Insulator surface contamination state analysis system and method based on hyperspectral imaging technology

Abstract

The invention relates to an insulator surface contamination state analysis system and method based on hyperspectral imaging technology, belonging to the technical field of insulator detection. The invention realizes a comprehensive consideration of factors affecting the contamination state of insulators through multi-dimensional data collection. In terms of weight calculation, for temperature and humidity data, advanced fuzzy clustering analysis and support vector regression model are used to accurately quantify their influence on the degree of contamination, and can adapt to complex environmental changes. For geographical location weight, factors such as distance from pollution source, topography and water flow are comprehensively considered to make the evaluation result more in line with the actual situation. The calculation of insulator material weight uses hierarchical structure and fuzzy judgment matrix to scientifically measure the influence of key performance of different materials on the contamination state, which can accurately evaluate the degree of contamination of insulators, provide a reliable basis for cleaning or replacement decisions, and effectively ensure the safe and stable operation of power systems.

Description

Technical Field

[0001] The present invention relates to the technical field of insulator detection, and in particular to an insulator surface contamination state analysis system and method based on hyperspectral imaging technology. Background Art

[0002] Insulators are one of the most important external insulating devices in the power system. The quality of their insulation performance directly affects whether the transmission line can operate safely and stably. The flashover problem of insulators has always been a concern, especially when pollution flashover occurs, the probability of line reclosing is low, which can easily cause long-term and large-scale power outages, causing huge losses to the power system. With the further development of industrial society, the concentration of pollutant particles in the atmosphere in some areas has increased and the composition has become more complex.

[0003] At present, it is a common method to detect the contamination status of insulator surfaces by using drones carrying hyperspectral imaging equipment. Although the contamination status of insulators can be accurately analyzed, the environment in which insulators are located is complex and diverse, including different climatic conditions, topography, and distribution of pollution sources. For example, insulators in coastal areas have an increase in conductive substances on the surface due to the influence of salt spray and high humidity, resulting in a significant decrease in insulation resistance, which is more likely to cause damage to the power system. Therefore, how to make the insulator contamination analysis system adapt to various complex environments and accurately evaluate the degree of contamination of insulators in different environments is a technical problem that needs to be solved. Summary of the invention

[0004] The purpose of the present invention is to provide an insulator surface contamination status analysis system and method based on hyperspectral imaging technology, which solves the problem that the existing insulator contamination status analysis system in the background technology is not convenient for comprehensive evaluation in combination with other external factors, resulting in errors in its results and ultimately affecting the normal operation of the power system.

[0005] To achieve the above object, the present invention provides the following technical solution: an insulator surface contamination state analysis system based on hyperspectral imaging technology, comprising:

[0006] Insulator preliminary detection unit: used to perform preliminary analysis on the degree of contamination of insulators after processing, analyzing and calculating the acquired images. The insulator preliminary detection unit includes a data acquisition module, a data analysis and processing module, and a judgment and evaluation module;

[0007] Weight calculation unit: The weight calculation unit includes temperature and humidity data weight calculation module, geographic location weight calculation module and insulator material weight calculation module. The function of the weight calculation unit is to determine the quantitative value of the importance of each influencing factor in the insulator contamination analysis;

[0008] Comprehensive weight calculation unit: used to accurately measure the influence of temperature and humidity, geographical location, distance from pollution source and insulator material factors on the degree of pollution, so as to comprehensively obtain a comprehensive weight that more accurately reflects the actual pollution status;

[0009] Comprehensive evaluation result output unit: Combines the preliminary test results of the insulator with the comprehensive weight, obtains the final pollution level evaluation result through calculation, divides different pollution level intervals according to the final evaluation result value, determines the current pollution level of the insulator, and outputs a detailed evaluation report;

[0010] Among them, the data acquisition module is used for:

[0011] Obtain insulator image information and insulator material information taken by the hyperspectral imaging equipment carried by the drone, and obtain local factory location information and pollutant emission database;

[0012] Obtain temperature and humidity data monitored by the temperature and humidity sensor carried by the drone;

[0013] Obtain the geographic location information of the insulator through the GIS system carried by the drone, which includes the distance and direction relationship between the insulator and the surrounding buildings and terrain features;

