Insulator degradation analysis method, system, equipment and medium

Abstract

The invention discloses an insulator degradation analysis method, system and device and a medium, and the method comprises the steps: inputting the real-time inspection data of an insulator into a multi-task deep learning identification model which is obtained through introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into a backbone network of a YOLOv5 model, and carrying out the degradation identification; obtaining degradation state information, and importing the degradation state information into a power grid geographic information comprehensive management system; according to historical inspection comprehensive data and real-time climate data of the insulator, an inspection polymorphic layer of the insulator is constructed through the system, a multi-dimensional algorithm is adopted to analyze the degradation rule of the insulator, and a degradation risk analysis result is obtained; on the basis of a power grid geographic information integrated management system, insulator degradation identification, map positioning and multi-dimensional algorithm fault law analysis are combined, so that the utilization rate of insulator degradation state information is improved, and solid data support and scientific basis are provided for formulating a power grid operation and maintenance strategy.

Description

Technical Field

[0001] This invention relates to the field of insulator degradation analysis technology, and in particular to an insulator degradation analysis method, system, equipment and medium. Background Technology

[0002] Insulators are crucial components ensuring the safe operation of overhead transmission lines. During operation, insulators are subjected to multiple stresses from electromechanical loads and the natural environment, leading to frequent occurrences of potential faults such as decreased insulation resistance, insulation cracking, and even breakdown, directly impacting power reliability. To ensure the reliability of power grid operation, research is needed on insulator degradation condition localization analysis techniques.

[0003] Existing methods for analyzing insulator degradation often focus on single detection steps, such as using drones to collect images or deploying sensors for online monitoring. However, these detection results are mostly isolated data points, and there is insufficient mining of historical detection data, resulting in low information utilization and making it difficult to provide systematic and forward-looking data support for the formulation of power grid operation and maintenance strategies.

[0004] Therefore, how to solve the problem of low data utilization rate in existing insulator degradation analysis schemes, which makes it difficult to provide systematic and forward-looking data support for the formulation of power grid operation and maintenance strategies, has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] This invention provides an insulator degradation analysis method, system, equipment, and medium, which solves the problem that existing insulator degradation analysis schemes have low data utilization and are difficult to provide systematic and forward-looking data support for the formulation of power grid operation and maintenance strategies.

[0006] To solve the above-mentioned technical problems, the first aspect of the present invention provides an insulator degradation analysis method, comprising: Acquire real-time inspection data, real-time climate data, and historical comprehensive inspection data of insulators; The real-time inspection data is input into a multi-task deep learning identification model for degradation identification to obtain degradation state information; the multi-task deep learning identification model is obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of the YOLOv5 model; Based on the degradation status information, a degradation event-location record set of the insulator is constructed and imported into the power grid geographic information integrated management system. Based on the historical inspection integrated data and the real-time climate data, the inspection multi-state layer of the insulator is constructed through the power grid geographic information integrated management system. Based on the inspection multi-state layer, a multi-dimensional algorithm is used to analyze the degradation pattern of the insulator, and the degradation risk analysis results of the insulator are obtained.

[0007] As one preferred embodiment, the insulator feature extraction branch includes a first feature path, a second feature path, and a third feature path; wherein... The step of inputting the real-time inspection data into a multi-task deep learning identification model for degradation identification to obtain degradation state information includes: The real-time inspection data is input to the insulator feature extraction branch, so that the first feature path uses the Sobel operator to extract the edge contour features of the metal parts in the insulator, the second feature path uses the local binary mode operator to extract the texture features of the porcelain or glass disk surface in the insulator, and the third feature path performs threshold segmentation on the infrared image in the real-time inspection data to extract the abnormal heating area features that exceed the set temperature difference. The edge contour features, texture features, and abnormal heat region features are weighted and fused through the cross-modal feature attention fusion branch to obtain enhanced fusion features; The enhanced fusion features are input into the neck network and detection head of the YOLOv5 model for processing to obtain the bounding box coordinates, degradation type label, and its identification confidence and infrared temperature correlation data of the insulator as the degradation state information.

[0008] As one preferred embodiment, the step of constructing a multi-state inspection layer for the insulators based on the historical inspection data and the real-time climate data through the power grid geographic information integrated management system includes: Several dimensions are set, and the real-time inspection data, the real-time climate data, and the historical comprehensive inspection data are packaged into standardized data units according to the dimensions and stored in the spatiotemporal database cube; According to user requirements, target data is extracted from the spatiotemporal database cube so that the power grid geographic information integrated management system renders the target data into several thematic layers, forming the inspection multi-state layer.

[0009] As one preferred embodiment, after analyzing the degradation pattern of the insulator using a multi-dimensional algorithm based on the inspection multi-state layer to obtain the degradation risk analysis result of the insulator, the method further includes: Based on the degradation risk analysis results, a differentiated operation and maintenance strategy is generated; wherein, the differentiated operation and maintenance strategy includes one or more of the following: priority maintenance targets, key monitoring areas, and optimized UAV inspection routes.

[0010] As one preferred embodiment, the degradation pattern of the insulator is analyzed using a multi-dimensional algorithm based on the inspection multi-state layer to obtain the degradation risk analysis results of the insulator, including: The inspection polymorphic layer is invoked to construct a degraded effective dataset and a multidimensional inspection dataset; An improved kernel density estimation algorithm incorporating adaptive bandwidth is used to perform thermal analysis on the degraded effective dataset to obtain a static high-risk thermal map of the insulator. Temporal analysis of the multidimensional inspection dataset is performed using an improved LSTM time series model to obtain the predicted degradation probability data of the insulator in a future preset period; the improved LSTM time series model is a two-layer LSTM structure that incorporates a climate feature attention layer. The static high-risk heat map and the degradation probability prediction data are superimposed and fused to obtain the degradation risk analysis results of the insulator.

