Power transformer monitoring method, device and equipment and storage medium

Abstract

The invention relates to a power transformer monitoring method, device and equipment and a storage medium, and the method comprises the steps: obtaining basic feature parameters and installation environment data of a plurality of power transformers, carrying out the classification matching based on the basic feature parameters and the installation environment data, and obtaining the arrangement information of a sensor; acquiring real-time working data of each power transformer according to the sensor arrangement information, and performing equipment parallel monitoring to obtain equipment parallel monitoring data; performing insulation degradation trend prediction on each power transformer based on the real-time working data and the equipment parallel monitoring data to obtain a single degradation prediction result; and carrying out collaborative management on a plurality of power transformer groups according to the equipment parallel monitoring data and the single degradation prediction result to obtain a transformer maintenance decision. According to the invention, the pertinence and accuracy of monitoring can be improved, and a more reliable data support is provided for the evaluation of the operation state of the transformer.

Description

Technical Field

[0001] The present invention relates to the technical field of power transformers, and in particular to a power transformer monitoring method, device, equipment and storage medium. Background Art

[0002] As a key device in the power system, power transformers play an irreplaceable role in the process of power transmission and voltage conversion. With the continuous advancement of smart grid construction and the continuous growth of electricity demand, higher requirements are placed on the real-time monitoring and scientific management of transformer operating status. Existing transformer monitoring methods usually adopt a unified monitoring standard and parameter system, which fails to fully consider the differences in structural characteristics of different types of transformers and the diversity of installation environments. This simplified monitoring method is difficult to meet the actual needs of differentiated monitoring. At the same time, traditional monitoring methods are often limited to the independent monitoring of a single device, lack the overall coordinated management of transformer groups, and fail to effectively optimize the sensor layout based on environmental factors. This affects the accuracy and comprehensiveness of the monitoring data, which is not conducive to the timely detection and prediction of possible insulation degradation problems in transformers. Summary of the Invention

[0003] The main purpose of the present invention is to provide a power transformer monitoring method, device, equipment and storage medium, which can improve the pertinence and accuracy of monitoring and provide more reliable data support for transformer operating status evaluation.

[0004] To achieve the above object, the present invention provides a power transformer monitoring method, comprising: Obtaining basic characteristic parameters and installation environment data of multiple power transformers, performing classification and matching based on the basic characteristic parameters and installation environment data, and obtaining sensor layout information; collecting real-time operating data of each of the power transformers according to the sensor arrangement information, performing equipment parallel monitoring, and obtaining equipment parallel monitoring data; Performing insulation degradation trend prediction on each of the power transformers based on the real-time working data and the equipment parallel monitoring data to obtain a single degradation prediction result; Collaborative management of multiple power transformer groups is performed based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions.

[0005] Furthermore, the obtaining of basic characteristic parameters and installation environment data of a plurality of power transformers, and performing classification matching based on the basic characteristic parameters and installation environment data to obtain sensor arrangement information includes: Obtaining equipment specifications, operating parameters, spatial layout data, equipment housing surface temperature distribution, and electromagnetic characteristic data of multiple power transformers, and constructing basic characteristics to obtain the basic characteristic parameters; Acquiring ambient temperature parameters, ambient humidity parameters, and electromagnetic interference parameters of multiple power transformers, and constructing installation environment characteristics to obtain the installation environment data; Performing fusion adaptation analysis on the basic characteristic parameters and the installation environment data to obtain a device adaptation distribution map; Performing compensation identification for power transformer contact monitoring blind areas on the equipment adaptation distribution map to obtain a three-dimensional coordinate deployment map; The device adaptation distribution map is read based on the three-dimensional coordinate deployment map to obtain the sensor layout information.

[0006] Furthermore, the device adaptation distribution map is constructed to compensate for the blind area of the power transformer contact monitoring to obtain a three-dimensional coordinate deployment map, including: Gridding the device adaptation distribution map to obtain monitoring area grid data; Performing a three-dimensional scan of the contact area of the power transformer according to the monitoring area grid data to obtain spatial parameters of the contact area; Identify the monitoring blind area of the spatial parameters of the contact area and perform multi-level decomposition to obtain hierarchical monitoring units; Calculating the sensor coverage radius of the hierarchical monitoring unit to obtain a sensor coverage group; Performing sensor density optimization on the monitoring area grid data according to the sensor coverage group to obtain density distribution data; performing sensor node redundancy optimization on the density distribution data to obtain a node optimization sequence; Sensor placement position constraint analysis is performed according to the node optimization sequence to obtain the three-dimensional coordinate deployment map.

[0007] Furthermore, the collecting of the real-time operating data of the power transformer and performing equipment parallel monitoring according to the sensor arrangement information and the real-time operating data to obtain equipment parallel monitoring data include: Partitioning the sensor layout information into monitoring point zones to obtain a monitoring area distribution map; According to the monitoring area distribution map, the sensor layout information is configured with monitoring point parameters, and data collection scheduling is performed to obtain a multi-point collection timing table; Performing cache management on the power transformer according to the multi-point acquisition timing table to obtain the real-time working data; Perform multi-dimensional parameter state identification on the power transformer according to the real-time working data to obtain equipment operating state parameters; The equipment operating status parameters are calculated in parallel, and multiple equipment are evaluated in real time based on a preset distributed computing framework to obtain the equipment parallel monitoring data.

[0008] Furthermore, the insulation degradation trend prediction of each power transformer is performed based on the real-time working data and the equipment parallel monitoring data to obtain a single degradation prediction result, including: Performing a single transformer parameter correlation analysis on the real-time working data based on the parallel monitoring data of the equipment to obtain an insulation state influencing factor; Classifying the insulation state influencing factors according to degradation characteristics to obtain an insulation degradation feature combination; Constructing a state transfer matrix based on the insulation degradation feature combination to obtain insulation performance change features; Performing data decomposition calculation on the insulation performance change characteristics to obtain insulation degradation change rules; Predicting insulation performance trends based on the insulation degradation variation law to obtain future insulation degradation development trends; A single evaluation is performed on each of the power transformers according to the future insulation degradation development trend to obtain the single degradation prediction result.

[0009] Furthermore, the insulation performance trend is predicted based on the insulation degradation variation law to obtain the future insulation degradation development trend, including: Performing an environmental load temperature correlation analysis on the insulation degradation variation law to obtain the insulation aging temperature dependence characteristics; Calculating the ratio of dissolved gas in the insulating oil of the power transformer according to the insulation aging temperature dependence characteristic to obtain a change rule of characteristic gas components in the insulating oil; Performing dielectric loss tangent analysis on the variation pattern of characteristic gas components of the insulating oil to obtain a degradation rate parameter; Calculating the insulation strength according to the degradation rate parameter to obtain an insulation strength attenuation curve; Extracting partial discharge characteristics from the insulation strength decay curve to obtain an insulation defect development state indicator; Calculating the winding deformation amount according to the insulation defect development state indicator to obtain a mechanical strength degradation warning value; Performing insulation breakdown field strength analysis on the mechanical strength degradation warning value to obtain the distribution of insulation weak points in the target area; A trend forecast is performed based on the distribution of insulation weak points to obtain the future insulation degradation development trend.

[0010] Furthermore, the collaborative management of multiple power transformer groups based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions includes: performing group feature clustering on the plurality of power transformers according to the equipment parallel monitoring data to obtain a transformer group operation feature matrix; performing correlation calculation on the single degradation prediction result according to the transformer group operation characteristic matrix to obtain a group degradation correlation index; Performing hierarchical decomposition on the group degradation correlation index to obtain a group degradation level evaluation result; performing load distribution analysis on the plurality of power transformers according to the group degradation level assessment result to obtain a group load distribution plan; Performing operation configuration matching on the group load distribution scheme to obtain group operation configuration parameters; Performing condition monitoring data fusion on a plurality of the power transformers according to the group operation configuration parameters to obtain a group comprehensive performance index; Maintenance resources are optimally configured according to the group comprehensive performance indicators to obtain the transformer maintenance decision.

