimize the failure risk of power transmission and transformation equipment and optimize the comprehensive cost throughout its entire life cycle. It takes the power outage time window provided by the power grid dispatching department, the N-1 safety verification result of the power grid during the maintenance period, the inventory of spare parts, and the available maintenance human resources as hard constraints.

[0029] The optimal maintenance strategy includes maintenance time, maintenance level, maintenance content, resource list and estimated duration, and is automatically checked against the power grid dispatch plan.

[0030] As a preferred embodiment of the present invention

[0031] The AI ​​fault prediction and health assessment module employs at least one of the following: a long short-term memory network model that integrates attention mechanisms, a Transformer time series model, or a gradient boosting decision tree model, to perform short-term fault prediction.

[0032] As a preferred embodiment of the present invention

[0033] The AI ​​fault prediction and health assessment module combines the mechanism analysis results of the digital twin model, uses a Bayesian network model to calculate the comprehensive health index of the device and predict its remaining service life.

[0034] As a preferred embodiment of the present invention

[0035] The dynamic maintenance decision optimization module uses at least one of the following algorithms for multi-objective optimization: genetic algorithm, particle swarm optimization algorithm, or simulated annealing algorithm.

[0036] As a preferred embodiment of the present invention

[0037] The device digital twin model module uses a simplified equivalent model based on physical mechanisms or a deep learning proxy model driven entirely by historical data to replace the multi-physics coupling model.

[0038] A method for generating maintenance and repair decisions for power transmission and transformation equipment includes the following steps:

[0039] Step S101: Multi-source data acquisition,

[0040] Collect multi-source heterogeneous data from power transmission and transformation equipment.

[0041] Step S102: Data fusion and feature extraction.

[0042] The system processes multi-source heterogeneous data to extract key feature indicators related to fault diagnosis of power transmission and transformation equipment, based on the characteristics of power data.

[0043] Step S103: Drive the digital twin model and update the virtual state.

[0044] Based on key characteristic indicators, the internal state of power transmission and transformation equipment is simulated using a digital twin model.

[0045] Step S104: Perform AI fault prediction and health assessment.

[0046] Based on simulation data from key characteristic indicators and digital twin models, fault prediction and health assessment are performed, generating prediction and assessment results.

[0047] Step S105: Perform multi-objective optimization to generate maintenance strategy.

[0048] Based on the prediction and evaluation results, and combined with power grid operation constraints, a multi-objective optimization algorithm is run to generate a maintenance strategy with optimal risk and cost.

[0049] Step S106: Output maintenance strategy and work order to guide on-site operations.

[0050] Output the optimal maintenance strategy and automatically generate standardized maintenance work orders integrated with the production management system to guide on-site operations.

[0051] A computer program product includes a computer program that, when executed by a processor, implements the steps of the power transmission and transformation equipment maintenance and repair decision generation method.

[0052] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the power transmission and transformation equipment maintenance and repair decision generation method.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] This invention provides an operation and maintenance system and method based on multi-source data fusion and digital twins, which deeply integrates the physical characteristics of power transmission and transformation equipment with power grid operation constraints. Specifically targeting key power equipment such as transformers and GIS, it enables early and accurate warnings and health trend predictions of typical fault modes. The warning and prediction process does not rely on the personal experience of operation and maintenance personnel. The entire operation and maintenance decision-making process is based on multiple objectives, including real-time equipment status, power grid risk assessment, maintenance costs, and power outage losses, strictly adhering to power grid safety operation procedures. It automatically generates dynamic maintenance plans that are risk-controllable and economically optimal, effectively avoiding over-maintenance and under-maintenance, and significantly improving the reliability, safety, and economy of power grid operation. It has the following advantages:

[0055] 1. A highly collaborative intelligent decision-making system has been established.

