Electrical insulation equipment salt deposit density analysis method based on online conductivity algorithm

Abstract

The invention discloses an electrical insulation equipment salt deposit density analysis method based on an on-line conductivity algorithm, and relates to the technical field of intelligent operation and maintenance and insulation state monitoring of electrical equipment, and the method comprises the steps: collecting conductivity and environmental parameters; constructing a hybrid parameter optimization model, performing iterative optimization on the hyper-parameter combination of the OPTICS clustering algorithm by using a particle swarm-multi-target grey wolf algorithm, outputting an optimal hyper-parameter combination, configuring the optimal hyper-parameter combination to an OPTICS clustering algorithm kernel, and generating an optimal salt deposit density clustering model; importing the conductivity and the environmental parameters into an optimal salt density clustering model, determining a core distance and a reachable distance to analyze a connected cluster structure, converting an unsupervised cluster into a supervised salt density reference value through a cluster salt density mapping mechanism, and outputting a predicted salt density value; based on the predicted salt deposit density value, in combination with a multi-level risk threshold, triggering an abnormal alarm; according to the invention, the accuracy and timeliness of insulator pollution state evaluation are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance and insulation condition monitoring technology for power equipment, specifically a salt density analysis method for electrical insulation equipment based on an online conductivity algorithm. Background Technology

[0002] The equivalent salt density value on the surface of insulators is a quantitative data for judging the severity of pollution of the external insulation of electrical porcelain. It is one of the important bases for classifying pollution levels and drawing pollution maps. Salt density measurement is of great significance to the safe operation of power systems. In traditional insulator salt density testing, traditional testing methods have the following limitations: On the one hand, destructive testing methods are often used, requiring the removal of insulators for manual cleaning and laboratory measurements. This not only leads to equipment downtime and affects power supply reliability, but also brings high risks due to high-altitude operations in transmission line insulators and other scenarios, causing maintenance costs to surge. At the same time, traditional testing methods have obvious lag and inefficiency. The testing cycle is usually quarterly or annual, making it difficult to capture sudden pollution situations such as sandstorms and industrial pollution in a timely manner. Furthermore, point sampling cannot reflect the differences in the distribution of pollution in equipment in different areas.

[0003] On the other hand, the single salt density value obtained by traditional monitoring methods is not associated with environmental parameters, making it impossible to establish a dynamic model of pollution accumulation. Moreover, relying on empirical thresholds for early warning leads to a high false alarm rate. This single, experience-based detection mode is difficult to adapt to the dynamic and precise requirements of power systems for equipment status perception, often resulting in problems such as untimely pollution early warning and insufficient targeted operation and maintenance decisions, posing potential risks to the safe operation of equipment and even increasing the probability of power grid failure.

[0004] Therefore, it is necessary to develop a non-destructive, real-time, efficient salt density detection technology that can correlate with environmental parameters. Summary of the Invention

[0005] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a salt density analysis method for electrical insulation equipment based on an online conductivity algorithm. The method involves immersing the insulator in a dissolving cup controlled by a lifting platform, collecting environmental parameters and conductivity data, obtaining optimal parameters through PSO-MOGWO hybrid optimization, constructing an OPTICS clustering model, and inputting the environmental parameters and conductivity into the model to complete the classification of salt density level clusters, salt density value prediction, and risk warning. The model output accurately identifies the salt density status of the insulator, effectively controlling operational risks of external power grid insulation equipment, ensuring equipment reliability, improving the accuracy and timeliness of insulator pollution status assessment, and solving the problems mentioned in the background technology.