[0011] The present invention further provides a power transformer monitoring device, which is applied to any one of the above-mentioned power transformer monitoring methods, comprising: An acquisition module is configured to acquire basic characteristic parameters and installation environment data of a plurality of power transformers, and perform classification and matching based on the basic characteristic parameters and installation environment data to obtain sensor arrangement information; An analysis module, configured to collect real-time operating data of each of the power transformers according to the sensor arrangement information, perform device parallel monitoring, and obtain device parallel monitoring data; A correlation module, configured to predict the insulation degradation trend of each power transformer based on the real-time operating data and the equipment parallel monitoring data to obtain a single degradation prediction result; A processing module is used to perform collaborative management of multiple power transformer groups based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions.

[0012] The present invention also provides a power transformer monitoring device, comprising: Memory, used to store programs; The processor is used to execute the program to implement each step of the power transformer monitoring method described in any one of the above.

[0013] The present invention also provides a storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute any of the above methods.

[0014] The present invention provides a power transformer monitoring method, device, equipment, and storage medium, which have the following beneficial effects: By categorizing and matching the basic characteristic parameters and installation environment data of multiple power transformers, differentiated monitoring plans can be developed based on the structural characteristics and environmental conditions of different transformer types. This overcomes the difficulty of traditional unified monitoring standards in adapting to diverse monitoring needs and improves the targeted and scientific nature of monitoring plans. Parallel device monitoring based on sensor layout information enables simultaneous data collection and analysis for multiple transformers, overcoming the limitations of traditional independent monitoring of individual devices and providing a data foundation for the integrated management of transformer groups. Insulation degradation trends are predicted by analyzing real-time operating data, enabling early detection of potential insulation problems and improving the accuracy and timeliness of fault warnings. Collaborative group management, combining parallel device monitoring data with individual degradation prediction results, enables coordinated analysis and optimized scheduling across multiple transformers, enhancing the efficiency and reliability of overall equipment management. Furthermore, by incorporating environmental factors into the optimized sensor layout, the accuracy and comprehensiveness of monitoring data are improved, providing a more reliable basis for decision-making regarding safe transformer operation and preventive maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a power transformer monitoring method provided by the present invention; Figure 2 This is a structural diagram of a power transformer monitoring device provided by the present invention; Figure 3 This is a structural diagram of a power transformer monitoring device provided by the present invention.

[0016] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0019] Reference Figure 1 As shown, the present invention provides a power transformer monitoring method, comprising: Step S1: acquiring basic characteristic parameters and installation environment data of multiple power transformers, performing classification and matching based on the basic characteristic parameters and installation environment data, and obtaining sensor layout information; Step S2: collecting real-time operating data of each power transformer according to the sensor layout information, performing equipment parallel monitoring, and obtaining equipment parallel monitoring data; Step S3: Predicting the insulation degradation trend of each power transformer based on the real-time working data and the equipment parallel monitoring data to obtain a single degradation prediction result; Step S4: Coordinated management of multiple power transformer groups is performed based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions.

[0020] Based on the above steps, the detailed steps are as follows: Step S1: The basic characteristic parameters of the power transformer include key information such as transformer capacity, voltage level, operating years, manufacturer, insulation structure, etc. The installation environment data covers environmental factors such as temperature, humidity, altitude, and degree of contamination. After obtaining these parameters through the data acquisition system, the transformers are grouped using a cluster analysis method. Based on similarity calculation, transformers with similar characteristics and environmental conditions are classified into the same category. For each category of transformer, the optimal sensor layout scheme is determined based on its characteristics, including sensor type, installation location, and quantity. The layout scheme takes into account the structural characteristics of the transformer to ensure that the sensors cover key monitoring points, such as winding temperature, oil level, gas content, etc. The classification and matching process uses a machine learning algorithm to establish a mapping relationship between characteristic parameters and sensor layout schemes, and realize the automatic generation of layout schemes. The output results of this step directly affect the collection quality of subsequent monitoring data.

[0021] Step S2: Based on the sensor layout information, various sensor devices are deployed to collect transformer operating data. The collected content includes multi-dimensional information such as electrical quantities (voltage, current, power), temperature field distribution, vibration signals, partial discharge, dissolved gas in oil, etc. A distributed data acquisition architecture is adopted to establish a real-time data transmission network. The acquisition frequency of each sensor is set according to the characteristics of the monitoring parameters to ensure the timeliness of the data. The parallel monitoring system simultaneously collects and processes data from multiple transformers, and uses edge computing technology for preliminary data filtering and feature extraction. Through the data fusion algorithm, the data of different types of sensors are time-aligned and correlated to form a complete description of the equipment status. A data quality assessment mechanism is established to mark and process abnormal data to ensure the reliability of monitoring data.

[0022] Step S3: Utilizing the acquired real-time operating data and equipment parallel monitoring data, a transformer insulation degradation prediction model is constructed. This prediction model comprehensively considers historical operating data, environmental factors, and load characteristics, employing a deep learning algorithm to establish the degradation development patterns under the influence of multiple factors. Time series analysis methods are used to extract the changing trends of degradation characteristic indicators and establish a degradation status assessment indicator system. Integrating the expert knowledge base, quantitative analysis results are combined with qualitative assessments to improve prediction accuracy. The prediction model considers the impact of changes in transformer operating conditions on the degradation process and dynamically adjusts prediction parameters. Monte Carlo simulation and other methods are used to assess the uncertainty of the prediction results and provide a confidence interval. The prediction results include key indicators such as the degree of insulation degradation, development rate, and remaining life.

[0023] Step S4: Based on the output results of steps S2 and S3, implement collaborative management of transformer groups. Establish a transformer health status evaluation system to convert single degradation prediction results into a unified health index. Through a multi-objective optimization algorithm, comprehensively consider factors such as equipment status, operating costs, and maintenance resources to formulate a group maintenance strategy. Establish a quantitative evaluation model for equipment status and compare the prediction results with the maintenance decision threshold. Determine the maintenance priority based on the transformer importance level and system operation requirements. Develop differentiated maintenance plans, including specific contents such as maintenance cycles, maintenance items, and spare parts configuration. Through collaborative optimization algorithms, achieve reasonable allocation of equipment maintenance resources within the group and improve maintenance efficiency. The maintenance decision results include specific implementation plans, resource requirements, and expected effect evaluation.

[0024] The present invention provides a power transformer monitoring method that analyzes transformer structural features and parameters to establish a differentiated monitoring parameter system. This allows for precise monitoring tailored to the characteristics of different transformer types, improving the pertinence and accuracy of monitoring and providing more reliable data support for transformer operating status assessment. By regionally matching installation environment data with the differentiated monitoring parameter system and optimizing sensor placement, the method achieves scientific configuration of monitoring equipment, improves data acquisition effectiveness, and avoids monitoring blind spots or resource waste caused by inappropriate sensor placement. The use of device parallel monitoring technology enables simultaneous monitoring and data analysis of multiple transformers, overcoming the limitations of traditional single-device monitoring. This provides a technical foundation for the collaborative management of transformer groups and improves the overall operational efficiency of the power system. Through comprehensive analysis of real-time operating data and device parallel monitoring data, a collaborative management mechanism for transformer groups is established. Combined with insulation degradation trend prediction, this mechanism enables timely identification of potential fault risks, provides a scientific basis for equipment maintenance decisions, effectively extends the transformer's service life, and reduces operation and maintenance costs. Based on differentiated monitoring and group collaborative management, comprehensive perception and intelligent early warning of transformer operating status are achieved, improving the reliability and security of the power system and providing a strong guarantee for the stable operation of the smart grid.

[0025] In one embodiment, basic characteristic parameters and installation environment data of a plurality of power transformers are obtained, and classification matching is performed based on the basic characteristic parameters and the installation environment data to obtain sensor arrangement information, including: During the power transformer monitoring process, the basic characteristic parameters and installation environment data of multiple power transformers are obtained, and the sensor layout information is obtained by classification and matching based on these data. The specific implementation method is as follows: The acquisition process of basic characteristic parameters involves the collection and construction of power transformer equipment specifications, operating parameters, spatial layout data, surface temperature distribution of the equipment casing, and electromagnetic characteristic data. The 3D scanning modeling of power transformer structures utilizes a variety of scanning technologies. Outdoor vacuum circuit breakers and box-type substations are externally scanned using high-precision laser scanners with a scanning accuracy of 0.1mm. Industrial CT scanning technology is used to image the internal structure of integrated primary and secondary equipment and high and low voltage equipment. Structured light scanning technology is used to capture microstructural features of high and low voltage components. The scanned data is fused from multiple perspectives using the ICP point cloud registration algorithm, and a complete 3D mesh model is generated using the Poisson reconstruction algorithm. The digital representation of the transformer structure includes information such as the equipment's geometric parameters, material properties, and connection relationships.