[0011] As one preferred embodiment, the step of performing thermal analysis on the degraded effective dataset using an improved kernel density estimation algorithm incorporating adaptive bandwidth to obtain a static high-risk thermal map of the insulator includes: An adaptive bandwidth is determined based on the equipment density and terrain complexity of the area where the insulator is located. A thermal analysis is then performed on the degraded effective dataset by introducing an improved kernel density estimation algorithm based on the adaptive bandwidth, resulting in a static high-risk thermal map of the insulator.

[0012] As one preferred embodiment, the step of performing temporal analysis on the multidimensional inspection dataset using an improved LSTM time series model to obtain the insulator's degradation probability prediction data for a future preset time period includes: The multidimensional inspection dataset is input into the climate feature attention layer to weight the key climate features contained in the multidimensional inspection dataset to obtain a weighted feature vector. The weighted feature vector is input into a two-layer LSTM for processing to obtain the degradation probability distribution of the insulator in the future preset time period; Based on the multidimensional inspection dataset and the degradation probability distribution, an improved Pearson correlation coefficient algorithm is used to quantify the correlation between the climate factors of the insulator and its degradation probability. The quantification results are then combined with the degradation probability distribution to obtain the degradation probability prediction data of the insulator in the future preset period.

[0013] A second aspect of the present invention provides an insulator degradation analysis system, comprising: The data acquisition module is used to acquire real-time inspection data, real-time climate data, and historical comprehensive inspection data of insulators. The degradation identification module is used to input the real-time inspection data into a multi-task deep learning identification model for degradation identification to obtain degradation state information; the multi-task deep learning identification model is obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of the YOLOv5 model; The layer construction module is used to construct a set of deterioration event-location records of the insulator based on the deterioration status information for import into the power grid geographic information integrated management system, and to construct a multi-state inspection layer of the insulator through the power grid geographic information integrated management system based on the historical inspection integrated data and the real-time climate data. The pattern analysis module is used to analyze the degradation pattern of the insulator based on the inspection multi-state layer and employ a multi-dimensional algorithm to obtain the degradation risk analysis results of the insulator.

[0014] A third aspect of the present invention provides an electronic device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the insulator degradation analysis method as described above.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium comprising a stored computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the insulator degradation analysis method as described above.

[0016] Compared with the prior art, the beneficial effects of the embodiments of the present invention are as follows: By introducing a multi-task deep learning identification model, the degradation status of insulators is automatically identified and analyzed, significantly reducing the time cost and subjective errors of traditional manual inspections and improving the speed of operation and maintenance response. The model adds an insulator feature extraction branch and a cross-modal feature attention fusion branch to the YOLOv5 backbone network, effectively integrating multimodal inspection data and significantly improving the accuracy and robustness of degradation identification in complex environments. By constructing a degradation event-location record set and integrating it with the power grid geographic information system, a multi-modal inspection layer covering historical and real-time status is formed, helping operation and maintenance personnel intuitively grasp the distribution, evolution trend, and environmental correlation of insulator degradation, achieving spatiotemporal visualization and control of equipment status. Based on the multi-modal layer, multi-dimensional algorithms are used to analyze the correlation patterns generated by degradation, which can not only identify current high-risk equipment but also predict potential degradation trends. On the basis of the power grid geographic information integrated management system, combined with insulator degradation identification, map positioning, and multi-dimensional algorithm fault pattern analysis, the utilization rate of insulator degradation status information is improved, providing solid data support and scientific basis for the formulation of power grid operation and maintenance strategies. Attached Figure Description

[0017] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of an insulator degradation analysis method provided in a certain embodiment of the present invention; Figure 2 This is a structural diagram of an insulator degradation analysis system provided in a certain embodiment of the present invention; Figure 3 This is a structural diagram of an electronic device provided in a certain embodiment of the present invention; Figure label: Among them, 10 is the data acquisition module; 20 is the degradation identification module; 30 is the layer construction module; 40 is the pattern analysis module; 5000 is the electronic device; 5001 is the processor; 5002 is the bus; 5003 is the memory; and 5004 is the transceiver. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings and examples. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art will be able to understand the specific meaning of the above terms in this application according to the specific circumstances.

[0021] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is merely for describing specific embodiments and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0022] Regarding research on insulator degradation status location analysis technology, patent CN107608375A discloses a degraded insulator detection system and method based on UAVs. This system uses UAVs for environmental inspection, photoelectric sensors to locate and collect the position information of degraded insulators, and electric field sensors to measure the axial electric field strength of the insulator string in real time. Based on this information, an electric field distribution curve is established to determine whether the insulator is degraded, thus improving detection accuracy. The paper "Application of Online Monitoring and Early Warning System for Hidden Dangers in Transmission Lines" proposes the use of hidden danger discharge traveling wave monitoring technology for preventable "gradual" faults such as insulator degradation in transmission lines. The system can monitor the potential discharge of the line in real time and determine whether there is a potential fault by combining historical data; another patent, CN117129815A, discloses a comprehensive detection method and system for multi-deteriorated insulators based on the Internet of Things. It collects working status data and images through built-in sensors and monitoring equipment of the insulators, and uploads them to the local server in real time via wireless communication. It then builds a multi-modal fusion database and a historical working data driving library of insulators, identifies the working status of the insulators in real time, judges the cause of insulator faults and the degree of deterioration, realizes real-time detection and prediction of insulator deterioration trends, and improves detection accuracy.

[0023] However, none of the aforementioned existing technologies consider subsequent macroscopic analysis of insulator degradation status information, resulting in low utilization of insulator degradation status information and a lack of data support and scientific basis for the formulation of power grid operation and maintenance strategies.

[0024] Based on this, in one embodiment, such as Figure 1 As shown, the first aspect of the present invention provides an insulator degradation analysis method, comprising: S1. Acquire real-time inspection data, real-time climate data, and historical comprehensive inspection data of insulators. Specifically, this invention uses drones and their carried cameras and sensors to replace manual high-altitude operations for inspecting transmission lines, performing detailed imaging of key components, and simultaneously acquiring high-definition visible light images and infrared thermal images of the insulator equipment. It also records the GNSS position information, drone attitude, and timestamp corresponding to each frame of the image, forming a raw image dataset with multi-dimensional spatiotemporal labels as real-time inspection data. Real-time temperature, humidity, and wind speed of the area where the insulators are located are also collected as real-time climate data. Historical comprehensive inspection data, including insulator inspection data, inspection time, climate data during inspection, and differentiated maintenance strategies after inspection, are stored in a historical inspection database and categorized by equipment number or equipment identifier for easy retrieval of historical data.