[0014] The calculation steps of the temperature and humidity data weight calculation module are as follows:

[0015] S1. Data preprocessing: Collect a large number of samples containing temperature and humidity data and corresponding insulator contamination degree labeled data, clean and normalize the data, map both the temperature and humidity data and the contamination degree data to the [0,1] interval to eliminate the influence of different dimensions and data ranges, and randomly divide the preprocessed data into training set and test set in a ratio of 7:3;

[0016] S2. Fuzzy clustering analysis: Use the fuzzy C-means clustering algorithm to determine the number of clusters K, run the fuzzy clustering algorithm, cluster the temperature and humidity data in the training set, and each data point x i =(T i , H i ) are assigned to a membership vector u ik , indicating the degree to which the data point belongs to the kth cluster, T i Indicates temperature, H i Indicates humidity, u ik The update formula is ,in Represents data point x i To cluster center C k =(T k , H k), m is the fuzzy index;

[0017] S3. Feature extraction and representation: For each cluster k, calculate its cluster center C k =(T k , H k ), where T k and H k are the mean values ​​of temperature and humidity in the cluster, respectively. For each data point x i , construct a new feature vector , ;

[0018] S4. Support vector regression model training: using the feature vectors of the data points in the training set and the corresponding pollution level value y i To train the support vector regression model, choose the Gaussian kernel function ,in is the kernel function parameter, which is determined by cross-validation. The other parameters of the support vector regression model are determined by trying different value combinations within the parameter range, selecting the parameter combination that performs best on the validation set, and using the training data and the selected parameters to train the support vector regression model to obtain a model function that predicts the degree of contamination based on the input feature vector. ,in and is the Lagrange multiplier, is the kernel parameter, b is the bias term;

[0019] S5. Model prediction evaluation: For the temperature and humidity data in the test set, fuzzy clustering analysis is first performed to obtain its membership vector and position information relative to the cluster center, and construct the feature vector , input the feature vector into the trained support vector regression model to obtain the predicted pollution level value , the root mean square error is used to evaluate the prediction performance of the model, and the calculation formula is , where m is the number of data points in the test set, is the actual level of filthiness, The predicted pollution degree value is based on the evaluation results, and the parameters of fuzzy clustering and support vector regression are adjusted, or the data preprocessing and feature extraction steps are repeated until a satisfactory model is obtained.

[0020] Furthermore, the geographic location weight calculation module determines the geographic location coordinates (x1, y1) of the insulator using the GIS system, and determines the emission characteristics and location coordinates (x2, y2) of the pollution source according to the local factory location information and the pollutant emission database;

[0021] The geographic location weight calculation module uses Euclidean distance to calculate the straight-line distance between the insulator and the pollution source. ;

[0022] The geographical location weight calculation module considers the correction of distance by topographic factors and introduces the terrain influence factor f terrain , the actual effective distance becomes d=d×f terrain For water bodies such as rivers, the water flow influencing factor f is introduced water， The actual effective distance becomes d=d / f water ;

[0023] The geographical location weight calculation module corrects the emission intensity according to the type of pollution source, and the original emission intensity E is multiplied by the correction coefficient c type Get E adjusted =E×c type , considering the influence of time factors, introducing the time function g(t), the final emission intensity E final =E×c type ×g(t);

[0024] The weight calculation formula of the geographical location weight calculation module is: , where k is the adjustment coefficient.

[0025] Furthermore, the calculation of the insulator material weight calculation module includes: establishing a hierarchical structure, constructing a fuzzy judgment matrix and calculating a weight vector, and the establishment of the hierarchical structure includes:

[0026] Target layer: Determine the comprehensive evaluation of insulator material weights;

[0027] Criteria layer: select key criteria such as pollution adsorption capacity, electrical insulation performance, mechanical strength and weather resistance;

[0028] Solution layer: various insulator materials;

[0029] The fuzzy judgment matrix is ​​constructed by inviting experts to compare different insulator materials in pairs under each criterion, giving fuzzy judgment values, which are expressed by triangular fuzzy numbers (l, m, u), and constructing a fuzzy matrix A=(a ij ), for each row element of the fuzzy judgment matrix, calculate its geometric mean, and normalize the calculated geometric mean, that is, divide each geometric mean by the sum of all geometric mean values ​​to obtain a weight vector. The geometric mean calculation formula is: , the weight vector obtained by normalizing the geometric mean The comprehensive weight calculation steps are as follows: Assume that for k criteria, the weight of the i-th material under the j-th criterion is w ij , the importance weight of the jth criterion is , then the comprehensive weight is .