[0056] This invention, targeting the operation and maintenance (O&M) scenarios of power transmission and transformation equipment, innovatively constructs a smart brain layer that includes data fusion and feature extraction, equipment digital twin models, AI fault prediction and health assessment, and dynamic maintenance decision optimization. Through the collaborative work of multiple modules, it achieves automated pipeline processing from raw data acquisition to intelligent decision output, solving the problem of the separation between analysis and decision-making in traditional O&M systems, and greatly improving the overall efficiency and systematic nature of O&M work.

[0057] 2. It has achieved true deep fusion of multi-source data.

[0058] This invention utilizes a specially designed data fusion and feature extraction module. The system effectively integrates and processes multi-source heterogeneous data from power-specific devices such as oil chromatography, partial discharge monitoring, infrared thermography, and vibration monitoring, extracting industry-recognized diagnostic features such as the three-ratio coding method. This fundamentally breaks down "information silos," enabling the effective correlation and capture of early, weak, and scattered fault symptoms across different monitoring systems, achieving early fault warnings and significantly reducing the false negative rate.

[0059] 3. An innovative predictive maintenance model based on mechanism and data-driven approaches has been developed.

[0060] This invention combines a digital twin model that integrates multi-physics coupling mechanisms with an AI prediction model, forming a unique "mechanism + data" dual-driven analysis capability. This not only enables accurate prediction of short-term fault probabilities but also allows for scientific assessment of long-term performance degradation trends such as transformer insulation aging. It provides core technical support for power transmission and transformation equipment to move from "periodic maintenance" and "condition-based maintenance" to "predictive maintenance," thereby scientifically extending equipment lifespan and optimizing the entire lifecycle operation and maintenance costs.

[0061] 4. It has achieved optimized maintenance decision-making that is deeply coupled with power grid operation.

[0062] This invention, through a dynamic maintenance decision optimization module, for the first time deeply integrates grid-specific constraints such as grid operation mode, N-1 safety criterion, and power outage losses into the decision model, establishing a mathematical model with the objectives of minimizing risk and optimizing overall cost, and solving it using an intelligent optimization algorithm. The generated maintenance strategy is a globally optimal solution that meets the requirements of grid safety and economic operation, significantly improving the overall efficiency of power company operation and maintenance work.

[0063] 5. It provides an immersive and interactive operation and maintenance experience.

[0064] This invention provides an integrated digital twin visualization and AR remote guidance panoramic cockpit through the application layer, which presents the complex internal status of equipment and data analysis results to maintenance personnel in an intuitive way, greatly facilitating monitoring and decision-making. It also achieves seamless integration of back-end expert knowledge and on-site maintenance operations through AR technology, improving the efficiency, accuracy and safety of handling complex faults.

[0065] 6. Possesses a high degree of industry expertise and market adaptability.

[0066] The overall architecture of this invention is closely integrated with the characteristics of the power industry. Its technical solution is specifically designed for the physical characteristics of power transmission and transformation equipment and the operating rules of the power grid, exhibiting a high degree of professionalism and practicality. The system supports various flexible deployment options, from local deployment to cloud-based SaaS services, providing reliable technical support for users of different sizes and needs, and possessing broad and reliable market adaptability.

[0067] In summary, this invention, through the systematic integration and innovation of technologies such as multi-source data fusion, digital twins, artificial intelligence, and optimized decision-making, has successfully constructed a comprehensive, intelligent, and forward-looking decision support system for the operation and maintenance of power transmission and transformation equipment. It has achieved significant improvements in early warning accuracy, decision-making scientificity, operation and maintenance economy, and system practicality, and is of great value in ensuring the safe, stable, and economical operation of the power grid. Attached Figure Description

[0068] Figure 1 This is a schematic diagram of the overall architecture of a power transmission and transformation equipment maintenance and repair decision generation system proposed in this invention;

[0069] Figure 2 This is a schematic diagram of the intelligent brain layer of a power transmission and transformation equipment maintenance and repair decision generation system proposed in this invention;

[0070] Figure 3 This is a flowchart illustrating a method for generating maintenance and repair decisions for power transmission and transformation equipment proposed in this invention.