[0025] The acquired data are then preprocessed, including image alignment, noise reduction, and enhancement, to improve data accuracy. Among these, the noise reduction of image data is performed using a bilateral filtering algorithm (which can also be replaced by other existing noise reduction algorithms, such as Gaussian filtering), to eliminate noise in the image (such as environmental interference during shooting, blur noise caused by device shake, etc.) and avoid noise affecting the accuracy of subsequent feature extraction and recognition.

[0026] S2. Input the real-time inspection data into a multi-task deep learning identification model for degradation identification to obtain degradation state information; the multi-task deep learning identification model is obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of the YOLOv5 model; wherein, the insulator feature extraction branch includes a first feature path, a second feature path and a third feature path; In one embodiment, step S2 includes: The real-time inspection data is input to the insulator feature extraction branch, so that the first feature path uses the Sobel operator to extract the edge contour features of the metal parts in the insulator, the second feature path uses the local binary mode operator to extract the texture features of the porcelain or glass disk surface in the insulator, and the third feature path performs threshold segmentation on the infrared image in the real-time inspection data to extract the abnormal heating area features that exceed the set temperature difference. The edge contour features, texture features, and abnormal heat region features are weighted and fused through the cross-modal feature attention fusion branch to obtain enhanced fusion features; The enhanced fusion features are input into the neck network and detection head of the YOLOv5 model for processing to obtain the bounding box coordinates, degradation type label, and its identification confidence and infrared temperature correlation data of the insulator as the degradation state information.

[0027] Specifically, this invention adds an insulator feature extraction branch and a cross-modal feature attention fusion branch to the backbone network CSPDarknet of the YOLOv5 model, while keeping the rest of the structure unchanged. The resulting multi-task deep learning identification model is used to identify real-time inspection data of insulators.

[0028] The preprocessed image data is input into the improved YOLOv5 backbone network. This network acts as a general feature extractor, responsible for extracting general visual features (such as edge, shape, color, and other contextual information) from different levels of the overall image to provide a basis for subsequent localization of the approximate insulator area. Specifically, the image is input into the insulator feature extraction branch, allowing its three feature paths to extract features from the image in parallel: the first feature path specifically targets the visible light image in real-time inspection data, using the Sobel operator to calculate and extract the edge contour features of the insulator's iron cap and its connecting metal components; the second feature path also targets the visible light image, applying the Local Binary Pattern (LBP) operator to extract the texture features of the insulator's ceramic or glass disc surface; the third feature path targets the infrared thermal imaging image in real-time inspection data. Based on historical data, this invention sets a reasonable temperature difference threshold to segment the image, extracting features of abnormally heated areas where the temperature is significantly higher or lower than the surrounding environment. This is because local overheating (hot spots) of the insulator is usually a direct manifestation of internal breakdown, degradation, or increased surface leakage current. It should be noted that if the algorithm or model is improved, this invention will specifically point it out. Unless otherwise specified, the application of LBP and Sobel operators, the subsequent training process of the model, etc. can be referred to the existing technology, and will not be elaborated here.

[0029] The features extracted by the insulator feature extraction branch are then input into the cross-modal feature attention fusion branch. This branch automatically analyzes the "information intensity" and "correlation" of the iron cap edge features, disk texture features, and infrared temperature features extracted from the current image based on an attention mechanism. In this branch, different attention weights are dynamically assigned to the three input features. If the image shows a strong temperature anomaly, the weight of the temperature feature is increased; if the texture feature is abnormally obvious, the weight of the texture feature is correspondingly increased. These weights are then used to perform a weighted sum with their corresponding features to obtain an enhanced fusion feature that contains global information and strengthens the degradation-sensitive features.

[0030] The fused enhanced feature maps are fed into the YOLOv5's neck network and head. The neck network performs multi-scale feature fusion on the enhanced features, enabling the model to identify both large and small / distant insulators. The head predicts the bounding box of the insulator based on the fused features, classifies the insulator's state, and determines its corresponding degradation type. Simultaneously, the model can regress quantitative information related to the insulator, such as the confidence score, specific temperature anomalies extracted from abnormal heating areas, and the highest temperature in hotspot areas, serving as infrared temperature correlation data. Finally, for each insulator in the input image, the improved YOLOv5 model outputs a structured result: degradation state information, including the insulator's bounding box coordinates, degradation type label, identification confidence score, and associated infrared temperature correlation data.

[0031] In this embodiment, the weight values ​​of edge contour features, texture features, and abnormal heating area features are calculated proportionally from geometric feature similarity (represented by cosine similarity) and temperature anomaly (which is the normalized degree of deviation of the temperature of the abnormal heating area from a preset reference temperature threshold); wherein, the weight value of edge contour features... This can be expressed by the following formula: In the formula, The proportionality coefficient is preferably 0.6; The geometric similarity between the edge contour features and the standard insulator edge contour vector is the cosine similarity between the two. This represents the geometric similarity between the texture features and the standard insulator texture vector, which is also known as the cosine similarity between the two.

[0032] Weight values ​​of texture features This can be expressed by the following formula: Weight values ​​of abnormal heating area characteristics This can be expressed by the following formula: In the formula, Temperature anomaly is a characteristic of abnormally heated areas.

[0033] Insulator feature attention mechanism optimizes loss function as follows: In the formula, For classification loss; For position loss; Loss due to temperature deviation; The weighting parameter for the location loss is 0.8 in this embodiment. The weighting parameter for temperature deviation loss (1.2 in this embodiment) enables the identification of insulator degradation types such as cracks, breakdowns, and overheating.