[0030] Furthermore, the comprehensive weight calculation unit performs an overall calculation on the weights calculated by the temperature and humidity data weight calculation module, the geographic location weight calculation module and the insulator material weight calculation module, and the calculation formula is: , , and are the distribution coefficients of temperature and humidity data weight, geographical location weight, and insulator material weight, and Finally, the insulator surface contamination degree value calculated by the insulator preliminary detection unit is multiplied by the weight calculated by the comprehensive weight calculation unit to obtain the final insulator contamination degree value. The contamination state is determined based on the value and it is judged whether cleaning or replacement is required.

[0031] Another technical solution proposed by the present invention: a method for analyzing the contamination state of an insulator surface based on hyperspectral imaging technology, comprising the following steps:

[0032] S1. Preliminary inspection of insulators:

[0033] S11. Data collection: Obtain insulator image information and material information through the hyperspectral imaging equipment carried by the drone, obtain local factory location information and pollutant emission database, monitor temperature and humidity data through temperature and humidity sensors, and obtain the geographical location information of the insulator through the GIS system;

[0034] S12. Analysis and processing: The collected data is processed by the data analysis and processing module, and the degree of contamination of the insulator is preliminarily analyzed by the judgment and evaluation module;

[0035] S2. Weight calculation:

[0036] S21. Calculation of temperature and humidity data weights:

[0037] S211. Data preprocessing: collect samples, clean and normalize data, and divide them into training and test sets;

[0038] S212. Fuzzy clustering analysis: The fuzzy C-means clustering algorithm is used to cluster the temperature and humidity data;

[0039] S213. Feature extraction and representation: Calculate cluster centers and construct new feature vectors of data points;

[0040] S214. Support vector regression model training: Use the training set data to train the model, select the Gaussian kernel function and determine the parameters;

[0041] S215. Model prediction evaluation: Make predictions on the test set data, evaluate model performance and adjust parameters;

[0042] S22. Geographic location weight calculation:

[0043] S221. Use GIS system to determine the location coordinates of insulators and pollution sources;

[0044] S222. Use Euclidean distance to calculate straight-line distance, and take into account the correction of topography and water flow factors on the distance;

[0045] S223. Modify the emission intensity according to the type of pollution source and time factors;

[0046] S224. Calculate the geographical location weight according to the weight calculation formula;

[0047] S23. Insulator material weight calculation:

[0048] S231. Establish a hierarchical structure, including the goal layer, the criteria layer and the program layer;

[0049] S232. Construct a fuzzy judgment matrix, invite experts to compare different materials in pairs under each criterion and give fuzzy judgment values;

[0050] S233. Calculate the weight vector, including calculating the geometric mean and performing normalization;

[0051] S3. Comprehensive weight calculation: Calculate the weights of temperature and humidity, geographical location and insulator material as a whole to obtain a comprehensive weight;

[0052] S4. Comprehensive evaluation result output: Combine the preliminary test results of the insulator with the comprehensive weight, calculate and obtain the final contamination level evaluation result, divide the contamination level interval, determine the contamination level and output a detailed evaluation report, and then determine whether cleaning or replacement is needed.

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

[0054] The insulator surface contamination state analysis system and method based on hyperspectral imaging technology provided by the present invention realizes a comprehensive consideration of factors affecting the contamination state of insulators through multi-dimensional data collection, including images and material information obtained by hyperspectral imaging equipment, temperature and humidity sensor monitoring data, GIS system positioning information, and factory and pollutant emission data. In terms of weight calculation, advanced fuzzy clustering analysis and support vector regression models are used for temperature and humidity data to accurately quantify their impact on the degree of contamination, and can adapt to complex environmental changes. For the geographical location weight, factors such as the distance from the pollution source, topography and water flow are comprehensively considered to make the evaluation results more in line with the actual situation. The calculation of the insulator material weight uses a hierarchical structure and a fuzzy judgment matrix to scientifically measure the impact of key performance of different materials on the contamination state, which can accurately evaluate the degree of contamination of insulators, provide a reliable basis for cleaning or replacement decisions, and effectively ensure the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of a module of the present invention;