[0071] Figure 4This is a schematic diagram of the user interface layout of a panoramic cockpit in a power transmission and transformation equipment maintenance and repair decision generation system proposed in this invention. Detailed Implementation

[0072] The specific embodiments of the present invention are described below with reference to the accompanying drawings and examples:

[0073] It should be noted that the structures, colors, proportions, sizes, etc. shown in the accompanying drawings are only used to complement the content disclosed in the specification, so that those skilled in the art can understand and read them, and are not intended to limit the conditions under which the present invention can be implemented. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0074] This invention provides a predictive maintenance and dynamic overhaul decision-making system and method for power transmission and transformation equipment based on multi-source data fusion and digital twins. The system is built on a hardware computing platform consisting of servers, network equipment, and storage devices, supporting various flexible deployment options: A local server deployment mode can be used to ensure data privacy, with the hardware computing platform located in the user's own operation and maintenance center or substation computer room. Alternatively, a cloud-based SaaS service model can be adopted, deploying the core functional modules of the system on a cloud server cluster. Users do not need to invest in hardware procurement and maintenance costs; they can simply access the service via the web or mobile terminal, providing users with a flexible, ready-to-use, maintenance-free service. A hybrid architecture combining edge computing and cloud computing can also be adopted. For latency-sensitive tasks (requiring millisecond-level response) such as real-time monitoring of power transmission and transformation equipment (e.g., real-time analysis of partial discharge signals, dynamic simulation of digital twin models), the architecture can be deployed on substation edge computing nodes (e.g., edge servers, smart gateways) to reduce data transmission link length and ensure real-time performance. Meanwhile, tasks requiring massive data support and high computing power, such as long-term equipment health trend analysis (e.g., insulation aging life assessment) and cross-regional maintenance resource optimization, can be handled in the cloud, balancing maintenance timeliness and scientific decision-making through a collaborative model. The core of this approach lies in achieving intelligent upgrades to the entire process of power transmission and transformation equipment operation and maintenance through hardware and software collaborative optimization.

[0075] like Figure 1 As shown, this invention provides a decision generation system for the maintenance and repair of power transmission and transformation equipment. Based on multi-source data fusion technology and digital twin technology, it includes three logical layers: a perception layer, a smart brain layer, and an application layer. The perception layer uploads multi-source heterogeneous data to the smart brain layer through a dedicated power network or wireless network, forming a closed-loop system for the entire process of "data acquisition - intelligent analysis - decision output".

[0076] The perception layer, as the data acquisition source of the system, is responsible for collecting multi-source heterogeneous data from power transmission and transformation equipment (transformers, GIS, circuit breakers, etc.) in the physical world, specifically including:

[0077] Fixed online monitoring unit: used to collect core operating data of power transmission and transformation equipment, including oil chromatography data (content and ratio of characteristic gases such as H2, CH4, C2H2, C2H4, and C2H6), partial discharge data (discharge quantity, phase, and frequency), winding temperature, oil temperature, bushing dielectric loss, and other electrical and chemical state quantities.

[0078] Mobile inspection units include inspection robots and drones equipped with infrared thermal imagers and visible light cameras, used to collect high-definition external images, infrared thermal images, and operating noise audio data of power transmission and transformation equipment, covering areas that are difficult for fixed monitoring units to reach.

[0079] Environmental information access unit: used to access external environmental and power grid disturbance data, including meteorological data (temperature, humidity, wind speed), lightning location data, short-circuit current data, etc., to provide external scenario support for equipment status analysis.

[0080] Operation and Maintenance History Database: Used to store basic data for the entire lifecycle of substation equipment. It covers equipment ledgers, historical defect records, maintenance reports, and electrical test data, providing historical evidence for trend analysis and fault tracing.

[0081] Preferably, the mobile inspection unit can also use a high-definition PTZ camera and acoustic sensor array fixedly installed in the equipment area to replace or assist inspection robots and drones, so as to realize routine automatic monitoring of specific areas and reduce the deployment cost and maintenance workload of mobile devices.