[0034] The multi-task deep learning identification model is trained using labeled samples containing a certain number of normal and deteriorated insulators (samples with different environments and different degrees of deterioration are required). By setting the number of iterations and the learning rate decay strategy, the mean accuracy (mAP) of the model is significantly improved compared with the traditional YOLOv5.

[0035] The model employed in this invention establishes three independent feature pathways to selectively extract core degradation features from different components of the insulator, achieving multi-dimensional professional feature extraction, accurately matching degradation physical mechanisms, and enhancing the model's perception and anti-interference capabilities in complex scenarios. Through dynamic learning via an attention mechanism, different feature pathways are assigned different weights to fuse the extracted features, improving the accuracy and reliability of comprehensive decision-making. This not only significantly improves the performance of degradation identification but also outputs richer and more action-oriented information. Furthermore, it effectively integrates visible light morphological texture information and infrared temperature information, significantly enhancing the accuracy and robustness of insulator degradation identification in complex backgrounds.

[0036] S3. Construct a degradation event-location record set for the insulator based on the degradation status information and import it into the power grid geographic information integrated management system. Based on the historical inspection data and the real-time climate data, construct a multi-state inspection layer for the insulator through the power grid geographic information integrated management system. Specifically, this invention binds the degradation status information output by the model to its corresponding GNSS coordinates or precise latitude and longitude obtained through georeferencing technology to generate a structured "degradation event-location" record set. This invention imports this record set into the power grid geographic information integrated management system in batches through a data interface or manual input. The insulator equipment is mapped to the equipment location in the power grid geographic information integrated management system according to latitude and longitude and optical character recognition equipment names. This allows the system to automatically render and label each record on the digital map with a highlighted, distinguishable icon (e.g., using different colors and shapes according to the degradation type) at the corresponding coordinate location, achieving a panoramic visualization and positioning of the insulator degradation status on a single map.

[0037] The system also uses icon click event binding technology to associate highlighted icons with the unique data identifiers of the devices (such as device ID and latitude / longitude combinations). When maintenance personnel click the icon of an insulator with deterioration in the power grid geographic information integrated management system, the system is triggered, allowing them to query all fault-related data corresponding to that icon. The system displays corresponding deterioration status information, location information, etc., presented in pop-up or sidebar format (compliant with the interaction design specifications of the power grid geographic information integrated management system). The information is modularly arranged, including basic information, deterioration details, and environmental associations, along with image previews (allowing users to view original deterioration images collected by drones), ensuring intuitiveness and ease of understanding. In addition to information display, the system also provides editing permissions, meeting the data update needs of maintenance personnel throughout the entire process from fault discovery to resolution, providing the maintenance team with an intuitive and clear information display platform.

[0038] In one embodiment, constructing the multi-state inspection layer of the insulator based on the historical inspection data and the real-time climate data through the power grid geographic information integrated management system includes: Several dimensions are set, and the real-time inspection data, the real-time climate data, and the historical comprehensive inspection data are packaged into standardized data units according to the dimensions and stored in the spatiotemporal database cube; According to user requirements, target data is extracted from the spatiotemporal database cube so that the power grid geographic information integrated management system renders the target data into several thematic layers, forming the inspection multi-state layer.

[0039] Specifically, this invention sets multiple dimensions, including equipment (tower number, insulator string number or its corresponding identifier), location (spatial coordinates of the insulator), time (time period corresponding to the inspection), status (deterioration type, etc.), environment (temperature, humidity, wind speed, etc. during inspection), and business (inspection method, differentiated operation and maintenance strategy, etc.). Real-time inspection data, real-time climate data, and historical comprehensive inspection data are classified and standardized according to these dimensions to encapsulate them into standardized data units. These units are then stored in a spatiotemporal database cube using PostgreSQL 15 + PostGIS 3.3. Leveraging its spatial index optimization and multi-dimensional data query capabilities, a three-layer cube structure is constructed. The bottom layer is the raw data layer, storing all standardized data units (including historical and real-time data). The middle layer is the aggregated data layer, pre-aggregating data at time granularity (day / week / month) + spatial granularity (line segment / county) + business granularity (defect level / climate type). The top layer is the index layer, establishing spatial R-tree indexes, time B-tree indexes, and business-dimensional hash indexes, supporting millisecond-level data retrieval. The standardization process is as follows: For numerical data, two decimal places are retained; for text data, enumerated values ​​are used for defect types, weather types, etc. (e.g., defect types: 01=pollution flashover, 02=lightning strike, 03=mechanical damage, 04=aging cracking); for spatial data, the coordinate system is uniformly adopted as WGS84, and the tower number and insulator string number follow the "Transmission Equipment Coding Standard"; for image data, the format is uniformly JPG (resolution 1920×1080), and the device ID and shooting timestamp are associated.

[0040] User requirements can be entered by dimension through the visual query interface of the power grid geographic information integrated management system. The system can then quickly locate the matching aggregated data layer through the index layer based on the query conditions. If there are no matching results in the aggregated data layer, the system will automatically go down to the original data layer and filter the target data through multi-dimensional joint query (spatial R-tree index + time B-tree index + business hash index). After the filtering is completed, a temporary aggregated result is generated so that the power grid geographic information integrated management system (developed based on ArcGIS Server) can render the extracted target data into several related thematic layers and overlay them to form a multi-modal inspection layer.