[0026] Multi-physics coupling analysis is performed based on the digital representation of the transformer structure. Electromagnetic field analysis uses the finite element method to solve Maxwell's equations and calculate the magnetic field distribution for different equipment types. For high-voltage equipment, the electric field intensity distribution in the insulating medium is analyzed, while for low-voltage equipment, the current density distribution in the conductor is analyzed. Thermal field analysis establishes a composite heat transfer model combining convection, conduction, and radiation. Mechanical stress analysis uses the structural mechanics finite element method to calculate the stress and strain state of the equipment under the influence of electromagnetic forces and thermal stresses.

[0027] Transformer physical characteristic data was analyzed using a multi-resolution wavelet transform for time-frequency analysis. A five-layer decomposition was performed using the db4 wavelet basis function. Statistical, time-domain, and frequency-domain features were extracted from each frequency band to form a feature vector. Analysis was performed by constructing a characteristic parameter association network. The Pearson correlation coefficient was used to calculate the strength of associations between parameters, and node topology indicators were calculated using complex network analysis methods. Community detection was performed on the network using a spectral clustering algorithm to generate transformer classification results.

[0028] Installation environment data is acquired using a spatial scanner to capture the spatial layout of the power transformer, with a resolution of 1mm and a scanning range covering the transformer body and a 3-meter area around it. An infrared thermal imager uses a 384×288 pixel array to capture the surface temperature gradient distribution of the equipment casing, with a temperature measurement accuracy of ±2°C. The humidity sensor array consists of 16 digital humidity sensors spaced 1 meter apart, with a sampling frequency of 1 time per minute. An electromagnetic interference detector collects electromagnetic interference spectrum characteristics, with a scanning frequency range of 0.1MHz-100MHz.

[0029] Fusion adaptation analysis uses a multi-source data fusion algorithm to correlate basic feature parameters with installation environment data. For each grid point in the device spatial feature matrix, the feature vector is extracted and weighted cosine similarity is calculated against each monitoring indicator. This calculation uses a sliding window approach with a window size of 5×5×5 grid points to form a device adaptation distribution map.

[0030] The blind spot compensation system for power transformer contact monitoring identifies and extracts areas with a degree of adaptation below 0.3 from the device adaptation distribution map as monitoring blind spots. Signal propagation paths are determined through field strength distribution calculations, and supplementary monitoring points are set where signal strength attenuation exceeds 20 dB. A ray tracing-based signal propagation model is established to calculate signal coverage connectivity at monitoring points, ensuring that at least one transmission path with a signal strength attenuation of less than 30 dB exists between any two points.

[0031] Sensor placement information is acquired by extracting information from the ultrasonic sensor array and infrared temperature measurement nodes. The ultrasonic sensor array consists of eight elements, and the infrared temperature measurement nodes use uncooled infrared detectors. A master-slave network architecture is used for time slot configuration, dividing the monitoring period into 1000 time slots. A dynamic parameter drift optimization model is established based on historical monitoring data, using a long-short-term memory network to predict parameter drift and dynamically adjust sensor positions.

[0032] The processing results include: basic characteristic parameter matrix, which records the basic characteristics of the transformer; installation environment characteristic vector, which describes the characteristics of the transformer operating environment; equipment adaptation distribution map, which shows the distribution of adaptation degree of monitoring points; three-dimensional coordinate deployment map, which marks the sensor installation location information; sensor layout information, which includes sensor three-dimensional coordinates, working parameter configuration, calibration cycle, etc.

[0033] The key technical indicators mentioned above include: basic characteristic parameter sampling frequency of no less than 1Hz, with an accuracy level of no less than 0.5; environmental parameter sampling interval of no more than 10 minutes; correlation coefficient threshold of 0.8 for fusion adaptation analysis; spatial resolution of blind spot compensation identification better than 0.1 meter; sensor node density of no less than one node per cubic meter; parameter drift prediction time window of 30 days; and position adjustment accuracy of 0.1mm. These technical indicators ensure the monitoring system's data collection accuracy and spatial coverage.

[0034] This embodiment uses a variety of scanning technologies to model the power transformer, combines multi-physics field coupling analysis and wavelet transform feature extraction, and achieves a comprehensive characterization of the physical characteristics of the transformer, providing a reliable data basis for differentiated monitoring. By constructing a characteristic parameter association network and spectral clustering analysis, accurate division of transformer types is achieved, making the monitoring parameter system more targeted. Based on spatial scanning and environmental data acquisition, combined with weighted cosine similarity calculation and sliding window analysis, an equipment adaptation distribution map is established to effectively identify monitoring blind spots. The sensor layout is optimized through a ray tracing model to ensure effective coverage of the monitoring point signal. The master-slave network architecture and parameter drift optimization model are used to achieve dynamic adjustment of the sensor position, improving the adaptability and reliability of the monitoring system. This method overcomes the problems of unreasonable sensor layout and difficulty in identifying monitoring blind spots in traditional monitoring schemes, and significantly improves the accuracy and comprehensiveness of power transformer monitoring.

[0035] In one embodiment, the equipment adaptation distribution map is constructed to compensate for the blind area of power transformer contact monitoring, and a three-dimensional coordinate deployment map is obtained, including: The device adaptation distribution map is a three-dimensional representation of the transformer equipment and its surrounding environment. Adaptive meshing is used for grid division, with finer meshes (5 cm × 5 cm × 5 cm) used in critical areas (such as near contacts) and coarser meshes (20 cm × 20 cm × 20 cm) used in non-critical areas. Grid cells are stored using an octree data structure, with each cell containing spatial coordinates, neighboring cell indices, and regional attribute identifiers. The monitoring area grid data is stored in a matrix format, with the matrix elements recording the attribute information of the grid cells.

[0036] A high-precision 3D laser scanner was used to set up multiple scanning stations around the transformer contacts, spaced 2 meters apart. Each station performed a 360-degree rotational scan. The scanning data acquisition frequency was 100kHz, and a single scan produced approximately 1 million point cloud data points. The acquired point cloud data was then registered, denoised, and reconstructed to generate a precise 3D model of the contact area. The contact area's spatial parameters include the contact's spatial position coordinates, geometric dimensions, surface features, and other data.

[0037] A multi-physics propagation model was constructed. An electromagnetic field model was established based on Maxwell's equations, and a temperature field model was established using the heat conduction equation. Finite element methods were used for numerical solution. The model considered the electromagnetic (electrical conductivity, magnetic permeability) and thermal (thermal conductivity, specific heat capacity) properties of the contact metal material, as well as the propagation characteristics of the air medium. An adaptive meshing method was used, with mesh refinement in areas with drastic changes in the electromagnetic and temperature fields to ensure computational accuracy.

[0038] Next, an improved ray tracing process is performed. A ray source is set up at each potential sensor location, and 1000 rays are initially launched in an importance sampling distribution within a 4π solid angle. The ray propagation paths are determined by solving a modified Eikonal equation that accounts for multi-physics effects. The ray energy attenuation model comprehensively considers geometric diffusion losses, material absorption losses, and the spatial distribution of electromagnetic field intensity.

[0039] Grid penetration statistics are then performed. The monitoring area is divided into adaptively sized statistical grids, whose size is dynamically adjusted based on the local field intensity gradient. The intersection of rays and the grid is recorded, and a multidimensional feature vector is constructed, containing information such as the number of ray penetrations, electric field strength, magnetic field strength, and temperature distribution. A comprehensive threshold based on multi-physics field characteristics is set, and grids below this threshold are marked as monitoring blind spots.