[0082] The intelligent brain layer, deployed within a backend server cluster, is the system's core processing and decision-making center. It receives and processes multi-source heterogeneous data uploaded by the perception layer to generate maintenance decisions. Through the collaborative work of multiple software functional modules, raw data is transformed into equipment health assessment results and maintenance decisions. Figure 2 As shown, the intelligent brain layer specifically includes:

[0083] The data fusion and feature extraction module receives multi-source heterogeneous data uploaded from the perception layer, performs power data-specific cleaning (such as removing outliers caused by grid disturbances), format alignment, and time synchronization to form unified and regular equipment status time series data. Subsequently, specifically for the health diagnosis needs of power transmission and transformation equipment, it extracts key feature indicators from the fused data, such as transformer three-ratio encoding, David triangle coordinates, PRPD spectrum statistical features of partial discharge signals (such as skewness and steepness), normalized temperature difference between equipment temperature and ambient temperature, and spectral features of vibration signals, providing high-quality data support for subsequent analysis.

[0084] The equipment digital twin model module establishes a high-fidelity virtual model integrating multi-physics mechanisms for each key physical device (such as the #1 main transformer). This digital twin model not only replicates the physical structure of the equipment but also incorporates mechanistic models reflecting its electro-thermal-mechanical coupling, insulation material aging, and fault evolution (such as a finite element-based thermo-fluid coupling model and an insulation paper polymerization degradation model). This module receives real-time field data from the data fusion and feature extraction module, driving the digital twin virtual model to perform synchronous simulations, thereby achieving real-time inversion and mapping of unmeasurable internal states of power transmission and transformation equipment (such as hotspot temperatures and insulation mechanical stress).

[0085] Preferably, the device digital twin model module can also use a simplified equivalent model based on physical mechanisms or a deep learning proxy model driven entirely by historical data to replace the high-fidelity multi-physics coupling model, so as to balance computing resources and model accuracy in specific scenarios.

[0086] The AI-powered fault prediction and health assessment module is the intelligent core of the system. Based on key feature indicators provided by the data fusion and feature extraction module and simulation data from the digital twin model of substation equipment, it performs in-depth analysis. For short-term prediction, it employs a long short-term memory network model incorporating an attention mechanism to predict the probability of specific fault modes (such as predicting overheating faults based on oil chromatography data and mechanical loosening based on vibration data) occurring within the next few days to weeks. For long-term health assessment, it combines the mechanistic analysis results of the digital twin model (such as the insulation paper life curve) and real-time monitoring data, using a Bayesian network equiprobability graphical model to calculate the comprehensive health index of the equipment and predict its remaining service life. The calculation of the health index comprehensively considers multiple dimensions of state, including electrical, chemical, and mechanical factors.

[0087] The LSTM model (Long Short-Term Memory Network with Attention Mechanism) used in the AI ​​fault prediction and health assessment module has a network structure consisting of an input layer, two hidden layers, and an output layer. The dimension of the input layer is the same as the number of key features extracted that are related to power transmission and transformation equipment faults, and the output layer is the probability value of a specific fault mode (such as discharge, overheating, or mechanical fault).

[0088] Preferably, the AI ​​fault prediction and health assessment module can also use a Transformer time series model or a gradient boosting decision tree (such as XGBoost) model instead of an LSTM model to achieve short-term trend prediction of equipment status and calculation of fault probability.

[0089] The dynamic maintenance decision optimization module establishes a multi-objective optimization model specifically for power grid maintenance decisions. It simultaneously receives prediction and evaluation results (such as fault probability, health index, and remaining lifetime) from the AI ​​fault prediction and health assessment module for solution. The solution process utilizes an improved genetic algorithm, ultimately outputting a recommended optimal maintenance strategy. This optimal strategy specifically includes suggested maintenance time, maintenance level (A / B / C / D), specific maintenance content, required resource list, and estimated duration, and automatically performs preliminary verification with the power grid dispatch plan.