[0041] The inspection multi-state layer includes a business-themed layer. This layer is rendered with the insulator's tower coordinates as the center point, setting thermal icons according to degradation type (e.g., yellow for flashover, orange for mechanical damage, red for aging cracking, etc.) or according to defect level (determined by the magnitude of the impact of each degradation type, or historical data) and thermal intensity (e.g., Level II defect: yellow, thermal radius 50 meters; Level III defect: orange, thermal radius 80 meters; Level IV defect: red, thermal radius 120 meters). Hovering the mouse over the defect displays the insulator string number, defect type, inspection time, and insulation resistance value; clicking allows viewing defect photos and historical inspection records. The climate-themed layer uses different icons to indicate climate type (e.g., heavy rain: blue raindrop icon, indicating daily precipitation; heavy fog: gray fog icon, indicating relative humidity). Icon size is positively correlated with the impact range (e.g., a heavy rain icon with 100mm of daily precipitation is larger than 50). The system can automatically link defect data within the same time and space range. Clicking the climate icon allows you to view the distribution of all defects under that weather condition. There is also a dedicated layer for equipment health status, which is rendered based on a comprehensive assessment of health status by defect level and installation age (green: no defects + installation age < 8 years; yellow: Level I-II defects + installation age 8-12 years; red: Level III-IV defects + installation age ≥ 12 years). Batch rendering is done by tower (when the health status of all insulator strings on the same tower is consistent, it is marked at the top of the tower; when there are differences, it is marked at the corresponding insulator string position). It also supports quick filtering by health status, hiding green healthy equipment and only displaying yellow / red high-risk equipment. Finally, there is a dedicated layer for historical defect comparison, which uses dashed lines to indicate the defect location in the same period of a certain year, and solid lines to indicate the current defect location in the following year. Defects from different years are distinguished by different colors. The system automatically calculates the defect growth rate for the same area and displays statistical charts in the corner of the layer.

[0042] In addition, the multiple thematic layers in the inspection multi-state layer can also be designed in layers according to dimensions, with different dimensions corresponding to independent sub-layers, and each layer can be linked and communicated. I will not go into details here.

[0043] This invention transforms discrete, multi-batch inspection data into a dynamic visualization layer that supports flexible switching and comparative analysis across multiple dimensions, including time, status, and environment, by standardizing the data and utilizing GIS layer management capabilities. This not only greatly improves the efficiency of data management and utilization but also provides a powerful platform for understanding the deterioration patterns of insulators and achieving precise control and intelligent operation and maintenance of power grid assets through spatiotemporal fusion and correlation analysis of multi-dimensional data.

[0044] S4. Based on the inspection multi-state layer, a multi-dimensional algorithm is used to analyze the degradation pattern of the insulator to obtain the degradation risk analysis results of the insulator; In one embodiment, step S4 includes: The inspection polymorphic layer is invoked to construct a degraded effective dataset and a multidimensional inspection dataset; An improved kernel density estimation algorithm incorporating adaptive bandwidth is used to perform thermal analysis on the degraded effective dataset to obtain a static high-risk thermal map of the insulator. Temporal analysis of the multidimensional inspection dataset is performed using an improved LSTM time series model to obtain the predicted degradation probability data of the insulator in a future preset period; the improved LSTM time series model is a two-layer LSTM structure that incorporates a climate feature attention layer. The static high-risk heat map and the degradation probability prediction data are superimposed and fused to obtain the degradation risk analysis results of the insulator.

[0045] Specifically, this invention calls upon the inspection polymorphic layer to filter out valid records containing fault point coordinates and degradation type labels from historical data. After removing invalid data (such as records with missing coordinates or blurred labels), it constructs a valid degradation dataset. Based on the inspection polymorphic layer, it integrates three core data categories: "inspection time, climate data, and fault status" to form a multidimensional inspection dataset—arranged in chronological order (e.g., by day / week / month), with each time node corresponding to "climate conditions, fault occurrence frequency, and degradation status information for that period."

[0046] Subsequently, an improved kernel density estimation algorithm (KDE) was used to analyze the deterioration effective dataset from a spatial distribution perspective to obtain a static high-risk heat map of the insulator. An improved LSTM time series model was then used to process the multidimensional inspection dataset from a temporal evolution perspective to analyze the temporal pattern of fault occurrence, predict the fault probability in a specific future period, and obtain the deterioration probability prediction data of the insulator in a future preset period.

[0047] Finally, by overlaying and fusing the obtained static high-risk heat map and degradation probability prediction data, the degradation risk analysis results of the insulators can be generated on the GIS map. This invention pioneers a spatiotemporally coupled deep analysis paradigm: it organically combines spatial density analysis (revealing geographical risk patterns) with temporal prediction analysis (revealing evolutionary risk trends), overcoming the one-sidedness of single-dimensional analysis.

[0048] In one embodiment, the step of using an improved kernel density estimation algorithm incorporating adaptive bandwidth to perform thermal analysis on the degraded effective dataset to obtain a static high-risk thermal map of the insulator includes: An adaptive bandwidth is determined based on the equipment density and terrain complexity of the area where the insulator is located. A thermal analysis is then performed on the degraded effective dataset by introducing an improved kernel density estimation algorithm based on the adaptive bandwidth, resulting in a static high-risk thermal map of the insulator.

[0049] Specifically, this invention employs an improved kernel density estimation algorithm, replacing the traditional fixed bandwidth with an adaptive bandwidth that dynamically adjusts based on regional device density (dense urban areas / sparse mountainous areas) and terrain complexity (plains / hills / mountains). This avoids the problems of "overly smoothed dense areas and distorted identification in sparse areas" caused by fixed bandwidth. Density calculation is then performed and accelerated based on the adaptive bandwidth. This process is expressed by the following formula: In the formula, n represents the number of fault points, which can also be considered as the number of deteriorated insulators; , Let be the coordinates of the i-th fault point; K be the Gaussian kernel function; h be the adaptive bandwidth, which is set according to the dynamic adjustment logic. The dynamic adjustment logic satisfies the following: the higher the density of equipment in the area, the smaller the adaptive bandwidth; the higher the terrain complexity, the larger the adaptive bandwidth. Its calculation process is expressed by the following formula: In the formula, C is the base bandwidth; C is the terrain complexity factor (used to quantify terrain complexity; in this embodiment, the terrain complexity factor for plains is assigned a value of 1, for hills it is assigned a value of 2, and for mountains it is assigned a value of 3); D is the regional device density factor (used to quantify regional device density; in this embodiment, when the regional device density is less than 1 device / km², the regional device density factor is assigned a value of 1; when the regional device density is greater than or equal to 1 device / km² and less than 10 devices / km², the regional device density factor is assigned a value of 2; when the regional device density is greater than or equal to 10 devices / km², the regional device density factor is assigned a value of 3). This is the sensitivity adjustment coefficient, used to adjust the sensitivity of the impact of terrain complexity and equipment density on bandwidth (in this embodiment, the sensitivity adjustment coefficient is 1).