[0006] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: This application provides a salt density analysis method for electrical insulation equipment based on an online conductivity algorithm, the method comprising: When the insulator enters the testing cycle, the dissolving cup and the environmental sensor array are controlled by the lifting platform to simultaneously collect conductivity and environmental parameters. A hybrid parameter optimization model is constructed, and the hyperparameter combination of the OPTICS clustering algorithm is iteratively optimized using the particle swarm optimization-multi-objective gray wolf algorithm. The optimal hyperparameter combination is output and configured into the kernel of the OPTICS clustering algorithm to generate the optimal salt density clustering model. Electrical conductivity and environmental parameters are imported into the optimal salt density clustering model to determine the core distance and reachability distance in order to resolve the connected cluster structure. Through the cluster salt density mapping mechanism, the unsupervised clusters are transformed into supervised salt density benchmark values. Combined with the reachability distance, the salt density cluster level to which the new sample belongs is matched, and the predicted salt density value is output. Based on the predicted salt density value, combined with a multi-level threshold discrimination mechanism, anomaly alarms are triggered or the processing strategy is dynamically adjusted.

[0007] Furthermore, conductivity and environmental parameters were collected, including: The lifting platform synchronously triggers and controls the dissolving cup and the environmental sensor array; Under the triggering of the lifting platform, the dissolving cup contacts the surface of the insulator through a sequence of actions including rising, immersing, emulsifying, and falling to obtain the conductivity of the solution. An environmental sensor array measures environmental parameters in real time and aligns them with conductivity data using timestamps. The conductivity and environmental parameters are preprocessed.

[0008] Furthermore, the hyperparameter combination includes the maximum neighborhood radius and the reachable distance threshold.

[0009] Furthermore, the optimal combination of hyperparameters is output, including: Determine the hyperparameter combination of the OPTICS clustering algorithm to construct a two-dimensional parameter search space; The particle swarm optimization algorithm is used to perform a global search. An initial population is randomly generated in the two-dimensional parameter search space. The population consists of m particles, and each particle has a unique position vector and velocity vector. At the same time, the OPTICS clustering algorithm is used to determine the clustering results of each particle to extract the silhouette coefficient and calculate the fitness. A multi-objective gray wolf algorithm is introduced for local mining: the population is sorted in descending order of fitness, the top N0 particles are selected and converted into the initial leader wolf position vector, and dynamic calibration is performed. If the maximum number of iterations is reached, the population is guided to move by adjusting the wolf position vector; otherwise, the optimal hyperparameter combination is output; where N0 is a positive integer greater than 0.

[0010] Further, determine the core distance and reachability distance, including: Determine the core distance: Call the maximum neighborhood radius to traverse the sample space and count the number of neighborhood samples of any sample point in the sample space within the maximum neighborhood radius. The number of neighborhood samples is compared with a preset minimum number of points threshold: if the number of neighborhood samples is greater than or equal to the preset minimum number of points threshold, the sample point is determined to be a core object; otherwise, the sample point is determined to be a noise point. Based on the core object, all sample points within the maximum neighborhood radius are sorted sequentially, and the minimum number of points threshold is marked as N1. The Euclidean distance between the N1th sample point and the nearest sample point is taken as the core distance; where N1 is a positive integer greater than 0. Determine reachable distance: Based on core distance: Select the core object and sample points in its neighborhood, and calculate the Euclidean distance between them; compare the core distance of the core object with the Euclidean distance, and take the larger value as the reachable distance of the sample point relative to the core object point.

[0011] Furthermore, the connected cluster structure is resolved, including: Construct a reachability graph based on reachability distance; The optimal reachability threshold of the hybrid parameter optimization model is called to cut the valleys in the reachability distance map, identify the valley regions, and mark the continuous valley regions as an independent connected cluster and assign a unique cluster ID; Based on the distribution location of connected clusters, they are divided into salt density clusters of different risk levels, including low, medium and high risk salt density clusters.

[0012] Further, the salt density cluster level to which the new sample belongs is calculated, and the predicted salt density value is output, including: Receive real-time collected conductivity and environmental parameters and use them as new samples; Online matching is initiated using the optimal salt density clustering model. All known connected cluster structures that have been resolved in the model are traversed. The reachability distance of the new sample relative to the core object of each connected cluster is calculated. All reachability distances are compared and analyzed. The connected cluster with the smallest reachability distance is selected as the target cluster to which the new sample belongs, so as to determine the salt density cluster level to which the new sample belongs. The cluster salt density mapping mechanism is invoked. Based on the unique cluster ID of the target cluster, the historical average salt density value corresponding to the cluster is retrieved and extracted as the salt density benchmark value. This salt density benchmark value is then directly assigned as the predicted salt density value of the current electrical insulation equipment.