[0056] Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] In order to solve the technical problem that the existing insulator contamination status analysis system is not convenient for comprehensive evaluation combined with other external factors, resulting in errors in its results, which ultimately affects the normal operation of the power system, such as Figure 1-Figure 2 As shown, the following preferred technical solutions are provided:

[0059] The insulator surface contamination status analysis system based on hyperspectral imaging technology includes:

[0060] Insulator preliminary detection unit: used to perform preliminary analysis on the degree of contamination of insulators after processing, analyzing and calculating the acquired images. The insulator preliminary detection unit includes a data acquisition module, a data analysis and processing module, and a judgment and evaluation module;

[0061] Weight calculation unit: The weight calculation unit includes temperature and humidity data weight calculation module, geographic location weight calculation module and insulator material weight calculation module. The function of the weight calculation unit is to determine the quantitative value of the importance of each influencing factor in the insulator contamination analysis;

[0062] Comprehensive weight calculation unit: used to accurately measure the influence of temperature and humidity, geographical location, distance from pollution source and insulator material factors on the degree of pollution, so as to comprehensively obtain a comprehensive weight that more accurately reflects the actual pollution status;

[0063] Comprehensive evaluation result output unit: Combines the preliminary test results of the insulator with the comprehensive weight, obtains the final pollution level evaluation result through calculation, divides different pollution level intervals according to the final evaluation result value, determines the current pollution level of the insulator, and outputs a detailed evaluation report;

[0064] Among them, the data acquisition module is used for:

[0065] Obtain insulator image information and insulator material information taken by the hyperspectral imaging equipment carried by the drone, and obtain local factory location information and pollutant emission database;

[0066] Obtain temperature and humidity data monitored by the temperature and humidity sensor carried by the drone;

[0067] Obtain the geographic location information of the insulator through the GIS system carried by the drone, which includes the distance and direction relationship between the insulator and the surrounding buildings and terrain features;

[0068] The calculation steps of the temperature and humidity data weight calculation module are as follows:

[0069] S1. Data preprocessing: Collect a large number of samples containing temperature and humidity data and corresponding insulator contamination degree labeled data, clean and normalize the data, map both the temperature and humidity data and the contamination degree data to the [0,1] interval to eliminate the influence of different dimensions and data ranges, and randomly divide the preprocessed data into training set and test set in a ratio of 7:3;

[0070] S2. Fuzzy clustering analysis: Use the fuzzy C-means clustering algorithm to determine the number of clusters K, run the fuzzy clustering algorithm, cluster the temperature and humidity data in the training set, and each data point x i =(T i , H i ) are assigned to a membership vector u ik , indicating the degree to which the data point belongs to the kth cluster, T i Indicates temperature, H i Indicates humidity, u ik The update formula is ,in Represents data point x i To cluster center C k =(T k , Hk ), m is the fuzzy index;

[0071] S3. Feature extraction and representation: For each cluster k, calculate its cluster center C k =(T k , H k ), where T k and H k are the mean values ​​of temperature and humidity in the cluster, respectively. For each data point x i , construct a new feature vector , ;

[0072] S4. Support vector regression model training: using the feature vectors of the data points in the training set and the corresponding pollution level value y i To train the support vector regression model, choose the Gaussian kernel function ,in is the kernel function parameter, which is determined by cross-validation. The other parameters of the support vector regression model are determined by trying different value combinations within the parameter range, selecting the parameter combination that performs best on the validation set, and using the training data and the selected parameters to train the support vector regression model to obtain a model function that predicts the degree of contamination based on the input feature vector. ,in and is the Lagrange multiplier, is the kernel parameter, b is the bias term;

[0073] S5. Model prediction evaluation: For the temperature and humidity data in the test set, fuzzy clustering analysis is first performed to obtain its membership vector and position information relative to the cluster center, and construct the feature vector , input the feature vector into the trained support vector regression model to obtain the predicted pollution level value , the root mean square error is used to evaluate the prediction performance of the model, and the calculation formula is , where m is the number of data points in the test set, is the actual level of filthiness, The predicted pollution degree value is based on the evaluation results, and the parameters of fuzzy clustering and support vector regression are adjusted, or the data preprocessing and feature extraction steps are repeated until a satisfactory model is obtained.