[0090] The multi-objective optimization model primarily aims to minimize equipment failure risk (risk value = failure probability × failure consequences, including economic losses and grid security impacts) and optimize the overall lifecycle cost (including direct maintenance costs and power outage loss costs). Its mathematical objective function can be: Min(α * risk value + β * cost value), where α and β are weighting coefficients that can be adjusted according to management strategies. The risk value is calculated by multiplying the failure probability by the failure consequences (comprehensively considering equipment damage costs, power outage losses, and impacts on grid stability). The cost value includes direct maintenance costs and load loss costs caused by power outages. Simultaneously, the dynamic maintenance decision optimization module uses the power outage time window provided by the grid dispatching department, the grid N-1 safety verification results during the maintenance period, spare parts inventory, and available maintenance human resources as hard constraints.

[0091] Preferably, the dynamic maintenance decision optimization module can also use particle swarm optimization algorithm or simulated annealing algorithm instead of genetic algorithm to solve the optimal maintenance strategy under multiple constraints.

[0092] The application layer, serving as the output interface for system functions, provides visual interaction for maintenance personnel via web clients or mobile terminals. The visual interaction includes at least a panoramic cockpit page, an early warning and reporting center page, and a maintenance work order management page. The panoramic cockpit page uses a three-dimensional virtual power plant model to display the integrated status and location information of equipment.

[0093] like Figure 4 As shown, the panoramic cockpit interface provided by the application layer includes at least the following areas in its layout:

[0094] The main visualization area includes the substation primary wiring diagram and status indicator lights, which intuitively display the real-time health status of each device using different colored (such as red, yellow, and green) lights. It also supports clicking to view the 3D model and internal simulation data of the device's digital twin.

[0095] The equipment warning list area displays early risk warning information proactively pushed by the system, including equipment name, warning level, suspected fault points derived from multi-source data fusion analysis (such as "transformer medium temperature overheating"), and handling suggestions.

[0096] The maintenance strategy recommendation area displays details of the optimal maintenance plan generated by the dynamic maintenance decision optimization module, along with a risk assessment report and cost-benefit analysis under that strategy.

[0097] The function entry area (AR remote collaboration entry) allows maintenance personnel to initiate calls and conduct remote collaborative maintenance with back-end experts from a first-person perspective through AR glasses. The system can overlay the fault location information analyzed by the digital twin model onto the real equipment on site to guide precise maintenance.

[0098] like Figure 3 As shown, this invention provides a method for generating maintenance and repair decisions for power transmission and transformation equipment, including the following steps, forming a complete process of "data acquisition - data processing - data analysis - decision generation - decision implementation":

[0099] Step S101: Multi-source data acquisition,

[0100] The power transmission and transformation equipment maintenance and repair decision generation system is launched. The fixed online monitoring unit, mobile inspection unit, and environmental information access unit of the system perception layer start working simultaneously. Operation data, environmental data, and power grid disturbance data are collected, and basic data such as equipment ledgers and historical maintenance records in the operation and maintenance history database are called up. All data are uploaded to the smart brain layer in real time through the power dedicated network or wireless network.

[0101] Step S102: Data fusion and feature extraction.

[0102] After receiving heterogeneous data from multiple sources, the data fusion and feature extraction module of the intelligent brain layer processes the data according to the characteristics of power data: first, it removes outliers caused by grid disturbances and sensor noise, completes data format alignment and time synchronization, and forms a unified equipment status time series dataset; then, in response to the fault diagnosis needs of power transmission and transformation equipment, it extracts key feature indicators such as transformer three-ratio method coding, partial discharge PRPD spectrum statistical features, and vibration signal spectrum features, providing core data support for subsequent analysis.

[0103] Step S103: Drive the digital twin model and update the virtual state.

[0104] The equipment digital twin model module receives the fused dataset and conducts real-time simulation based on multiphysics mechanisms: it calibrates the virtual model parameters by inputting field data, updates the virtual operating status of the equipment synchronously, and inverts parameters that cannot be directly measured by physical sensors, such as internal hot spot temperature and insulation mechanical stress, to achieve accurate virtual-real mapping between the physical equipment and the virtual model.