[0050] By accelerating kernel density matrix operations through GPU parallel computing architecture, the problem of low computational efficiency for massive historical data is solved, and a static high-risk heat map is generated. The darker the color in the heat map, the higher the density of fault points and the greater the risk of degradation. The heat map is overlaid on the power grid geographic information integrated management system to identify high-risk areas. After algorithm improvement, the high-risk area identification accuracy is improved by more than 30% compared with the traditional fixed bandwidth KDE, thereby helping operation and maintenance personnel to quickly locate fault-concentrated areas and providing guidance for targeted operation and maintenance measures.

[0051] This invention dynamically determines adaptive bandwidth based on equipment density and terrain complexity, enabling the algorithm to intelligently adapt to the actual conditions of different areas, significantly improving the spatial accuracy and authenticity of high-risk area identification, and successfully overcoming the inherent limitations of traditional methods in spatial risk analysis of power equipment.

[0052] In one embodiment, the step of performing temporal analysis on the multidimensional inspection dataset using an improved LSTM time series model to obtain the insulator's degradation probability prediction data over a future preset period includes: The multidimensional inspection dataset is input into the climate feature attention layer to weight the key climate features contained in the multidimensional inspection dataset to obtain a weighted feature vector. The weighted feature vector is input into a two-layer LSTM for processing to obtain the degradation probability distribution of the insulator in the future preset time period; Based on the multidimensional inspection dataset and the degradation probability distribution, an improved Pearson correlation coefficient algorithm is used to quantify the correlation between the climate factors of the insulator and its degradation probability. The quantification results are then combined with the degradation probability distribution to obtain the degradation probability prediction data of the insulator in the future preset period.

[0053] Specifically, this invention employs an improved LSTM time series model to perform temporal analysis on a multidimensional inspection dataset. This model includes a climate feature attention layer (adapting to the number of data features), a two-layer LSTM (dynamically adjusting the number of nodes), and an output layer (classifying the corresponding fault severity). Furthermore, it uses a sliding window (with a window size that adapts to the historical data period) to predict the probability of degradation in a future preset time period. In other words, for each day's multidimensional inspection dataset within the sliding window, it is input into the climate feature attention layer. A learnable attention weight matrix evaluates the importance of each climate feature (such as rainfall, humidity, and wind speed) to the task of "predicting the next stage of degradation" based on all current input information (including historical states). The importance is normalized to form weights, which are then used to weight the corresponding climate features to obtain a weighted feature vector. The weight allocation method of the aforementioned climate feature attention layer is a well-known and conventional technique in the field of deep learning, which can be implemented by those skilled in the art based on common knowledge, and will not be elaborated upon here.

[0054] The weighted feature vector of the entire sliding window is input into the first LSTM layer in chronological order to capture short-period, fine-grained local dependency patterns in the input sequence, such as how several consecutive days of humid weather gradually affect surface resistance. This yields a hidden state sequence, which is then processed by the second LSTM layer to capture long-period, macroscopic evolution trends and deep-seated patterns, such as the long-term correlation between seasonal climate change (e.g., rainy season, high-temperature season) and insulator performance degradation. The hidden state of the second LSTM layer at the last time step, which encapsulates the evolutionary information of the entire historical window, is then passed through a fully connected layer in the output layer and activated using either the Sigmoid or Softmax activation function. This yields a probability value between 0 and 1 (when using Sigmoid, it represents the probability of degradation in the future period) or a probability distribution (when using Softmax, it represents the probability of the insulator being in "normal", "mild", or "severe" states in the future), i.e., the probability distribution of insulator degradation in the future preset period.

[0055] Simultaneously, based on the multidimensional inspection dataset and degradation probability distribution, an improved Pearson correlation coefficient algorithm is used to calculate the correlation between the failure frequency (failure can also be understood as degradation) and climatic factors in different regions. The formula is as follows: In the formula, r is the improved Pearson correlation coefficient, which ranges from [-1, 1]. The closer r is to 1 or -1, the stronger the linear correlation. The closer r is to 0, the weaker the linear correlation. T is a single observation of a climate factor, such as specific climate index data such as rainfall, temperature, and wind force level in a certain region on a certain day. T is the average of all observed values ​​of climate factors, that is, the arithmetic mean of all T values ​​of a certain climate index over a period of time; F is a single observed value of failure frequency, such as the number of equipment failures or failure rate in the same area on the same day corresponding to T. This is the average of all observed values ​​for the fault frequency.

[0056] By combining the probability distribution of insulator degradation over a specific future period as direct information with key influencing climate factors and their correlation coefficients as auxiliary information, the resulting data predicts the insulator degradation probability over a predetermined period, providing clear guidance for operation and maintenance strategies. The improved LSTM time series model demonstrates 25% greater stability in correlation calculations compared to traditional algorithms in small-sample scenarios, thus providing a precise basis for differentiated operation and maintenance. Based on the above technical solutions, the power grid geographic information integrated management system displays the entire voltage level power grid structure and geographical orientation on a single map. It allows users to view transformer ledger information and low-voltage lines in each transformer area, and statistically display the scale and distribution of various users within that area. It possesses the capability for hierarchical and tiered management of power grid equipment across all voltage levels, and comprehensive coverage of power sources, grids, loads, and energy storage.

[0057] This invention introduces a climate feature attention layer to achieve intelligent weighting of input features. This mechanism can automatically learn and dynamically highlight the climate features most relevant to the degradation process, while downplaying the influence of irrelevant or redundant information, significantly improving prediction accuracy and physical rationality. An improved Pearson correlation coefficient algorithm is used to quantify the dynamic correlation between key climate factors and degradation probability, enhancing the interpretability of the prediction model and the credibility of decision-making.