[0074] The geographic location weight calculation module uses the GIS system to determine the geographic location coordinates (x1, y1) of the insulator, and determines the emission characteristics and location coordinates (x2, y2) of the pollution source according to the local factory location information and the pollutant emission database.

[0075] The geographic location weight calculation module uses Euclidean distance to calculate the straight-line distance between the insulator and the pollution source. .

[0076] The geographical location weight calculation module considers the correction of distance by topographic factors and introduces the terrain influence factor f terrain , the actual effective distance becomes d=d×f terrain For water bodies such as rivers, the water flow influencing factor f is introduced water， The actual effective distance becomes d=d / f water .

[0077] The geographical location weight calculation module corrects the emission intensity according to the type of pollution source, and the original emission intensity E is multiplied by the correction coefficient c type Get E adjusted =E×c type , considering the influence of time factors, introducing the time function g(t), the final emission intensity E final =E×c type ×g(t).

[0078] The weight calculation formula of the geographical location weight calculation module is: , where k is the adjustment coefficient.

[0079] The calculation of the insulator material weight calculation module includes: establishing a hierarchical structure, constructing a fuzzy judgment matrix and calculating a weight vector, wherein the establishment of the hierarchical structure includes:

[0080] Target layer: Determine the comprehensive evaluation of insulator material weights;

[0081] Criteria layer: select key criteria such as pollution adsorption capacity, electrical insulation performance, mechanical strength and weather resistance;

[0082] Solution layer: various insulator materials.

[0083] The fuzzy judgment matrix is ​​constructed by inviting experts to compare different insulator materials in pairs under each criterion, giving fuzzy judgment values, which are expressed by triangular fuzzy numbers (l, m, u), and constructing a fuzzy matrix A=(a ij ), for each row element of the fuzzy judgment matrix, calculate its geometric mean, and normalize the calculated geometric mean, that is, divide each geometric mean by the sum of all geometric mean values ​​to obtain a weight vector. The geometric mean calculation formula is: .

[0084] The weight vector obtained by normalizing the geometric mean The comprehensive weight calculation steps are as follows: Assume that for k criteria, the weight of the i-th material under the j-th criterion is w ij, the importance weight of the jth criterion is , then the comprehensive weight is .

[0085] The comprehensive weight calculation unit performs overall calculation on the weights calculated by the temperature and humidity data weight calculation module, the geographical location weight calculation module and the insulator material weight calculation module, and its calculation formula is: , , and are the distribution coefficients of temperature and humidity data weight, geographical location weight, and insulator material weight, and Finally, the insulator surface contamination degree value calculated by the insulator preliminary detection unit is multiplied by the weight calculated by the comprehensive weight calculation unit to obtain the final insulator contamination degree value. The contamination state is determined based on the value and it is judged whether cleaning or replacement is required.

[0086] Another technical solution proposed by the present invention: a method for analyzing the contamination state of an insulator surface based on hyperspectral imaging technology, comprising the following steps:

[0087] S1. Preliminary inspection of insulators:

[0088] S11. Data collection: Obtain insulator image information and material information through the hyperspectral imaging equipment carried by the drone, obtain local factory location information and pollutant emission database, monitor temperature and humidity data through temperature and humidity sensors, and obtain the geographical location information of the insulator through the GIS system;

[0089] S12. Analysis and processing: The collected data is processed by the data analysis and processing module, and the degree of contamination of the insulator is preliminarily analyzed by the judgment and evaluation module;

[0090] S2. Weight calculation:

[0091] S21. Calculation of temperature and humidity data weights:

[0092] S211. Data preprocessing: collect samples, clean and normalize data, and divide them into training and test sets;

[0093] S212. Fuzzy clustering analysis: The fuzzy C-means clustering algorithm is used to cluster the temperature and humidity data;