[0013] Furthermore, the multi-level threshold discrimination mechanism includes: Compare the predicted salt density value with the preset first and second security thresholds: Level 1 detection: When the predicted salt density value is greater than or equal to the first safety threshold, an abnormal alarm signal is triggered; Secondary discrimination: When the predicted salt density value is greater than the first safety threshold and less than the second safety threshold, the detection cycle of the lifting platform control dissolving cup is automatically shortened; Level 3 discrimination: When the predicted salt density value is less than or equal to the first safety threshold, the device is determined to be in a safe operating range. The current sample is incorporated into the historical database using a moving average strategy, and the baseline mean of the corresponding salt density cluster is updated to achieve adaptive calibration of the model.

[0014] Furthermore, environmental parameters include temperature, humidity, wind speed, and pollutant concentration.

[0015] (III) Beneficial Effects This invention provides a salt density analysis method for electrical insulation equipment based on an online conductivity algorithm, which has the following advantages: This invention uses a lifting platform to control the immersion of the insulator in a dissolving cup. The system collects raw data such as the conductivity of the solution and environmental parameters according to a preset detection cycle. Subsequently, the data enters the preprocessing stage, where calibration, temperature compensation, and other operations are performed to provide high-quality input for model analysis. This invention imports conductivity and environmental parameters into the optimal salt density clustering model, effectively identifies natural clustering patterns of data and filters outlier noise interference by calculating core distance and reachability distance, thereby improving the robustness of the model. It constructs a quantitative bridge from unsupervised geometric structure to supervised physical benchmark, ensuring that the prediction results have clear physical meaning and numerical accuracy. It can achieve online salt density cluster level identification based on the reachability distance matching of new samples, making the monitoring results highly real-time, providing a reliable basis for subsequent risk warning and operation and maintenance decisions, and improving the system's anti-interference ability and operational robustness. This invention uses the optimal hyperparameter combination obtained by the Particle Swarm Optimization-Grey Wolf algorithm hybrid optimization to construct the OPTICS clustering model, thereby training a clustering and prediction system adapted to the salt density scenario of insulators. In this process, the silhouette coefficient is used as the evaluation index, and the iteration stops when the predetermined maximum number of iterations is reached; otherwise, it returns to the PSO-MOGWO hybrid parameter optimization stage to continue iterating, thereby improving the robustness and classification accuracy of insulator salt density state identification. This invention, by using salt density prediction values ​​combined with a multi-level threshold discrimination mechanism to trigger abnormal alarms or dynamically adjust processing strategies, triggers maintenance early warnings, thereby achieving accurate identification and risk control of insulator salt density status and ensuring the operational reliability of external grid insulation equipment. Attached Figure Description