[0040] Adaptive density clustering based on electromagnetic field characteristics replaces Euclidean distance with a characteristic distance that accounts for field intensity attenuation, and the density threshold is adaptively adjusted based on the field intensity distribution. The temperature field is also introduced as a clustering weight, giving hotspots higher clustering priority. Next, hierarchical clustering coupled with multiple physical fields is performed to establish a mapping between field intensity and monitoring difficulty, automatically constructing a hierarchical blind spot structure. Finally, constrained clustering based on monitoring effectiveness is implemented, integrating monitoring coverage requirements, sensor blind spots, and device structural characteristics as clustering constraints.

[0041] Establish hierarchical monitoring units. Based on clustering results, calculate the comprehensive characteristics of blind spot clusters, including spatial geometry (volume, center of gravity), physical field characteristics (field intensity distribution, temperature distribution), and monitoring characteristics (coverage difficulty, priority). Based on the similarity and spatial correlation of multidimensional features, construct an adaptive hierarchical tree structure. Each node in the tree represents a monitoring unit of different scales, recording unit attributes and hierarchical relationships.

[0042] Form hierarchical monitoring units. Allocate monitoring resources and strategies to each hierarchical monitoring unit, including sensor type selection, installation location optimization, and data collection plan. Establish a collaborative mechanism between monitoring units to achieve multi-scale, multi-layered integrated monitoring. Output includes complete information about the monitoring units, providing decision support for subsequent sensor network optimization and monitoring system deployment.

[0043] A finite element model is created for each layered monitoring unit to simulate the detection characteristics of different sensor types (such as magnetic field sensors and temperature sensors). The signal strength attenuation patterns of sensors at different installation angles and distances are calculated. The effective monitoring range is determined based on sensor performance indicators (sensitivity, signal-to-noise ratio, etc.). The calculated results are integrated into a sensor coverage group, which contains spatial coverage data for each installation location.

[0044] The sensor coverage group data is mapped onto the monitoring area grid data, and the coverage count for each grid cell is calculated. A genetic algorithm is used to optimize the sensor distribution, with the objective function including coverage rate, repeat coverage, and number of sensors. Through multiple iterations of optimization, density distribution data is generated, describing the required sensor installation density for different areas.

[0045] A sensor coverage relationship graph is constructed based on density distribution data. Nodes in the graph represent sensor locations, and edges represent overlapping coverage areas. A minimum dominating set algorithm is used to remove redundant nodes to ensure that the coverage of the remaining nodes meets monitoring requirements. Nodes are ranked based on their importance (coverage area, number of blind spots, etc.) to generate an optimized node sequence.

[0046] A structural constraint model of the transformer is established, including geometric information such as the device housing, internal components, and connectors. Collision detection is performed on locations in the node optimization sequence to eliminate locations that do not meet installation space requirements. The impact of electromagnetic interference between sensors is analyzed to ensure minimum spacing between adjacent sensors. Maintenance accessibility at the installation location is assessed, taking into account factors such as maintenance access and operating space. Finally, a 3D coordinate deployment map is generated, including the precise sensor installation coordinates, installation orientation, and coverage area.

[0047] This embodiment, by constructing a multi-physics field propagation model and an improved ray tracing method, can fully consider the coupling effects of the electromagnetic field and the temperature field, achieve accurate identification of the contact area monitoring blind spots, and improve the reliability of the monitoring system. Through the innovative adaptive density clustering algorithm, multi-dimensional features such as electromagnetic field characteristics and temperature distribution are integrated into the clustering process, so that the blind spot identification results are more in line with actual monitoring needs and the pertinence of the monitoring scheme is improved. By establishing hierarchical monitoring units and layered monitoring schemes, the optimal configuration of monitoring resources is achieved, which not only ensures the monitoring effect of key areas, but also avoids excessive redundant deployment of sensors and reduces system costs. Through the fusion of multi-physics field coupling analysis and engineering practice constraints, the monitoring system has stronger adaptability and reliability in actual applications, providing more scientific technical support for transformer contact status monitoring.

[0048] In one embodiment, real-time operating data of the power transformer is collected, and equipment parallel monitoring is performed based on sensor layout information and the real-time operating data to obtain equipment parallel monitoring data, including: Monitoring point zoning is based on the physical structure of the power transformer, dividing the transformer surface into functional units such as the tank area, cooler area, bushing area, and tap changer area. A set of boundary coordinate points is defined for each area, and a polygonal region partitioning algorithm is used to calculate the area of each area. The sensor density within the area is calculated by counting the number of sensors per unit area. The system integrates the information for each area into a monitoring area distribution map data structure, which includes attribute fields such as area identifier, boundary coordinate set, area value, and sensor density.

[0049] Differentiated parameter configurations were implemented for each monitoring point based on the monitoring area distribution map. The temperature sensor sampling frequency was set to 5 minutes, with a data accuracy of 0.1°C and a temperature alarm threshold of 85°C. The vibration sensor sampling frequency was set to 1 minute, with a data accuracy of 0.01g and a vibration alarm threshold of 2g. The noise sensor sampling frequency was set to 10 minutes, with a data accuracy of 0.1dB and a noise alarm threshold of 75dB. The data collection scheduling mechanism accessed each monitoring point sequentially in a round-robin manner, writing the collected data, along with information such as timestamp, device number, and sensor type, into a multi-point collection sequence table.

[0050] A distributed cache architecture is used to store real-time working data, sharding it by device ID. Each data shard contains data collected by all sensors on that device within a specified time window. Data shards are replicated across multiple cache nodes, using a master-slave replication mechanism to ensure data consistency. A cache eviction strategy uses a least-recently-used (LRU) algorithm to clear historical data that falls outside the time window. The cache query interface supports filtering data by time range, sensor type, and other criteria.

[0051] The state recognition algorithm preprocesses real-time operating data, including steps such as data cleaning, outlier processing, and data normalization. The feature extraction module calculates characteristic quantities such as the temperature change rate, vibration spectrum, and noise power spectrum. The pattern recognition module, based on a pretrained device state model, maps the extracted characteristics into device operating state parameters. These parameters include load factor (the ratio of the device's actual load to its rated capacity), temperature rise rate (temperature change per unit time), and vibration amplitude (the effective value of the vibration signal).

[0052] The distributed computing framework utilizes a master-slave architecture, consisting of a master node (MasterNode) and multiple compute nodes (Compute Nodes). The MasterNode is responsible for distributing and scheduling power transformer monitoring tasks, while the Compute Nodes are responsible for executing specific transformer condition assessment tasks. The framework also includes a distributed storage system for storing power transformer operating data and assessment results.

[0053] The master control node consists of three core components: a task scheduler, a resource manager, and a status monitor. The task scheduler is responsible for breaking down the power transformer status assessment task into subtasks and assigning tasks based on the load of the computing nodes. The resource manager monitors the computing resource usage of each computing node, including CPU utilization and memory usage. The status monitor monitors the execution status of the transformer assessment task in real time and handles abnormal situations such as task failure. The computing node consists of three core components: a task executor, a data processor, and a result collector. The task executor receives and executes the assigned transformer assessment tasks, supporting multi-threaded parallel processing. The data processor is responsible for operations such as preprocessing and feature calculation of transformer operating data. The result collector summarizes the transformer status assessment results and returns them to the master control node.

[0054] The parallel computing process is implemented through both data parallelism and task parallelism. Data parallelism shards the operating status parameters of multiple power transformers according to transformer ID, with each computing node processing a portion of the transformer data. Task parallelism decomposes the status assessment task for each transformer into subtasks such as normalized status parameter calculation, transformer health score calculation, fault risk assessment, and operating trend analysis.

[0055] The real-time assessment execution process is divided into data preparation, parallel computing, and result aggregation. The data preparation phase reads power transformer operating status parameters from the distributed storage system, shards the data by transformer ID, and distributes it to compute nodes. In the parallel computing phase, the compute nodes perform standardized processing on the assigned transformer data, calculate transformer health scores based on a pre-set assessment model, execute fault diagnosis algorithms to identify potential risks, and calculate transformer performance trends. In the result aggregation phase, the compute nodes return the processed results to the master control node, which merges the results from multiple compute nodes to generate the final parallel monitoring data for multiple power transformers.

[0056] The assessment result data structure includes basic transformer information, condition assessment results, and trend analysis results. Basic transformer information includes transformer ID, transformer type, and installation location; condition assessment results include health score, fault risk level, and key parameter anomaly flags; trend analysis results include performance degradation rate, expected remaining life, and maintenance recommendations.