[0058] In one embodiment, after step S4, the method further includes: Based on the degradation risk analysis results, a differentiated operation and maintenance strategy is generated; wherein, the differentiated operation and maintenance strategy includes one or more of the following: priority maintenance targets, key monitoring areas, and optimized UAV inspection routes.

[0059] Specifically, based on the degradation risk analysis results and the multi-state inspection layers invoked, differentiated operation and maintenance strategies are automatically recommended or assisted in being generated. For example, for equipment located in high-risk areas with a high predicted probability, it is recommended to immediately arrange special inspections or maintenance; for equipment with an increased predicted probability but located outside the current risk area, it is suggested to increase the monitoring frequency; and based on the risk distribution, optimized routes focusing on inspections in the next cycle are planned for the drones, thereby directly transforming the analysis results into executable operation and maintenance action plans to achieve data-driven and precise decision-making.

[0060] This invention transforms complex analysis results into intuitive and operable instructions, reducing intermediate steps and time delays between expert interpretation of analysis reports and the formulation of work plans. It achieves near real-time automated decision-making from risk warning to strategy generation, effectively supporting the digital transformation and intelligent operation and maintenance of the power grid, and greatly improving the efficiency and scientific nature of power grid operation and maintenance.

[0061] This application addresses the problem that existing insulator degradation analysis schemes suffer from low data utilization, making it difficult to provide systematic and forward-looking data support for power grid operation and maintenance strategy formulation. To address this, an insulator degradation analysis method is designed. This method introduces a multi-task deep learning identification model to automatically identify and analyze the degradation state of insulators, significantly reducing the time cost and subjective errors of traditional manual inspections and improving operation and maintenance response speed. The model adds an insulator feature extraction branch and a cross-modal feature attention fusion branch to the YOLOv5 backbone network, effectively fusing multi-modal inspection data and significantly improving the accuracy and robustness of degradation identification in complex environments. By constructing a degradation event... The system integrates insulator location records with the power grid geographic information system to form a multi-state inspection layer covering historical and real-time status. This helps maintenance personnel intuitively grasp the distribution, evolution trend, and environmental correlation of insulator degradation, enabling spatiotemporal visualization and control of equipment status. Based on the multi-state layer, multi-dimensional algorithms are used to analyze the correlation patterns of degradation, which can not only identify current high-risk equipment but also predict potential degradation trends. On the basis of the power grid geographic information integrated management system, combined with insulator degradation identification, map positioning, and multi-dimensional algorithm fault pattern analysis, the utilization rate of insulator degradation status information is improved, providing solid data support and scientific basis for the formulation of power grid operation and maintenance strategies.

[0062] It should be noted that although the steps in the flowchart above are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order requirement for the execution of these steps, and they can be executed in other orders.

[0063] In another embodiment, such as Figure 2 As shown, a second aspect of the present invention provides an insulator degradation analysis system, comprising: Data acquisition module 10 is used to acquire real-time inspection data, real-time climate data and historical comprehensive inspection data of insulators; The degradation identification module 20 is used to input the real-time inspection data into a multi-task deep learning identification model for degradation identification to obtain degradation state information; the multi-task deep learning identification model is obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of the YOLOv5 model; The layer construction module 30 is used to construct a set of deterioration event-location records of the insulator based on the deterioration status information for import into the power grid geographic information integrated management system, and to construct a multi-state inspection layer of the insulator through the power grid geographic information integrated management system based on the historical inspection integrated data and the real-time climate data. The pattern analysis module 40 is used to analyze the degradation pattern of the insulator based on the inspection multi-state layer and using a multi-dimensional algorithm to obtain the degradation risk analysis result of the insulator.

[0064] It should be noted that each module in the aforementioned insulator degradation analysis system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module. For specific limitations regarding an insulator degradation analysis system, please refer to the limitations regarding an insulator degradation analysis method above; both have the same function and role, and will not be repeated here.

[0065] A third aspect of the present invention provides an electronic device comprising: Processor, memory, and bus; The bus is used to connect the processor and the memory; The memory is used to store operation instructions; The processor is configured to execute instructions by calling the operation instructions, causing the processor to perform operations corresponding to an insulator degradation analysis method as shown in the first aspect of this application.

[0066] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The illustrated electronic device 5000 includes a processor 5001 and a memory 5003. The processor 5001 and the memory 5003 are connected, for example, via a bus 5002. Optionally, the electronic device 5000 may also include a transceiver 5004. It should be noted that in practical applications, the transceiver 5004 is not limited to one type, and the structure of this electronic device 5000 does not constitute a limitation on the embodiments of this application.

[0067] Processor 5001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 5001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0068] Bus 5002 may include a path for transmitting information between the aforementioned components. Bus 5002 may be a PCI bus or an EISA bus, etc. Bus 5002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0069] The memory 5003 may be a ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0070] The memory 5003 is used to store application code that executes the scheme of this application, and its execution is controlled by the processor 5001. The processor 5001 is used to execute the application code stored in the memory 5003 to implement the content shown in any of the foregoing method embodiments.

[0071] Among them, electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers.

[0072] The fourth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements an insulator degradation analysis method as shown in the first aspect of the present application.

[0073] Another embodiment of this application provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the aforementioned method embodiments.

[0074] Furthermore, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0075] In summary, this invention relates to the field of insulator degradation analysis technology, and discloses an insulator degradation analysis method, system, equipment, and medium. Real-time insulator inspection data is input into a multi-task deep learning identification model obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of a YOLOv5 model for degradation identification, obtaining degradation state information, which is then imported into a power grid geographic information integrated management system. Based on historical inspection data and real-time climate data of the insulators, a multi-state inspection layer of the insulators is constructed through this system to analyze the degradation patterns of the insulators using multi-dimensional algorithms, obtaining degradation risk analysis results. Based on the power grid geographic information integrated management system, and combining insulator degradation identification, map positioning, and multi-dimensional algorithm fault pattern analysis, the utilization rate of insulator degradation state information is improved, providing solid data support and scientific basis for the formulation of power grid operation and maintenance strategies.