[0105] Step S104: Perform AI fault prediction and health assessment.

[0106] The AI ​​fault prediction and health assessment module calls the key feature indicators extracted in step S102 and the simulation data output in step S103 to conduct bidirectional analysis: on the one hand, it uses an LSTM model with an attention mechanism to predict the probability and risk level of specific fault modes in the next few days to weeks; on the other hand, it combines the mechanism analysis results of the digital twin model (such as the aging curve of insulation materials) and real-time monitoring data, calculates the comprehensive health index of the equipment through a Bayesian network, accurately predicts the remaining service life, and finally generates a comprehensive assessment report that includes short-term fault risk and long-term health status.

[0107] Step S105: Perform multi-objective optimization to generate maintenance strategy.

[0108] The dynamic maintenance decision optimization module receives the evaluation report from step S104 and, in conjunction with the power grid operation constraints (power outage time window, N-1 safety criterion, spare parts inventory, and maintenance human resources), initiates a multi-objective optimization model: with the objective of "lowest fault risk + optimal life cycle cost", it solves the global optimal solution through an improved genetic algorithm, generates a complete maintenance strategy including maintenance time, maintenance level, specific content, resource list, and expected duration, and automatically performs a preliminary check with the power grid dispatch plan to ensure the feasibility of the strategy.

[0109] Step S106: Output maintenance strategy and work order to guide on-site operations.

[0110] The application layer pushes the maintenance strategy generated in step S105 to the maintenance personnel (web client or mobile terminal) in a visual manner. At the same time, it automatically generates standardized maintenance work orders that are seamlessly integrated with the production management system (PMS). The work orders include information such as fault location, operation process, safety precautions, and a list of required tools and spare parts, which directly guide the on-site maintenance personnel to carry out maintenance work.

[0111] Example 1:

[0112] A provincial power grid company manages 20 substations of 220kV and above. It needs to meet the requirements of real-time monitoring of multiple substations, cross-regional resource coordination, and independent data control. An edge-cloud hybrid deployment solution is adopted.

[0113] Hardware deployment: Each substation is equipped with an edge server and a dedicated power communication gateway, while the provincial operation and maintenance center deploys a cloud server cluster and distributed storage devices.

[0114] Data Acquisition: The fixed online monitoring unit collects oil chromatography and GIS partial discharge data of the main transformer every 5 minutes; the inspection robot collects infrared thermal images and high-definition images every week; the environmental information access unit synchronizes meteorological and lightning data in real time, and uploads them to the cloud after preprocessing by the edge node.

[0115] The intelligent brain layer operates as follows: the edge side completes data cleaning and feature extraction in real time, driving the digital twin model to invert the main transformer hotspot temperature; the cloud uses the LSTM model to predict the probability of failure in the next two weeks, and generates a health report by combining insulation aging mechanism analysis; the dynamic maintenance decision optimization module coordinates cross-regional maintenance resources and grid N-1 constraints to generate the optimal maintenance plan (e.g., for the risk of overheating of the #1 main transformer in a certain substation, it is recommended to carry out Class C maintenance during the low load period of the following month).

[0116] Application results: Early detection rate of equipment failure increased by 40%, maintenance costs decreased by 25%, and cross-regional resource scheduling efficiency increased by 30%.

[0117] Example 2:

[0118] A power supply company in a prefecture-level city manages eight 110kV substations. Due to limitations in hardware budget and the size of its operation and maintenance team, it adopted a cloud-based SaaS deployment solution.

[0119] Hardware deployment: No local hardware investment is required. Users access the cloud system through a web client, and data is uploaded to the cloud storage node via an encrypted network.

[0120] Data Acquisition: Relying on existing online monitoring devices to collect data on circuit breaker vibration and transformer oil temperature, and commissioning third-party drones to collect infrared thermal images and visible light images every month.