[0057] The reliability of transformer assessment calculations is ensured through mechanisms such as task retry upon task failure, data consistency checks, load balancing, and real-time monitoring. When a compute node fails, tasks are automatically reallocated; data consistency checks ensure the accuracy of transformer assessment results; load balancing dynamically adjusts task allocation to avoid compute node overload; and real-time monitoring monitors the transformer assessment process to promptly identify and address anomalies.

[0058] This embodiment achieves efficient parallel monitoring of power transformer groups by adopting a distributed computing framework with a master-slave architecture. The task scheduler of the master node can flexibly allocate evaluation tasks according to the number of transformers and the load of the computing nodes, thereby improving the scalability of the system's processing capabilities. Through the dual mechanisms of data parallelism and task parallelism, the system can simultaneously process the status evaluation of multiple transformers, significantly improving the monitoring efficiency. The computing nodes adopt a multi-threaded parallel processing mechanism to fully utilize computing resources and speed up the transformer status evaluation. Reliability assurance mechanisms such as task failure retry and data consistency check are set to ensure the accuracy of the transformer evaluation results and the stability of the system operation. The transformer operation data is managed through a distributed storage system, achieving efficient reading and writing and secure storage of data. The framework supports dynamic expansion of computing nodes to adapt to the monitoring needs of transformer groups of different sizes and has good system adaptability.

[0059] In one embodiment, insulation degradation trend prediction is performed on each power transformer based on real-time operating data and equipment parallel monitoring data to obtain a single degradation prediction result, including: During the single transformer parameter correlation analysis phase, preprocessing of the equipment's parallel monitoring data and real-time operating data involves data normalization and outlier removal. Data normalization unifies data of varying dimensions to the same scale, while outlier removal uses the 3σ criterion to remove significantly deviating data points. During parameter correlation analysis, a correlation coefficient matrix is calculated between each monitoring parameter, using a monthly timeframe. This matrix reflects the strength of correlation between parameters. By setting thresholds, significantly correlated parameter pairs are selected, which constitute the factors influencing insulation condition.

[0060] During the degradation feature classification phase, features of factors influencing insulation status are extracted. This feature extraction includes time-domain features (mean, variance, and peak) and frequency-domain features (spectral energy distribution). Based on the extracted features, a classification distance metric is set to cluster factors with high feature similarity. During the clustering process, a minimum inter-class distance threshold is set, and classes are merged when the inter-class distance falls below the threshold. The classification results form multiple feature subsets, each representing a type of degradation feature. Together, all subsets constitute the insulation degradation feature set.

[0061] During the state transition matrix construction phase, insulation degradation characteristics are combined and quantified into discrete state values. The state values are divided into multiple state intervals using an equidistant approach. The frequency of state value transitions between adjacent time points is counted to calculate the state transition probability. The rows of the state transition probability matrix represent the current state, the columns represent the next state, and the matrix element values represent the state transition probability. This matrix is used to analyze the insulation performance variation characteristics.

[0062] During the data decomposition phase, a multi-scale analysis of insulation performance variation characteristics is performed. Local extreme points in the data sequence are found, upper and lower envelopes are constructed, and the mean envelope is calculated. The mean envelope is subtracted from the original signal to obtain the intrinsic mode components. This process is repeated until the remaining signal becomes a monotonic function. Each component represents variation characteristics at different time scales. Comprehensive analysis of these characteristics reveals the law of insulation degradation variation.

[0063] During the insulation performance trend prediction phase, a prediction model is established based on the patterns of insulation degradation. This model uses an additive generation sequence to reduce data randomness and establishes a differential equation to describe data trends. The differential equation is solved to obtain the predicted value, which is then restored to the original sequence through cumulative subtraction. This prediction represents the future trend of insulation degradation.

[0064] During the single assessment phase, a multi-level evaluation indicator system is constructed. The evaluation indicators include the insulation degradation rate (reflecting the rate of degradation), the degradation degree (reflecting the current level of degradation), and the remaining service life (reflecting the remaining service life). Each indicator is assigned a weight, determined using the analytic hierarchy process. The evaluation value for each indicator is calculated, and the weighted sum of the evaluation value and the weight is calculated to obtain a comprehensive evaluation score. This score serves as the single degradation prediction result and is used to characterize the transformer's insulation degradation status.

[0065] The construction and evaluation process of the multi-level evaluation indicator system in a single evaluation phase are as follows: The multi-level evaluation index system adopts a three-tiered design: target layer, criterion layer, and indicator layer. The target layer assesses the insulation degradation status of the transformer, the criterion layer includes insulation degradation rate, degradation degree, and remaining life, and the indicator layer includes specific evaluation indicators.

[0066] The insulation degradation rate criterion uses four evaluation indicators: the growth rate of dissolved gas in oil, the change rate of dielectric loss tangent, the insulation resistance drop rate, and the partial discharge growth rate. These indicators reflect the dynamic changes in insulation degradation.

[0067] The degradation criteria include five evaluation indicators: dissolved gas content in oil, dielectric loss tangent, insulation resistance, partial discharge, and water content in oil. These indicators reflect the current state of insulation degradation.

[0068] The remaining life criterion sets three evaluation indicators: thermal aging of insulation paper, degradation of insulation oil, and load history. These indicators reflect the cumulative loss of the insulation system.

[0069] The indicator weighting process uses the Analytic Hierarchy Process (AHP) to determine the weights of indicators at each level by constructing a judgment matrix. This judgment matrix is based on expert experience and historical data analysis, and uses a 1-9 scale to compare the importance of each indicator pairwise. The relative weight of each indicator is determined by calculating the eigenvalues and eigenvectors of the judgment matrix.

[0070] During the evaluation calculation process, scoring criteria are set for each specific indicator. These criteria are determined based on relevant technical specifications and historical operational experience. The indicator values are divided into different grading intervals, with each interval corresponding to a specific score. The interval and corresponding score for each indicator are determined based on actual monitoring data.

[0071] The final evaluation result is calculated through a comprehensive weighted approach. For each criterion-level indicator, the scores of its subordinate indicator layers are multiplied by the corresponding weights and summed to obtain the criterion-level score. The criterion-level score is then multiplied by the criterion-level weights and summed to obtain the final comprehensive evaluation score. This score reflects the overall state of transformer insulation degradation, with higher scores indicating better insulation condition.

[0072] The evaluation results are divided into four levels: Excellent (90-100 points), Good (75-89 points), Fair (60-74 points), and Poor (under 60 points). Different levels correspond to different O&M recommendations: For Excellent, maintain normal operation; for Good, increase monitoring frequency; for Fair, develop a preventive maintenance plan; and for Poor, perform timely repairs or replacements.

[0073] In this embodiment, by performing correlation analysis on the real-time working data of the power transformer and the parallel monitoring data of the equipment, the accurate identification of the factors affecting the insulation state is achieved, and the accuracy of the transformer insulation state assessment is effectively improved. The fuzzy clustering algorithm is used to classify the factors affecting the insulation state, forming a systematic insulation degradation feature combination, which enhances the reliability of feature identification. By constructing a state transition matrix, the dynamic tracking of the insulation performance change characteristics is achieved, and the scientific nature of the insulation degradation trend prediction is improved. The empirical mode decomposition method is used to analyze the insulation performance change characteristics, revealing the inherent laws of insulation degradation and providing a reliable basis for the establishment of a prediction model. Based on a comprehensive evaluation of the multi-level evaluation index system, a comprehensive evaluation of the transformer insulation degradation state is achieved, providing scientific guidance for operation and maintenance decisions. This method overcomes the limitations of the traditional single monitoring method, improves the comprehensiveness and accuracy of the transformer insulation state assessment, and has important guiding significance for preventive maintenance.

[0074] In one embodiment, the group insulation performance trend is predicted based on the group insulation degradation variation law to obtain the future insulation degradation development trend, including: During the ambient load-temperature correlation analysis phase, operating temperature data for the transformer group was collected, including oil temperature, winding temperature, and ambient temperature. A mathematical model linking temperature and insulation aging rate was established using the Arrhenius equation. This model reflects the characteristic that the insulation aging rate doubles with every 8-10°C increase in temperature. Based on this mathematical model, the temperature dependence of insulation aging was derived, which characterizes the impact of temperature on insulation life.