[0016] Figure 1 This is a schematic diagram illustrating the steps of a salt density analysis method for electrical insulation equipment according to an exemplary embodiment. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: This invention provides a method for salt density analysis of electrical insulation equipment based on an online conductivity algorithm; Figure 1 This is a schematic diagram illustrating the steps of a salt density analysis method for electrical insulation equipment according to an exemplary embodiment; please refer to... Figure 1 The method includes the following steps: S1: When the insulator enters the detection cycle, the dissolving cup and the environmental sensor array are controlled by the lifting platform to synchronously collect conductivity and environmental parameters; among which, the environmental parameters include temperature, humidity, wind speed and pollutant concentration; Step S1 includes: The lifting platform synchronously triggers and controls the dissolving cup and the environmental sensor array; During the testing cycle, the lifting platform and dissolving cup assembly are controlled by PLC or industrial control computer. Under the trigger of the lifting platform, the dissolving cup contacts the surface of the insulator through a sequence of actions of rising, immersing, emulsifying and falling to obtain the solution conductivity, which is used to indirectly reflect the salt density. Ascending: The lifting platform raises the dissolving cup containing deionized water to the insulator position; Immersion: Ensure that the surface of the insulator to be tested is completely immersed in water; Emulsification: Maintain a certain time period to allow the salt on the surface of the insulator to fully dissolve in the water, forming the solution to be tested; Descent: After dissolution is complete, the dissolution vessel falls back down, and the sensor probe is connected to the solution; It should be noted that data is collected every time interval T, where T is the time it takes for the lifting platform to complete the entire detection cycle of rising, immersion, emulsification, and falling, in order to ensure that the insulator is fully immersed and the salt is efficiently dissolved. Environmental parameters are collected by deploying an array of environmental sensors, including temperature and humidity sensors, wind speed sensors, and pollutant concentration sensors. None of these sensors are shown in the diagram, and they are installed adaptively. Preprocessing of conductivity and environmental parameters, including anomaly removal, missing data imputation, and normalization; Outlier removal: Sensor fault data is removed using the 3σ criterion; Missing data filling: For gaps caused by data transmission packet loss, KNN interpolation is used to fill in the missing values; Normalization: Due to the huge differences in the dimensions of temperature, humidity, wind speed and pollutant concentration, in order to avoid the large numerical features dominating the model, Min-Max is used to map all data to the interval [0, 1]. The significance of the above analysis is that: by controlling the dissolving cup to immerse the insulator through the lifting platform, the system collects raw data such as the conductivity of the solution and environmental parameters according to the preset detection cycle; then, the data enters the preprocessing stage to perform calibration, temperature compensation and other operations, providing high-quality input for model analysis.