[0121] The intelligent brain layer operates as follows: data fusion and feature extraction are completed in the cloud, and a simplified digital twin model is used to balance computing power; the Transformer model is used to predict the risk of equipment failure within one month, and the dynamic maintenance decision optimization module simplifies the model, focusing on balancing maintenance costs and power outage losses, and generating maintenance plans adapted to local resources (such as the risk of mechanical jamming of circuit breakers, recommending the replacement of Class D local components).

[0122] Application results: The rate of false alarms for equipment failures is reduced by 35%, the efficiency of maintenance plan execution is improved by 50%, no additional maintenance personnel are required, and the application threshold is greatly reduced.

[0123] Example 3:

[0124] Two main transformers at a large thermal power plant are core equipment with extremely high requirements for data security and reliability. A local server deployment solution is adopted.

[0125] Hardware deployment: The power plant's operation and maintenance computer room is equipped with two redundant high-performance servers, industrial-grade switches, and a large-capacity storage array, with data stored locally and not transmitted across domains.

[0126] Data acquisition: Fixed online monitoring units collect oil chromatography, winding temperature, and bushing dielectric loss data in real time; drones collect images of high-altitude parts of the equipment every quarter; and the operation and maintenance history database stores data throughout the entire lifecycle of the equipment since its commissioning.

[0127] The intelligent brain layer operates by using a high-fidelity digital twin multiphysics coupling model to invert the internal hot spot temperature and insulation aging status of the main transformer; it uses the XGBoost model to predict the risk of overheating and insulation moisture within one month, and generates maintenance strategies in combination with the unit's power generation plan (such as recommending Class A maintenance during the unit's overhaul period if the main transformer insulation aging is accelerated).

[0128] Application results: The unplanned outage rate of the main transformer was reduced to below 0.1%, the replacement cycle of insulation components was extended by 15%, and the reliability of power generation was significantly improved.

[0129] A computer program product, such as computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the present invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0130] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The memory and the processor are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the steps of a cable aging detection method based on ultra-wideband pulse radar.

[0131] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

[0132] Many other changes and modifications can be made without departing from the concept and scope of this invention. It should be understood that this invention is not limited to the specific embodiments, and the scope of this invention is defined by the appended claims.

Claims

1. A decision generation system for maintenance and repair of power transmission and transformation equipment, characterized in that, Based on multi-source data fusion technology and digital twin technology, including: The perception layer is used to collect multi-source heterogeneous data from power transmission and transformation equipment. This multi-source heterogeneous data at least includes equipment operating status data, environmental data, power grid disturbance data, and historical operation and maintenance data. The intelligent brain layer, deployed in the backend server cluster, is the core processing and decision-making center of the system. It is used to receive and process multi-source heterogeneous data uploaded by the perception layer to generate maintenance decisions. The application layer, serving as the output interface for system functions, is used to provide visual interaction for operations and maintenance personnel through a client. The perception layer uploads multi-source heterogeneous data to the intelligent brain layer via the network. The intelligent brain layer includes: The data fusion and feature extraction module receives multi-source heterogeneous data from the perception layer, performs power data-specific cleaning, format alignment, and time synchronization, and extracts key feature indicators related to the health diagnosis of power transmission and transformation equipment. The equipment digital twin model module is used to establish a digital twin model of the power transmission and transformation equipment, and to drive the digital twin model based on the data processed by the data fusion and feature extraction module to simulate the internal state of the power transmission and transformation equipment. The AI-based fault prediction and health assessment module is used to perform short-term fault prediction and long-term health assessment based on the key feature indicators and simulation data from the digital twin model, generating prediction and assessment results. The dynamic maintenance decision optimization module receives the prediction and evaluation results from the AI ​​fault prediction and health assessment module, and, combined with power grid operation constraints, generates the optimal maintenance strategy through multi-objective optimization. The visual interaction includes at least a panoramic cockpit page, an early warning and reporting center page, and a maintenance work order management page. The panoramic cockpit page includes at least a main visualization area, an equipment early warning list area, a maintenance strategy recommendation area, and a function entry area.