[0075] The calculation of dissolved gas ratios in insulating oil is based on the IEC60599 standard. Gas chromatography is used to measure characteristic gases dissolved in transformer oil, including hydrogen, methane, ethane, ethylene, and acetylene. Three ratios are calculated: acetylene / ethylene, methane / hydrogen, and ethylene / ethane. The temporal evolution of these ratios is analyzed. The trends in these ratios reflect the type and severity of internal transformer faults.

[0076] The dielectric loss tangent (tanδ) analysis uses a capacitance bridge measurement method. The dielectric loss tangent of insulating oil is measured under standard test voltage and temperature conditions. This value characterizes the loss characteristics of insulating oil under an alternating electric field. By analyzing the time series of tanδ values, a degradation rate parameter model is established that quantifies the rate of degradation of insulating oil performance over time.

[0077] Among them, the degradation rate parameter model in the dielectric loss tangent analysis is a multi-level evaluation system, which quantitatively analyzes the degradation rate of insulating oil performance through multiple dimensions.

[0078] At the model's foundational level, tan delta values obtained through capacitance bridge measurements serve as raw data input. These measurements are performed under standard conditions, with a test voltage set at 2000V and a temperature controlled at 90±0.5°C. Each measurement is taken every 30 days, with no less than 24 months of data collected continuously to form a time series dataset.

[0079] At the data processing level, the raw tanδ values are decomposed into a time series to extract three components: trend, periodic, and random. The trend term reflects the long-term direction of changes in insulating oil performance, the periodic term reflects the influence of periodic factors such as ambient temperature, and the random term includes measurement errors and occasional fluctuations.

[0080] At the feature extraction level, four key characteristic parameters are extracted from the decomposed data: average growth rate, fluctuation amplitude, acceleration coefficient, and stability index. The average growth rate describes the overall upward trend of the tanδ value, the fluctuation amplitude reflects the degree of dispersion of the values, the acceleration coefficient characterizes the nonlinear characteristics of the degradation process, and the stability index assesses the reliability of the data.

[0081] At the model construction level, these four characteristic parameters are integrated into a comprehensive degradation rate indicator. This indicator uses a weighted average method, with a weight of 0.4 for the average growth rate, 0.3 for the acceleration coefficient, 0.2 for the fluctuation range, and 0.1 for the stability index. The weighting is determined based on historical operating experience and expert evaluation results.

[0082] The assessment criteria for the comprehensive degradation rate are divided into five levels: normal (less than 0.005 / year), mild degradation (0.005-0.01 / year), moderate degradation (0.01-0.015 / year), severe degradation (0.015-0.02 / year), and serious degradation (greater than 0.02 / year). These thresholds are determined based on statistical analysis of extensive field data.

[0083] At the temporal level, based on the current degradation rate level and combined with the changing patterns of historical data, we predict the trend of tanδ values over a certain period of time. This prediction uses a sliding time window approach with a 12-month window length, rolling forward one month at a time, to dynamically track the degradation process.

[0084] At the early warning mechanism level, three levels of warning thresholds are set. Warnings are triggered when the predicted tanδ value is likely to reach a higher level of degradation within the next six months. A yellow warning corresponds to a transition from mild to moderate degradation, an orange warning corresponds to a transition from moderate to severe degradation, and a red warning corresponds to a transition from severe to severe degradation.

[0085] At the correction and update level, the model possesses adaptive learning capabilities. Each time new measurement data is acquired, the characteristic parameters and weight coefficients are automatically updated to ensure that the model's predictions are more consistent with the actual degradation process. The correction process utilizes a sliding average algorithm to ensure a balance between the model's sensitivity to new data and the stability of historical data.

[0086] This multi-level model structure enables accurate quantitative assessment of the insulating oil degradation rate, providing reliable data support for transformer maintenance decisions. The model not only reflects the current degradation state but also predicts future trends, demonstrating its practical value.

[0087] Insulation strength is calculated using a breakdown voltage test. Under standard clearances, the breakdown voltage of the insulating oil is measured, and an insulation strength decay curve is plotted based on the test results. This curve reflects the change in the insulating oil's electrical strength over time, providing a basis for assessing the remaining life of the insulation system.

[0088] Partial discharge feature extraction utilizes ultra-high frequency (UHF) detection technology. UHF sensors installed on the transformer collect partial discharge signals. Time-frequency analysis is performed on the signals to extract characteristic parameters such as discharge amplitude, repetition rate, and phase distribution. This allows for a comprehensive assessment of the development of insulation defects and the formation of a status indicator system.

[0089] Winding deformation calculation is based on frequency response analysis (FRA). Using swept-frequency excitation, the transformer winding's transfer function characteristics are measured. By comparing and analyzing frequency response curves at different times, winding deformation is calculated and a threshold for mechanical strength degradation is set. When the deformation exceeds the threshold, it indicates that the winding's mechanical strength has reached the warning level.

[0090] Frequency response analysis is a nondestructive testing technique that applies a swept-frequency signal to a transformer winding and measures the amplitude ratio and phase difference between the input and output signals to determine the winding's frequency response. The measurement frequency range is typically 20 Hz to 2 MHz, covering low, medium, and high frequencies.

[0091] During the measurement process, a sweep signal is generated by a signal generator, with the signal amplitude kept below 5V to avoid damage to the windings. Connections are made end-to-end, with both the high-voltage and low-voltage windings measured separately. The measured data includes amplitude-frequency and phase-frequency curves, which together constitute the transfer function of the transformer windings.

[0092] Frequency response curve analysis is divided into three frequency bands: the low frequency band (20Hz-5kHz) reflects the magnetic circuit characteristics of the winding; the mid-frequency band (5kHz-500kHz) reflects the capacitive coupling characteristics of the winding; and the high frequency band (500kHz-2MHz) reflects the local resonance characteristics of the winding. Different types of winding deformation will produce characteristic changes in different frequency bands.

[0093] Comparative analysis uses the benchmark curve method to compare the currently measured frequency response curve with the curve obtained during the initial operation of the transformer or during the last test. Analysis includes characteristics such as curve shape changes, resonance point shifts, and amplitude differences. The correlation coefficient and frequency band deviation coefficient are two important evaluation indicators.

[0094] Calculation of winding deformation is based on deviation analysis of the frequency response curve. At low frequencies, curve deviation primarily reflects axial deformation; at mid-frequency ranges, deviation reflects radial deformation; and at high frequencies, deviation reflects localized deformation. By establishing a corresponding relationship between frequency response characteristics and deformation, quantitative calculation of deformation is achieved.

[0095] Mechanical strength degradation warning thresholds are set using a tiered system. For axial deformation, a correlation coefficient below 0.6 is considered severe; for radial deformation, a frequency band deviation exceeding 5dB is considered severe; and for localized deformation, a resonance point offset exceeding 10% is considered severe. If any of these indicators reaches the severe deformation level, a warning is triggered.

[0096] Deformation monitoring is performed regularly, with the inspection cycle determined by the transformer's operating conditions, typically six months to one year. Important transformers should be promptly inspected after experiencing a short-circuit shock. The inspection results are incorporated into the transformer condition assessment system and serve as a key basis for maintenance decisions.

[0097] Data processing is performed using specialized software with features such as curve smoothing, noise filtering, and feature extraction. Analysis results are output in graphical and numerical form, including deformation values, trend graphs, and warning status information. This information provides an intuitive basis for assessing the transformer's mechanical condition.

[0098] The testing environment requires strict control. The transformer should be completely out of service during measurement, and the measurement environment temperature should be stable to avoid electromagnetic interference. Regular calibration of measuring instruments ensures measurement accuracy. Ensure good grounding during measurement to ensure data reliability.

[0099] Complete test records are maintained, including measurement conditions, raw data, and analysis results. A historical database is established for long-term trend analysis. Through statistical analysis of historical data, early warning thresholds are continuously optimized to improve early warning accuracy.

[0100] The formula for calculating the winding deformation includes: ; is the feature point correlation coefficient; is the amplitude of the characteristic point of the reference curve; is the amplitude of the characteristic point of the current measurement curve; and are the arithmetic mean of the amplitudes of the feature points; k is the number of feature points.