[0019] S2: Construct a hybrid parameter optimization model, use the particle swarm optimization-multi-objective gray wolf algorithm to iteratively optimize the hyperparameter combination of the OPTICS clustering algorithm, use the silhouette coefficient as the fitness evaluation index, output the optimal hyperparameter combination when the index converges, and configure it into the kernel of the OPTICS clustering algorithm to generate the optimal salt density clustering model. The hyperparameter combination includes the maximum neighborhood radius and the reachable distance threshold; Maximum neighborhood radius: Defines the neighborhood range of a sample point; Reachable distance threshold: Defines the dividing line between clusters; Iterative optimization of hyperparameter combinations in the OPTICS clustering algorithm using the particle swarm optimization-multi-objective gray wolf algorithm includes: Constructing the parameter space and initializing the population: Determine the hyperparameter combination of the OPTICS clustering algorithm to construct a two-dimensional parameter search space; use a uniform distribution function to randomly generate an initial parameter population in the two-dimensional parameter search space, set the maximum number of iterations to Tmax, the population consists of m particles, the population contains Q0 particles, and each particle m represents a set of potential parameter solutions, represented as a position vector, and an initial velocity vector is randomly assigned. Fitness assessment: Perform OPTICS clustering on each particle and extract the silhouette coefficient S from the clustering results. In the formula, is the average distance within the cluster, which refers to the average distance between sample i and other samples in the same cluster. The smaller the value, the higher the homogeneity within the cluster. is the average distance between the nearest heterogeneous clusters, which refers to the average distance between sample i and the nearest heterogeneous cluster sample. The larger the value, the more significant the difference between clusters; n represents the sample size; max{·} represents maximizing intra-cluster homogeneity and inter-cluster variability, that is, the intra-cluster compactness and inter-cluster separation used to quantify the clustering results. Physical meaning: S takes values ​​from -1 to 1, with values ​​closer to 1 indicating better clustering results. In the formula, For cluster C i The number of samples in the middle; j is the number of samples used to traverse cluster C. i Inside; d(x) i ,x j ) is the sample x i and sample x j Distance metric between; In the formula, k is the index of other clusters; C k For sample k, it belongs to a cluster other than the cluster it belongs to. For clusters The number of samples in the middle; j is the number of clusters traversed C. k An index of all samples within; The arithmetic mean of the silhouette coefficients of all sample points in the dataset is used as the fitness of the population evolution: Global Search: The particle swarm optimization algorithm is launched to perform a global search. The fitness value of all particles in the swarm is calculated in turn. Based on the fitness value, the individual historical best position Pbest of each particle and the global best position Gbest of the entire swarm are determined. Using inertia weight and learning factor, the velocity vector and position vector of particle m are updated based on the individual historical best position Pbest and the global best position Gbest, driving the entire swarm to drift towards the region with higher fitness, thus completing the coarse-grained capture of the parameter space features. Local Mining: A multi-objective gray wolf algorithm is introduced to perform local mining on the population after global search. This includes: sorting the updated population according to fitness values ​​from high to low; selecting the top N0 particles (usually N0 is set to 3), which are then mapped to Alpha wolves, Beta wolves, and Delta wolves, respectively, as the leadership layer guiding the population movement; defining the remaining particles as the Omega wolf pack as the follower layer; calculating the distance vector between the Omega wolves and the three leader wolves; and dynamically adjusting the position vector of each Omega wolf under the coordinated guidance of the leadership layer to make it approach the global optimum, thereby achieving a refined search of the dominant region. Threshold judgment: Real-time monitoring of fitness value changes. When the increment of fitness value is less than the preset convergence threshold or the number of iterations reaches Tmax, the algorithm is judged to have converged. The current position vector of Alpha Wolf is locked and output as the optimal hyperparameter combination. Finally, the optimal hyperparameter combination is configured into the kernel of the OPTICS algorithm to instantiate and generate the optimal salt density clustering model for subsequent salt density state identification. The significance of the above analysis lies in the following: by using the optimal hyperparameter combination obtained through the Particle Swarm Optimization-Grey Wolf algorithm hybrid optimization, and constructing the OPTICS clustering model, a clustering and prediction system adapted to the salt density scenario of insulators is trained; the iteration stops when the predetermined maximum number of iterations is reached; otherwise, it returns to the PSO-MOGWO hybrid parameter optimization stage to continue iterating. By continuously repeating the iteration process, using the silhouette coefficient as the evaluation index, the shortcomings of traditional clustering models that rely on human experience and are prone to getting trapped in local optima are effectively overcome. The silhouette coefficient-guided mechanism maximizes the intra-class homogeneity and inter-class differences of salt density clusters, ensuring that the generated model can accurately adapt to the changing environmental parameter characteristics, thereby significantly improving the robustness and classification accuracy of insulator salt density state identification.