[0076] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0077] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for analyzing insulator degradation, characterized in that, include: Acquire real-time inspection data, real-time climate data, and historical comprehensive inspection data of insulators; The real-time inspection data is input into a multi-task deep learning identification model for degradation identification to obtain degradation state information; the multi-task deep learning identification model is obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of the YOLOv5 model; Based on the degradation status information, a degradation event-location record set of the insulator is constructed and imported into the power grid geographic information integrated management system. Based on the historical inspection integrated data and the real-time climate data, the inspection multi-state layer of the insulator is constructed through the power grid geographic information integrated management system. Based on the inspection multi-state layer, a multi-dimensional algorithm is used to analyze the degradation pattern of the insulator, and the degradation risk analysis results of the insulator are obtained.

2. The insulator degradation analysis method according to claim 1, characterized in that, The insulator feature extraction branch includes a first feature path, a second feature path, and a third feature path; wherein... The step of inputting the real-time inspection data into a multi-task deep learning identification model for degradation identification to obtain degradation state information includes: The real-time inspection data is input to the insulator feature extraction branch, so that the first feature path uses the Sobel operator to extract the edge contour features of the metal parts in the insulator, the second feature path uses the local binary mode operator to extract the texture features of the porcelain or glass disk surface in the insulator, and the third feature path performs threshold segmentation on the infrared image in the real-time inspection data to extract the abnormal heating area features that exceed the set temperature difference. The edge contour features, texture features, and abnormal heat region features are weighted and fused through the cross-modal feature attention fusion branch to obtain enhanced fusion features; The enhanced fusion features are input into the neck network and detection head of the YOLOv5 model for processing to obtain the bounding box coordinates, degradation type label, and its identification confidence and infrared temperature correlation data of the insulator as the degradation state information.

3. The insulator degradation analysis method according to claim 1, characterized in that, The step of constructing a multi-state inspection layer for the insulators based on the historical inspection data and the real-time climate data through the power grid geographic information integrated management system includes: Several dimensions are set, and the real-time inspection data, the real-time climate data, and the historical comprehensive inspection data are encapsulated into standardized data units according to the dimensions and stored in the spatiotemporal database cube; According to user requirements, target data is extracted from the spatiotemporal database cube so that the power grid geographic information integrated management system renders the target data into several thematic layers, forming the inspection multi-state layer.

4. The insulator degradation analysis method according to claim 1, characterized in that, Based on the inspection multi-state layer, a multi-dimensional algorithm is used to analyze the degradation pattern of the insulator, and the degradation risk analysis results of the insulator are obtained, including: The inspection polymorphic layer is invoked to construct a degraded effective dataset and a multidimensional inspection dataset; An improved kernel density estimation algorithm incorporating adaptive bandwidth is used to perform thermal analysis on the degraded effective dataset to obtain a static high-risk thermal map of the insulator. Temporal analysis of the multidimensional inspection dataset is performed using an improved LSTM time series model to obtain the predicted degradation probability data of the insulator in a future preset period; the improved LSTM time series model is a two-layer LSTM structure that incorporates a climate feature attention layer. The static high-risk heat map and the degradation probability prediction data are superimposed and fused to obtain the degradation risk analysis results of the insulator.

5. The insulator degradation analysis method according to claim 4, characterized in that, The improved kernel density estimation algorithm, incorporating adaptive bandwidth, is used to perform thermal analysis on the degraded effective dataset to obtain a static high-risk thermal map of the insulator, including: An adaptive bandwidth is determined based on the equipment density and terrain complexity of the area where the insulator is located. A thermal analysis is then performed on the degraded effective dataset by introducing an improved kernel density estimation algorithm based on the adaptive bandwidth, resulting in a static high-risk thermal map of the insulator.

6. The insulator degradation analysis method according to claim 4, characterized in that, The improved LSTM time series model is used to perform temporal analysis on the multidimensional inspection dataset to obtain the insulator's degradation probability prediction data for a future preset time period, including: The multidimensional inspection dataset is input into the climate feature attention layer to weight the key climate features contained in the multidimensional inspection dataset to obtain a weighted feature vector. The weighted feature vector is input into a two-layer LSTM for processing to obtain the degradation probability distribution of the insulator in the future preset time period; Based on the multidimensional inspection dataset and the degradation probability distribution, an improved Pearson correlation coefficient algorithm is used to quantify the correlation between the climatic factors of the insulator and its degradation probability. The quantification results are then combined with the degradation probability distribution to obtain the degradation probability prediction data of the insulator in the future preset period.

7. The insulator degradation analysis method according to claim 1, characterized in that, After analyzing the degradation pattern of the insulator using a multi-dimensional algorithm based on the inspection multi-state layer to obtain the degradation risk analysis result of the insulator, the process further includes: Based on the degradation risk analysis results, a differentiated operation and maintenance strategy is generated; wherein, the differentiated operation and maintenance strategy includes one or more of the following: priority maintenance objects, key monitoring areas, and optimized UAV inspection routes.

8. An insulator degradation analysis system, characterized in that, include: The data acquisition module is used to acquire real-time inspection data, real-time climate data, and historical comprehensive inspection data of insulators. The degradation identification module is used to input the real-time inspection data into a multi-task deep learning identification model for degradation identification to obtain degradation state information; the multi-task deep learning identification model is obtained by introducing an insulator feature extraction branch and a cross-modal feature attention fusion branch into the backbone network of the YOLOv5 model; The layer construction module is used to construct a set of deterioration event-location records of the insulator based on the deterioration status information for import into the power grid geographic information integrated management system, and to construct a multi-state inspection layer of the insulator through the power grid geographic information integrated management system based on the historical inspection integrated data and the real-time climate data. The pattern analysis module is used to analyze the degradation pattern of the insulator based on the inspection multi-state layer and employ a multi-dimensional algorithm to obtain the degradation risk analysis results of the insulator.

9. An electronic device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the insulator degradation analysis method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein when the device containing the computer-readable storage medium executes the computer program, it implements the insulator degradation analysis method as described in any one of claims 1 to 7.

Images