[0101] Feature point selection rules: resonance peak point, resonance valley point, curve inflection point, and frequency band boundary point.

[0102] when When it is less than 0.6, it indicates that the winding is severely deformed.

[0103] The finite element method is used to analyze the insulation breakdown field strength. A three-dimensional model of the transformer insulation structure is constructed to calculate the electric field distribution. By analyzing areas of high field strength, the spatial distribution of insulation weaknesses is determined. These weak points are potential locations for insulation breakdown, posing a threat to the safe operation of the transformer.

[0104] Trend prediction utilizes time series analysis. Using the aforementioned analysis results as input variables, a multivariate prediction model is developed. This model comprehensively considers factors such as temperature dependence, gas ratio, loss characteristics, strength decay, discharge characteristics, and mechanical deformation to predict insulation performance trends over a specific period. This model identifies future insulation degradation trends, including insulation life estimation, failure risk assessment, and maintenance recommendations.

[0105] This embodiment achieves an accurate quantitative assessment of the degradation rate of insulating oil performance by establishing a multi-level dielectric loss tangent value analysis model. The model integrates multiple functional modules such as data processing, feature extraction, and comprehensive evaluation, significantly improving the accuracy and reliability of the assessment. By setting up a hierarchical early warning mechanism, the trend of insulation performance degradation can be discovered in a timely manner, providing a scientific basis for preventive maintenance. The winding deformation calculation method based on frequency response analysis can accurately judge the type and degree of transformer winding deformation by analyzing the characteristics of different frequency bands. This method uses non-destructive testing technology to avoid the potential damage to the equipment caused by traditional testing methods. At the same time, through strict environmental control and data processing, the reliability of the measurement results is ensured. By establishing a complete historical database and dynamically optimized early warning thresholds, the transformer status assessment is made more forward-looking, effectively reducing the risk of equipment failure and extending the service life of the transformer.

[0106] In one embodiment, multiple power transformer groups are collaboratively managed based on the equipment parallel monitoring data and the single degradation prediction results to obtain transformer maintenance decisions, including: Real-time operating data includes parameters that directly reflect the transformer's current operating status, such as load current, voltage, oil temperature, and winding temperature. Parallel equipment monitoring data includes historical operating data, ambient temperature, humidity, vibration data, and past fault records.

[0107] Clustering of multiple power transformers was performed based on their group characteristics. The collected data was preprocessed based on parallel monitoring data from the equipment to remove outliers and noise, and the data was standardized to ensure comparability across different dimensions. Next, a K-means clustering algorithm was used to perform cluster analysis on the preprocessed data, grouping transformers with similar operating characteristics. This generated a transformer group operating characteristic matrix. This matrix includes key operating parameters such as the average load factor, average oil temperature, and average winding temperature of the transformers in each group.

[0108] The correlation between the individual degradation prediction results and the transformer group's operational characteristic matrix is calculated to obtain a group degradation correlation index. This index is derived from the Pearson correlation coefficient and reflects the correlation between the degradation states of the individual transformers in the group. Its value range is [-1, 1], where positive values indicate positive correlation and negative values indicate negative correlation. Larger absolute values indicate stronger correlation.

[0109] The group degradation correlation index is hierarchically decomposed using the Analytic Hierarchy Process (AHP) to obtain the group degradation level assessment results. The specific process is as follows: a judgment matrix is constructed and obtained by comparing the relative importance of the degradation status correlation indicators of each transformer in the group. Based on practical experience and data analysis results, experts compare the correlation between each two transformers and fill in the judgment matrix. The eigenvector of the judgment matrix is calculated. The eigenvector represents the weight of each transformer in the group degradation state. The group degradation level assessment results are calculated based on the eigenvector, and the transformer group is divided into three levels: severe degradation (Grade A), moderate degradation (Grade B), and mild degradation (Grade C).

[0110] Based on the group degradation level assessment results, a load distribution analysis was conducted on multiple power transformers to generate a group load allocation plan. This load distribution analysis aims to optimize the load configuration of the transformers and minimize overall operating costs. A particle swarm optimization algorithm was used to find the optimal load allocation plan by simulating the movement of particles in the solution space. This load allocation plan takes into account the capacity, efficiency, and degradation level of the transformers to ensure optimal load distribution for each transformer.

[0111] The group load distribution scheme is matched with the operational configuration to obtain the group operational configuration parameters. The operational configuration parameters include operational constraints such as the load limit and temperature limit of each transformer, which are used to guide the coordinated operation of the transformer group.

[0112] Condition monitoring data from multiple power transformers is fused based on group operating configuration parameters to generate comprehensive group performance indicators. Data fusion technology is used to integrate multiple monitoring data sources, including temperature, vibration, and partial discharge. The data fusion process includes data preprocessing, feature extraction, and the establishment of a data fusion model. Combined with the group operating configuration parameters, comprehensive group performance indicators are calculated to reflect the overall operating status of the transformer group. These comprehensive performance indicators include the group's overall load factor, overall temperature status, and overall vibration status.

[0113] Maintenance resources are optimally allocated based on the group's comprehensive performance indicators to arrive at transformer maintenance decisions. The maintenance resource optimization process includes the following steps: First, the types and quantities of maintenance resources, including manpower, equipment, and materials, are determined. Then, based on the group's comprehensive performance indicators, the health of the transformers is assessed, and transformers requiring priority maintenance are identified. Next, a maintenance plan is developed, determining the maintenance cycle and maintenance methods for each transformer. Finally, maintenance resources are allocated to ensure that maintenance tasks can be carried out as planned. Transformer maintenance decisions are formed by comprehensively considering the availability of maintenance resources, the urgency of the maintenance task, and the health of the transformer. This decision provides a scientific basis for preventive maintenance of the transformer group, ensuring its reliable operation.

[0114] This embodiment achieves comprehensive monitoring of the operating status of the transformer group through comprehensive analysis of equipment parallel monitoring data and real-time working data, thereby improving the accuracy of status assessment. The degradation correlation index of the group is decomposed through the hierarchical analysis method, and a scientific transformer degradation level assessment system is established, providing a reliable basis for maintenance decision-making. The particle swarm optimization algorithm is used for load distribution to achieve the optimal operating configuration of the transformer group and reduce the overall operating cost. The multi-source monitoring data is processed using data fusion technology to improve the accuracy and reliability of status assessment. Through the optimal allocation of maintenance resources, scientific decision-making for preventive maintenance is achieved, the operation of equipment with defects and excessive maintenance are avoided, the service life of the transformer is extended, and the operational reliability and economy of the equipment are improved.

[0115] Reference Figure 2 As shown, the present invention further provides a power transformer monitoring device, which is applied to any of the above-mentioned power transformer monitoring methods, comprising: An acquisition module is used to obtain basic characteristic parameters and installation environment data of multiple power transformers, classify and match the basic characteristic parameters and installation environment data, and obtain sensor layout information; An analysis module is used to collect real-time operating data of each power transformer based on sensor layout information, perform equipment parallel monitoring, and obtain equipment parallel monitoring data; The correlation module is used to predict the insulation degradation trend of each power transformer based on real-time working data and equipment parallel monitoring data to obtain a single degradation prediction result; The processing module is used to collaboratively manage multiple power transformer groups based on equipment parallel monitoring data and single degradation prediction results to obtain transformer maintenance decisions.

[0116] The present invention provides a power transformer monitoring device that analyzes transformer structural features and parameters to establish a differentiated monitoring parameter system. This allows for precise monitoring tailored to the characteristics of different transformer types, improving the pertinence and accuracy of monitoring and providing more reliable data support for transformer operating status assessment. By regionally matching installation environment data with the differentiated monitoring parameter system and optimizing sensor placement, the device achieves scientific configuration of monitoring equipment, improves data acquisition effectiveness, and avoids monitoring blind spots or resource waste caused by inappropriate sensor placement. The device's parallel monitoring technology enables simultaneous monitoring and data analysis of multiple transformers, overcoming the limitations of traditional single-device monitoring. This provides a technical foundation for the coordinated management of transformer groups and improves the overall operational efficiency of the power system. Through comprehensive analysis of real-time operating data and device parallel monitoring data, a transformer group coordinated management mechanism is established. Combined with insulation degradation trend prediction, this mechanism enables timely identification of potential fault risks, provides a scientific basis for equipment maintenance decisions, effectively extends transformer service life, and reduces operation and maintenance costs. Based on differentiated monitoring and group coordinated management, comprehensive perception and intelligent early warning of transformer operating status are achieved, improving the reliability and security of the power system and providing a strong guarantee for the stable operation of the smart grid.