[0020] S3: Input conductivity and environmental parameters into the optimal salt density clustering model to determine the core distance and reachability distance in order to resolve the connected cluster structure. Through the cluster salt density mapping mechanism, the unsupervised clusters are transformed into supervised salt density benchmark values. Combined with the reachability distance, the salt density cluster level to which the new sample belongs is matched, and the predicted salt density value is output. Determining the core distance and reachable distance includes: By calling the maximum neighborhood radius, local density detection is performed on each sample point in the sample space, including: Neighborhood search: Traverse the sample space, count the number of neighboring samples of any sample point within the maximum neighborhood radius, and divide the sample space into several neighborhood ranges. Core object determination: If the number of neighboring samples is greater than or equal to the preset minimum number of points threshold, the corresponding sample point is marked as a core object, and the Euclidean distance between it and its nearest neighbor with the minimum number of points threshold is calculated and defined as the core distance of the sample point; if the number of neighboring samples is less than the minimum number of points threshold, the sample point is marked as a noise point, and its core distance is recorded as infinity 1; it should be noted that the core distance reflects the minimum radius required for the sample point to reach the minimum density, and the smaller the value, the higher the local density; Distance calculation: For the core object, all sample points within the maximum neighborhood radius are sorted sequentially, and the minimum number of points threshold is marked as N1. The Euclidean distance between the N1th sample point and the nearest sample point is taken as the core distance; where N1 is a positive integer greater than 0. Based on the core distance, the connection strength between any two adjacent sample pairs in the sample space is calculated, including: Object selection: Select a core object and a test sample within its neighborhood; Original distance calculation: Calculate the Euclidean distance between the sample to be tested and the core object, which is the straight-line distance in geometric space; Maximum value comparison: Compare the core distance of the core object with the Euclidean distance between the current core point and the test samples in its neighborhood, and take the larger value as the reachable distance of the sample point of the core object relative to the test sample; it should be noted that the reachable distance ensures that the distance from the core object to the neighbor is at least equal to the core distance, thereby smoothing density fluctuations, making the growth of the cluster more stable, and effectively resisting noise interference. The connected cluster structure was resolved, including: Based on reachability distance, a reachability distance graph is generated using a sorting operation; where the horizontal axis of the reachability distance graph represents the processing order of samples, and the vertical axis represents the numerical value of reachability distance. The optimal reachability threshold of the hybrid parameter optimization model is used to cut the reachability distance map into valleys. Sample segments with continuous reachability distances significantly lower than the optimal reachability threshold are identified as high-density regions, i.e., valley regions. Continuous valley regions are marked as independent connected clusters, and low-density regions with reachability distances higher than the optimal reachability threshold, i.e., discrete noise points, are filtered out. At the same time, a unique cluster ID is assigned to each connected cluster. Based on the distribution position of each cluster in the feature space, they are divided into salt density level cluster structures representing different risk levels, including low, medium, and high risk salt density level clusters. The unsupervised clustering model is transformed into a supervised salt density benchmark through a cluster salt density mapping mechanism, including: Historical label extraction: Several connected cluster structures are parsed out. By traversing each parsed connected cluster, the measured salt density labels corresponding to all historical samples contained in the cluster are extracted. It should be noted that although OPTICS clustering is based on conductivity and environmental parameters (unsupervised features), in the historical dataset during the training phase, each sample is associated with a measured surface salt density value (i.e., a real label obtained through manual or laboratory measurement). Salt density benchmark calculation: Calculate the arithmetic mean of all measured salt density tags and define it as the salt density benchmark value of the cluster to eliminate the influence of single-point measurement errors and random fluctuations; Mapping establishment: Construct a mapping table between cluster ID and salt density baseline value, for example: cluster ID = 1 - baseline value 0.03 mg / cm³. 2 Cluster ID=2 — Baseline value 0.15 mg / cm³ 2 It realizes the transformation of abstract mathematical clustering results into specific physical salt density values; Calculate the salt density cluster level to which the new sample belongs, and output the predicted salt density value, including: Receive real-time collected conductivity and environmental parameters and use them as new samples; Online matching is initiated using the optimal salt-dense clustering model. All known connected cluster structures that have been resolved in the model are traversed, and the reachability distance of the new sample relative to the core object of each connected cluster is calculated. The reachability distance is the larger of the intrinsic core distance of each core object and the Euclidean distance from the new sample to the core object. This is used to quantify the degree of correlation between the new sample and each established salt-dense pattern. By comparing all reachability distances, the connected cluster with the smallest reachability distance is selected as the target cluster to which the new sample belongs. The new sample is determined to have the highest homogeneity with the target cluster in the density space, so as to determine the salt density cluster level to which the new sample belongs. The cluster salt density mapping mechanism is invoked. Based on the unique cluster ID of the target cluster, the historical average salt density value corresponding to the cluster is retrieved and extracted as the salt density benchmark value. This salt density benchmark value is directly assigned as the predicted salt density value of the current electrical insulation equipment. The significance of the above analysis lies in the following: by importing conductivity and environmental parameters into the optimal salt density clustering model, the model effectively identifies natural clustering patterns of data and filters outlier noise interference through the calculation of core distance and reachability distance, thereby improving the robustness of the model. It also constructs a quantitative bridge from unsupervised geometric structure to supervised physical benchmark, ensuring that the prediction results have clear physical meaning and numerical accuracy. Furthermore, it enables online salt density cluster level identification based on the reachability distance of new samples, resulting in highly real-time monitoring results. This provides a reliable basis for subsequent risk warning and operation and maintenance decisions, and improves the system's anti-interference ability and operational robustness.