[0094] S213. Feature extraction and representation: Calculate cluster centers and construct new feature vectors of data points;

[0095] S214. Support vector regression model training: Use the training set data to train the model, select the Gaussian kernel function and determine the parameters;

[0096] S215. Model prediction evaluation: Make predictions on the test set data, evaluate model performance and adjust parameters;

[0097] S22. Geographic location weight calculation:

[0098] S221. Use GIS system to determine the location coordinates of insulators and pollution sources;

[0099] S222. Use Euclidean distance to calculate straight-line distance, and take into account the correction of topography and water flow factors on the distance;

[0100] S223. Modify the emission intensity according to the type of pollution source and time factors;

[0101] S224. Calculate the geographical location weight according to the weight calculation formula;

[0102] S23. Insulator material weight calculation:

[0103] S231. Establish a hierarchical structure, including the goal layer, the criteria layer and the program layer;

[0104] S232. Construct a fuzzy judgment matrix, invite experts to compare different materials in pairs under each criterion and give fuzzy judgment values;

[0105] S233. Calculate the weight vector, including calculating the geometric mean and performing normalization;

[0106] S3. Comprehensive weight calculation: Calculate the weights of temperature and humidity, geographical location and insulator material as a whole to obtain a comprehensive weight;

[0107] S4. Comprehensive evaluation result output: Combine the preliminary test results of the insulator with the comprehensive weight, calculate and obtain the final contamination level evaluation result, divide the contamination level interval, determine the contamination level and output a detailed evaluation report, and then determine whether cleaning or replacement is needed.

[0108] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0109] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. The insulator surface contamination status analysis system based on hyperspectral imaging technology is characterized by: include: Insulator preliminary detection unit: used to perform preliminary analysis on the degree of contamination of insulators after processing, analyzing and calculating the acquired images. The insulator preliminary detection unit includes a data acquisition module, a data analysis and processing module, and a judgment and evaluation module; Weight calculation unit: The weight calculation unit includes temperature and humidity data weight calculation module, geographic location weight calculation module and insulator material weight calculation module. The function of the weight calculation unit is to determine the quantitative value of the importance of each influencing factor in the insulator contamination analysis; Comprehensive weight calculation unit: used to accurately measure the influence of temperature and humidity, geographical location, distance from pollution source and insulator material factors on the degree of pollution, so as to comprehensively obtain a comprehensive weight that more accurately reflects the actual pollution status; Comprehensive evaluation result output unit: Combines the preliminary test results of the insulator with the comprehensive weight, obtains the final pollution level evaluation result through calculation, divides different pollution level intervals according to the final evaluation result value, determines the current pollution level of the insulator, and outputs a detailed evaluation report; Among them, the data acquisition module is used for: Obtain insulator image information and insulator material information taken by the hyperspectral imaging equipment carried by the drone, and obtain local factory location information and pollutant emission database; Obtain temperature and humidity data monitored by the temperature and humidity sensor carried by the drone; Obtain the geographic location information of the insulator through the GIS system carried by the drone, which includes the distance and direction relationship between the insulator and the surrounding buildings and topographic features; The calculation steps of the temperature and humidity data weight calculation module are as follows: S1. Data preprocessing: Collect a large number of samples containing temperature and humidity data and corresponding insulator contamination degree labeled data, clean and normalize the data, map both the temperature and humidity data and the contamination degree data to the [0,1] interval to eliminate the influence of different dimensions and data ranges, and randomly divide the preprocessed data into training set and test set in a ratio of 7:3; S2. Fuzzy clustering analysis: Use the fuzzy C-means clustering algorithm to determine the number of clusters K, run the fuzzy clustering algorithm, cluster the temperature and humidity data in the training set, and each data point x i =(T i , H i ) are assigned to a membership vector u ik , indicating the degree to which the data point belongs to the kth cluster, T i Indicates temperature, H i Indicates humidity, u ik The update formula is ,in Represents data point x i To cluster center C k =(T k , H k ), m is the fuzzy index; S3. Feature extraction and representation: For each cluster k, calculate its cluster center C k =(T k , H k ), where T k and H k are the mean values ​​of temperature and humidity in the cluster, respectively. For each data point x i , construct a new feature vector , ; S4. Support vector regression model training: using the feature vectors of the data points in the training set and the corresponding pollution level value y i To train the support vector regression model, choose the Gaussian kernel function ,in is the kernel function parameter, which is determined by cross-validation. The other parameters of the support vector regression model are determined by trying different value combinations within the parameter range, selecting the parameter combination that performs best on the validation set, and using the training data and the selected parameters to train the support vector regression model to obtain a model function that predicts the degree of contamination based on the input feature vector. ,in and is the Lagrange multiplier, is the kernel parameter, b is the bias term; S5. Model prediction evaluation: For the temperature and humidity data in the test set, fuzzy clustering analysis is first performed to obtain its membership vector and position information relative to the cluster center, and construct the feature vector , input the feature vector into the trained support vector regression model to obtain the predicted pollution level value , the root mean square error is used to evaluate the prediction performance of the model, and the calculation formula is , where m is the number of data points in the test set, is the actual level of filthiness, The predicted pollution degree value is based on the evaluation results, and the parameters of fuzzy clustering and support vector regression are adjusted, or the data preprocessing and feature extraction steps are repeated until a satisfactory model is obtained.

2. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 1, characterized in that: The geographic location weight calculation module uses the GIS system to determine the geographic location coordinates (x1, y1) of the insulator, and determines the emission characteristics and location coordinates (x2, y2) of the pollution source according to the local factory location information and the pollutant emission database.

3. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 2, characterized in that: The geographic location weight calculation module uses Euclidean distance to calculate the straight-line distance between the insulator and the pollution source. .

4. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 3 is characterized in that: The geographic location weight calculation module considers the correction of distance by topographic factors and introduces the terrain influence factor f terrain , the actual effective distance becomes d=d×f terrain For water bodies such as rivers, the water flow influencing factor f is introduced water， The actual effective distance becomes d=d / f water .

5. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 4, characterized in that: The geographical location weight calculation module corrects the emission intensity according to the type of pollution source, and the original emission intensity E is multiplied by the correction coefficient c type Get E adjusted =E×c type , considering the influence of time factors, introducing the time function g(t), the final emission intensity E final =E×c type ×g(t).

6. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 5, characterized in that: The weight calculation formula of the geographical location weight calculation module is: , where k is the adjustment coefficient.

7. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 6, characterized in that: The calculation of the insulator material weight calculation module includes: establishing a hierarchical structure, constructing a fuzzy judgment matrix and calculating a weight vector, wherein the establishment of the hierarchical structure includes: Target layer: Determine the comprehensive evaluation of insulator material weights; Criteria layer: select key criteria such as pollution adsorption capacity, electrical insulation performance, mechanical strength and weather resistance; Solution layer: various insulator materials.

8. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 7, characterized in that: The fuzzy judgment matrix is ​​constructed by inviting experts to compare different insulator materials in pairs under each criterion, giving fuzzy judgment values, which are expressed by triangular fuzzy numbers (l, m, u), and constructing a fuzzy matrix A=(a ij ), for each row element of the fuzzy judgment matrix, calculate its geometric mean, and normalize the calculated geometric mean, that is, divide each geometric mean by the sum of all geometric mean values ​​to obtain a weight vector. The geometric mean calculation formula is: .

9. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 8, characterized in that: The weight vector obtained by normalizing the geometric mean The comprehensive weight calculation steps are as follows: Assume that for k criteria, the weight of the i-th material under the j-th criterion is w ij , the importance weight of the jth criterion is , the comprehensive weight is .

10. The insulator surface contamination state analysis system based on hyperspectral imaging technology according to claim 9, characterized in that: The comprehensive weight calculation unit performs overall calculation on the weights calculated by the temperature and humidity data weight calculation module, the geographical location weight calculation module and the insulator material weight calculation module, and its calculation formula is: , , and are the distribution coefficients of temperature and humidity data weight, geographical location weight, and insulator material weight, and Finally, the insulator surface contamination degree value calculated by the insulator preliminary detection unit is multiplied by the weight calculated by the comprehensive weight calculation unit to obtain the final insulator contamination degree value. The contamination state is determined based on the value and it is judged whether cleaning or replacement is required.

Images