[0117] Reference Figure 3 As shown, the present invention also provides a power transformer monitoring device, comprising: Memory, used to store programs; The processor is used to execute the program to implement each step of any one of the above-mentioned power transformer monitoring methods.

[0118] In this embodiment, the processor and memory may be connected via a bus or other means. The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, a digital signal processor, an application-specific integrated circuit, or one or more integrated circuits configured to implement the embodiments of the present invention.

[0119] The present invention also provides a storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute any of the above methods.

[0120] It should be noted that, those skilled in the art will clearly understand that, for the sake of convenience and brevity of description, the specific working processes of the above-described system and each module can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0121] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields, is also included in the patent protection scope of the present invention.

Claims

1. A method for monitoring a power transformer, characterized in that: include: Obtaining basic characteristic parameters and installation environment data of multiple power transformers, performing classification and matching based on the basic characteristic parameters and installation environment data, and obtaining sensor layout information; collecting real-time operating data of each of the power transformers according to the sensor arrangement information, performing equipment parallel monitoring, and obtaining equipment parallel monitoring data; Performing insulation degradation trend prediction on each of the power transformers based on the real-time working data and the equipment parallel monitoring data to obtain a single degradation prediction result; Collaborative management of multiple power transformer groups is performed based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions.

2. The power transformer monitoring method according to claim 1, characterized in that: The obtaining of basic characteristic parameters and installation environment data of a plurality of power transformers, and performing classification matching based on the basic characteristic parameters and the installation environment data to obtain sensor arrangement information includes: Obtaining equipment specifications, operating parameters, spatial layout data, equipment housing surface temperature distribution, and electromagnetic characteristic data of multiple power transformers, and constructing basic characteristics to obtain the basic characteristic parameters; Acquiring ambient temperature parameters, ambient humidity parameters, and electromagnetic interference parameters of multiple power transformers, and constructing installation environment characteristics to obtain the installation environment data; Performing fusion adaptation analysis on the basic characteristic parameters and the installation environment data to obtain a device adaptation distribution map; Performing compensation identification for power transformer contact monitoring blind areas on the equipment adaptation distribution map to obtain a three-dimensional coordinate deployment map; The device adaptation distribution map is read based on the three-dimensional coordinate deployment map to obtain the sensor layout information.

3. The power transformer monitoring method according to claim 2, characterized in that: The device adaptation distribution map is subjected to power transformer contact monitoring blind zone compensation construction to obtain a three-dimensional coordinate deployment map, including: Gridding the device adaptation distribution map to obtain monitoring area grid data; Performing a three-dimensional scan of the contact area of the power transformer according to the monitoring area grid data to obtain spatial parameters of the contact area; Identify the monitoring blind area of the spatial parameters of the contact area and perform multi-level decomposition to obtain hierarchical monitoring units; Calculating the sensor coverage radius of the hierarchical monitoring unit to obtain a sensor coverage group; Performing sensor density optimization on the monitoring area grid data according to the sensor coverage group to obtain density distribution data; performing sensor node redundancy optimization on the density distribution data to obtain a node optimization sequence; Sensor placement position constraint analysis is performed according to the node optimization sequence to obtain the three-dimensional coordinate deployment map.

4. The power transformer monitoring method according to claim 1, characterized in that: The collecting the real-time operating data of the power transformer and performing equipment parallel monitoring according to the sensor arrangement information and the real-time operating data to obtain equipment parallel monitoring data include: Partitioning the sensor layout information into monitoring point zones to obtain a monitoring area distribution map; According to the monitoring area distribution map, the sensor layout information is configured with monitoring point parameters, and data collection scheduling is performed to obtain a multi-point collection timing table; Performing cache management on the power transformer according to the multi-point acquisition timing table to obtain the real-time working data; Perform multi-dimensional parameter state identification on the power transformer according to the real-time working data to obtain equipment operating state parameters; The equipment operating status parameters are calculated in parallel, and multiple equipment are evaluated in real time based on a preset distributed computing framework to obtain the equipment parallel monitoring data.

5. The power transformer monitoring method according to claim 1, characterized in that: The performing insulation degradation trend prediction on each power transformer based on the real-time working data and the equipment parallel monitoring data to obtain a single degradation prediction result includes: Performing a single transformer parameter correlation analysis on the real-time working data based on the parallel monitoring data of the equipment to obtain an insulation state influencing factor; Classifying the insulation state influencing factors according to degradation characteristics to obtain an insulation degradation feature combination; Constructing a state transfer matrix based on the insulation degradation feature combination to obtain insulation performance change features; Performing data decomposition calculation on the insulation performance change characteristics to obtain insulation degradation change rules; Predicting insulation performance trends based on the insulation degradation variation law to obtain future insulation degradation development trends; A single evaluation is performed on each of the power transformers according to the future insulation degradation development trend to obtain the single degradation prediction result.

6. The power transformer monitoring method according to claim 5, characterized in that: The method of predicting the insulation performance trend based on the insulation degradation variation law to obtain the future insulation degradation development trend includes: Performing an environmental load temperature correlation analysis on the insulation degradation variation law to obtain the insulation aging temperature dependence characteristics; Calculating the ratio of dissolved gas in the insulating oil of the power transformer according to the insulation aging temperature dependence characteristic to obtain a change rule of characteristic gas components in the insulating oil; Performing dielectric loss tangent analysis on the variation pattern of characteristic gas components of the insulating oil to obtain a degradation rate parameter; Calculating the insulation strength according to the degradation rate parameter to obtain an insulation strength attenuation curve; Extracting partial discharge characteristics from the insulation strength decay curve to obtain an insulation defect development state indicator; Calculating the winding deformation amount according to the insulation defect development state indicator to obtain a mechanical strength degradation warning value; Performing insulation breakdown field strength analysis on the mechanical strength degradation warning value to obtain the distribution of insulation weak points in the target area; A trend forecast is performed based on the distribution of insulation weak points to obtain the future insulation degradation development trend.

7. The power transformer monitoring method according to claim 1, characterized in that: The collaborative management of multiple power transformer groups based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions includes: performing group feature clustering on the plurality of power transformers according to the equipment parallel monitoring data to obtain a transformer group operation feature matrix; performing correlation calculation on the single degradation prediction result according to the transformer group operation characteristic matrix to obtain a group degradation correlation index; Performing hierarchical decomposition on the group degradation correlation index to obtain a group degradation level evaluation result; performing load distribution analysis on the plurality of power transformers according to the group degradation level assessment result to obtain a group load distribution plan; Performing operation configuration matching on the group load distribution scheme to obtain group operation configuration parameters; Performing condition monitoring data fusion on a plurality of the power transformers according to the group operation configuration parameters to obtain a group comprehensive performance index; Maintenance resources are optimally configured according to the group comprehensive performance indicators to obtain the transformer maintenance decision.

8. A power transformer monitoring device, characterized in that: The power transformer monitoring method according to any one of claims 1 to 7 comprises: An acquisition module is configured to acquire basic characteristic parameters and installation environment data of a plurality of power transformers, and perform classification and matching based on the basic characteristic parameters and installation environment data to obtain sensor arrangement information; An analysis module, configured to collect real-time operating data of each of the power transformers according to the sensor arrangement information, perform device parallel monitoring, and obtain device parallel monitoring data; A correlation module, configured to predict the insulation degradation trend of each power transformer based on the real-time operating data and the equipment parallel monitoring data to obtain a single degradation prediction result; A processing module is used to perform collaborative management of multiple power transformer groups based on the equipment parallel monitoring data and the single degradation prediction result to obtain transformer maintenance decisions.

9. A power transformer monitoring device, characterized in that: include: Memory, used to store programs; The processor is configured to execute the program to implement the steps of the power transformer monitoring method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: Computer instructions are stored, and the computer instructions are used to make a computer execute the method according to any one of claims 1 to 7.

Images