2. The power transmission and transformation equipment maintenance and repair decision generation system as described in claim 1, Its features are, The sensing layer includes: Fixed online monitoring units are used to collect oil chromatography data, partial discharge data, winding temperature, oil temperature, and bushing dielectric loss from power transmission and transformation equipment. The mobile inspection unit, comprising inspection robots and drones, is used to collect high-definition images, infrared thermal images, and operational noise audio data of power transmission and transformation equipment. The environmental information access unit is used to access external environmental data and power grid disturbance data. The operation and maintenance history database is used to store ledgers, defect records, maintenance reports, and electrical test data of power transmission and transformation equipment.

3. The power transmission and transformation equipment maintenance and repair decision generation system as described in claim 1, Its features are, The dynamic maintenance decision optimization module aims to minimize the failure risk of power transmission and transformation equipment and optimize the comprehensive cost throughout its entire life cycle. It takes the power outage time window provided by the power grid dispatching department, the N-1 safety verification result of the power grid during the maintenance period, the inventory of spare parts, and the available maintenance human resources as hard constraints. The optimal maintenance strategy includes maintenance time, maintenance level, maintenance content, resource list and estimated duration, and is automatically checked against the power grid dispatch plan.

4. The power transmission and transformation equipment maintenance and repair decision generation system as described in claim 1, Its features are, The AI ​​fault prediction and health assessment module employs at least one of the following: a long short-term memory network model that integrates attention mechanisms, a Transformer time series model, or a gradient boosting decision tree model, to perform short-term fault prediction.

5. The power transmission and transformation equipment maintenance and repair decision generation system as described in claim 1, Its features are, The AI ​​fault prediction and health assessment module combines the mechanism analysis results of the digital twin model, uses a Bayesian network model to calculate the comprehensive health index of the device and predict its remaining service life.

6. The power transmission and transformation equipment maintenance and repair decision generation system as described in claim 1, Its features are, The dynamic maintenance decision optimization module uses at least one of the following algorithms for multi-objective optimization: genetic algorithm, particle swarm optimization algorithm, or simulated annealing algorithm.

7. The power transmission and transformation equipment maintenance and repair decision generation system as described in claim 1, Its features are, The device digital twin model module uses a simplified equivalent model based on physical mechanisms or a deep learning proxy model driven entirely by historical data to replace the multi-physics coupling model.

8. A method for generating maintenance and repair decisions for power transmission and transformation equipment, employing the power transmission and transformation equipment maintenance and repair decision generation system described in any one of claims 1-7. Its features are, Includes the following steps, Step S101: Multi-source data acquisition, Collect multi-source heterogeneous data from power transmission and transformation equipment. Step S102: Data fusion and feature extraction. The system processes multi-source heterogeneous data to extract key feature indicators related to fault diagnosis of power transmission and transformation equipment, based on the characteristics of power data. Step S103: Drive the digital twin model and update the virtual state. Based on key characteristic indicators, the internal state of power transmission and transformation equipment is simulated using a digital twin model. Step S104: Perform AI fault prediction and health assessment. Based on simulation data from key characteristic indicators and digital twin models, fault prediction and health assessment are performed, generating prediction and assessment results. Step S105: Perform multi-objective optimization to generate maintenance strategy. Based on the prediction and evaluation results, and combined with power grid operation constraints, a multi-objective optimization algorithm is run to generate a maintenance strategy with optimal risk and cost. Step S106: Output maintenance strategy and work order to guide on-site operations. Output the optimal maintenance strategy and automatically generate standardized maintenance work orders integrated with the production management system to guide on-site operations.

9. A computer program product, comprising a computer program, Its features are, When the computer program is executed by the processor, it implements the steps of the power transmission and transformation equipment maintenance and repair decision generation method as described in claim 8.

10. A computer device, comprising: Memory, processor, and computer programs stored in memory and capable of running on the processor. The processor is characterized in that it executes the computer program to implement the steps of the power transmission and transformation equipment maintenance and repair decision generation method of claim 8.

Images