[0021] S4: Based on the predicted salt density value and combined with multi-level risk thresholds, trigger maintenance early warning; Compare the predicted salt density value with the preset first and second security thresholds: Level 1 Judgment: When the predicted salt density value is greater than or equal to the first safety threshold, it is determined that the salt density accumulation on the surface of the insulator has reached or exceeded the insulation withstand limit. When encountering damp weather (such as fog, dew, drizzle), it is very easy to cause a flashover accident. This is an emergency fault precursor, triggering an abnormal alarm signal, immediately activating the audible and visual alarm, and sending the highest level red warning to the monitoring center through the SCADA system. The current sample is marked as a high-risk sample, associated with the insulator number to generate an insulator cleaning work order, and packaged and fed back to the operation and maintenance terminal. Secondary discrimination: When the predicted salt density value is greater than the first safety threshold and less than the second safety threshold, it is determined that the trend of dirt accumulation is obvious. If the original low-frequency sampling is maintained, the abrupt change point from quantitative to qualitative change may be missed, triggering a strategy to automatically shorten the detection cycle of the lifting platform control dissolving cup. For example, the default sampling once every 24 hours is automatically adjusted to sampling once every 4 hours. Level 3 discrimination: When the predicted salt density value is less than or equal to the first safety threshold, it is determined that the insulator surface is clean or slightly dirty and the equipment is in a safe operating range. The current sample is incorporated into the historical database using a moving average strategy, and the baseline mean of the corresponding salt density cluster is updated to achieve adaptive calibration of the model. By comparing the alarm accuracy and false alarm rate under different thresholds through cross-validation, the threshold that optimizes the accuracy is selected as the first safety threshold and the second safety threshold. The significance of the above analysis lies in the fact that by triggering abnormal alarms or dynamically adjusting processing strategies through a multi-level threshold discrimination mechanism, maintenance early warnings can be triggered, enabling accurate identification and risk control of insulator salt density status, and ensuring the operational reliability of external grid insulation equipment.

[0022] In the application, the various formulas mentioned are all calculated by removing dimensions and taking their numerical values. The formulas are derived from the most recent real-world situation by collecting a large amount of data and conducting software simulations. The formulas are set by those skilled in the art according to the actual situation.

[0023] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0024] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0025] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for salt density analysis of electrical insulation equipment based on online conductivity algorithm, characterized in that, The method includes: When the insulator enters the testing cycle, the dissolving cup and the environmental sensor array are controlled by the lifting platform to simultaneously collect conductivity and environmental parameters. A hybrid parameter optimization model is constructed, and the hyperparameter combination of the OPTICS clustering algorithm is iteratively optimized using the particle swarm optimization-multi-objective gray wolf algorithm. The optimal hyperparameter combination is output and configured into the kernel of the OPTICS clustering algorithm to generate the optimal salt density clustering model. Electrical conductivity and environmental parameters are imported into the optimal salt density clustering model to determine the core distance and reachability distance in order to resolve the connected cluster structure. Through the cluster salt density mapping mechanism, the unsupervised clusters are transformed into supervised salt density benchmark values. Combined with the reachability distance, the salt density cluster level to which the new sample belongs is matched, and the predicted salt density value is output. Based on the predicted salt density value, combined with a multi-level threshold discrimination mechanism, anomaly alarms are triggered or the processing strategy is dynamically adjusted.

2. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 1, characterized in that, The collected conductivity and environmental parameters include: The lifting platform synchronously triggers and controls the dissolving cup and the environmental sensor array; Under the triggering of the lifting platform, the dissolving cup contacts the surface of the insulator through a sequence of actions including rising, immersing, emulsifying, and falling to obtain the conductivity of the solution. An environmental sensor array measures environmental parameters in real time and aligns them with conductivity data using timestamps. The conductivity and environmental parameters are preprocessed.

3. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 1, characterized in that, The hyperparameter combination includes the maximum neighborhood radius and the reachable distance threshold.

4. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 3, characterized in that, The optimal combination of output hyperparameters includes: Determine the hyperparameter combination of the OPTICS clustering algorithm to construct a two-dimensional parameter search space; The particle swarm optimization algorithm is used to perform a global search. An initial population is randomly generated in the two-dimensional parameter search space. The population consists of m particles, and each particle has a unique position vector and velocity vector. At the same time, the OPTICS clustering algorithm is used to determine the clustering results of each particle to extract the silhouette coefficient and calculate the fitness. A multi-objective gray wolf algorithm is introduced for local mining: the population is sorted in descending order of fitness, the top N0 particles are selected and converted into the initial leader wolf position vector, and dynamic calibration is performed. If the maximum number of iterations is reached, the population is guided to move by adjusting the wolf position vector; otherwise, the optimal hyperparameter combination is output; where N0 is a positive integer greater than 0.

5. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 1, characterized in that, The determination of the core distance and reachable distance includes: Determine the core distance: Call the maximum neighborhood radius to traverse the sample space and count the number of neighborhood samples of any sample point in the sample space within the maximum neighborhood radius. The number of neighborhood samples is compared with a preset minimum number of points threshold: if the number of neighborhood samples is greater than or equal to the preset minimum number of points threshold, the sample point is determined to be a core object; otherwise, the sample point is determined to be a noise point. Based on the core object, all sample points within the maximum neighborhood radius are sorted sequentially, and the minimum number of points threshold is marked as N1. The Euclidean distance between the N1th sample point and the nearest sample point is taken as the core distance; where N1 is a positive integer greater than 0. Determine reachable distance: Based on core distance: Select the core object and sample points in its neighborhood, and calculate the Euclidean distance between them; compare the core distance of the core object with the Euclidean distance, and take the larger value as the reachable distance of the sample point relative to the core object point.

6. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 5, characterized in that, The parsed connected cluster structure includes: Construct a reachability graph based on reachability distance; The optimal reachability threshold of the hybrid parameter optimization model is called to cut the valleys in the reachability distance map, identify the valley regions, and mark the continuous valley regions as an independent connected cluster and assign a unique cluster ID; Based on the distribution location of connected clusters, they are divided into salt density clusters of different risk levels, including low, medium and high risk salt density clusters.

7. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 1, characterized in that, The calculation of the salt density cluster level to which the new sample belongs and the output of the predicted salt density value include: Receive real-time collected conductivity and environmental parameters and use them as new samples; Online matching is initiated using the optimal salt density clustering model. All known connected cluster structures that have been resolved in the model are traversed. The reachability distance of the new sample relative to the core object of each connected cluster is calculated. All reachability distances are compared and analyzed. The connected cluster with the smallest reachability distance is selected as the target cluster to which the new sample belongs, so as to determine the salt density cluster level to which the new sample belongs. The cluster salt density mapping mechanism is invoked. Based on the unique cluster ID of the target cluster, the historical average salt density value corresponding to the cluster is retrieved and extracted as the salt density benchmark value. This salt density benchmark value is then directly assigned as the predicted salt density value of the current electrical insulation equipment.

8. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 1, characterized in that, The multi-level threshold discrimination mechanism includes: Compare the predicted salt density value with the preset first and second security thresholds: Level 1 detection: When the predicted salt density value is greater than or equal to the first safety threshold, an abnormal alarm signal is triggered; Secondary discrimination: When the predicted salt density value is greater than the first safety threshold and less than the second safety threshold, the detection cycle of the lifting platform control dissolving cup is automatically shortened; Level 3 discrimination: When the predicted salt density value is less than or equal to the first safety threshold, the device is determined to be in a safe operating range. The current sample is incorporated into the historical database using a moving average strategy, and the baseline mean of the corresponding salt density cluster is updated to achieve adaptive calibration of the model.

9. The method for salt density analysis of electrical insulation equipment based on online conductivity algorithm according to claim 1, characterized in that, The environmental parameters include temperature, humidity, wind speed, and pollutant concentration.

Images