A method for detecting the charge of a lightning arrester

Abstract

The present application discloses a method for detecting live lightning arresters, comprising: obtaining historical detection logs of live lightning arrester detection and historical detection data in the historical detection logs, preprocessing the historical detection data to obtain initial detection data; classifying the initial detection data according to detection type, and extracting characteristic detection data and characteristic factors of the characteristic detection data in different historical detection logs of each detection type; performing detection quality evaluation on the characteristic factors, and determining the optimal detection type and preferred detection data based on the detection quality evaluation results; constructing a predictive detection model, obtaining real-time detection data based on the optimal detection type, and inputting the data into the predictive detection model to obtain a predictive detection result; calculating the credibility of the predictive detection result, improving detection accuracy, reducing detection costs and interference of the surrounding environment on detection, ensuring the detection accuracy of the lightning arrester state, and improving the operational reliability of the lightning arrester.

Description

Technical Field

[0001] The present application relates to the technical field of lightning arrester live detection, and in particular to a lightning arrester live detection method. Background Art

[0002] Lightning arresters are critical electrical devices used to protect electrical equipment from operational and atmospheric overvoltages. Their operational reliability directly impacts power grid security. Long-term operation in high voltage and harsh environments can lead to internal moisture buildup and insulation degradation. Therefore, proper lightning arrester testing is crucial. Existing methods for detecting live lightning arresters suffer from low measurement accuracy, long testing times, and high costs. Summary of the Invention

[0003] In order to solve the above technical problems, the present application provides a method for detecting the power supply of a lightning arrester. By analyzing the characteristic detection data and characteristic factors of each detection type, the detection quality evaluation results of the characteristic detection data are determined, the optimal detection type and preferred detection data are obtained, and a predictive detection model is constructed to improve the detection accuracy, reduce the detection cost and the interference of the surrounding environment on the detection, ensure the detection accuracy of the lightning arrester status, and improve the operation reliability of the lightning arrester.

[0004] In some embodiments of the present application, a method for detecting a charged lightning arrester is provided, comprising:

[0005] Obtain historical detection logs of live detection of lightning arresters and historical detection data in the historical detection logs, preprocess the historical detection data, and obtain initial detection data;

[0006] Classify the initial detection data according to the detection type, and extract the characteristic detection data and characteristic factors of the characteristic detection data from different historical detection logs of each detection type;

[0007] Performing a quality evaluation on the characteristic factors of the characteristic detection data, and determining the optimal detection type and the corresponding preferred detection data based on the detection quality evaluation results;

[0008] Generate a prediction detection model based on the preferred detection data and the corresponding detection results, obtain the real-time detection data of the current lightning arrester according to the optimal detection type, and input it into the prediction detection model to obtain the prediction detection results;

[0009] Calculate the credibility of the predicted detection results and determine whether to send an alarm signal based on the credibility.

[0010] In some embodiments of the present application, the preprocessing includes normalizing time units, removing outliers, filling missing values, and data cleaning.

[0011] In some embodiments of the present application, the initial detection data is classified according to the detection type, and characteristic detection data and characteristic factors of the characteristic detection data in different historical detection logs of each detection type are extracted, including:

[0012] Obtain the initial detection data of the same historical detection log, analyze the data source of the initial detection data, determine the sensor type of the initial detection data on the same historical detection day, and determine the detection type of the initial detection data in the corresponding historical detection log based on the sensor type;

[0013] Preset feature detection data for each detection type;

[0014] Analyze the initial detection data in different historical detection logs of the corresponding detection type according to the preset feature detection data to obtain feature detection data in different historical detection logs of the corresponding detection type;

[0015] Preset a preset collection time interval for each detection type, obtain feature detection data from the same historical detection log of the corresponding detection type according to the preset collection time interval, and map it to the time reference line constructed by the historical detection duration in the corresponding historical detection log, and generate a feature detection data reference graph of multiple historical detection logs of the current detection type;

[0016] The characteristic detection data reference graphs of multiple historical detection logs of each detection type are analyzed to determine the characteristic factors of the corresponding characteristic detection data.

[0017] In some embodiments of the present application, determining a characteristic factor corresponding to the characteristic detection data includes:

[0018] Obtaining the integrity of the feature detection data in the feature detection data reference graph, and when the integrity is greater than a preset integrity threshold, generating a feature detection data change curve for the corresponding feature detection data reference graph;

[0019] Acquire multiple fluctuation values ​​of adjacent preset acquisition time intervals of each characteristic detection data change curve, generate a stability evaluation value based on the multiple fluctuation values, and set the stability evaluation value as a first characteristic factor of the corresponding characteristic detection data;

[0020] Obtain the historical detection accuracy, historical detection cost, and historical detection timeliness of each feature detection data change curve, generate a detection evaluation value for the corresponding feature detection data change curve based on the historical detection accuracy, historical detection cost, and historical detection timeliness, and set the detection evaluation value as the second characteristic factor of the corresponding feature detection data;

[0021] Obtain historical environmental information within the historical detection period of each feature detection data change curve, determine the environmental information interval in which the historical environmental information is located, and the mutation time node of the historical environmental information within the historical detection period;

[0022] The mutation time node is marked in the corresponding feature detection data change curve, and the change value of the feature detection data at the corresponding mutation time node is obtained. An anti-interference evaluation value is generated according to the change values ​​at multiple mutation time nodes in the same feature detection data change curve, and the anti-interference evaluation value is set as the third characteristic factor of the corresponding feature detection data.

[0023] In some embodiments of the present application, the detection quality evaluation of the characteristic factors of the characteristic detection data includes:

[0024] Comparing the first characteristic factor, the second characteristic factor, and the third characteristic factor of each feature detection data with the first standard characteristic factor, the second standard characteristic factor, and the third standard characteristic factor of the preset feature detection data of the corresponding detection type, respectively, to obtain a first characteristic factor difference, a second characteristic factor difference, and a third characteristic factor difference;

[0025] Determine a corresponding detection quality evaluation library according to the detection type of each feature detection data, the detection quality evaluation library including a plurality of preset first feature factor difference values, preset second feature factor difference values, and preset third feature factor difference values ​​of the preset feature detection data, and each of the preset first feature factor difference values, the preset second feature factor difference values, and the preset third feature factor difference values ​​is associated with a specific preset detection quality evaluation value;

[0026] Perform similarity analysis on the first characteristic factor difference, the second characteristic factor difference, and the third characteristic factor difference of each characteristic detection data with the preset first characteristic factor difference, the preset second characteristic factor difference, and the preset third characteristic factor difference in the corresponding detection quality evaluation library, respectively, and combine the preset detection quality evaluation values ​​of the preset first characteristic factor difference, the preset second characteristic factor difference, and the preset third characteristic factor difference with the greatest similarity with the weight coefficients of the first characteristic factor, the second characteristic factor, and the third characteristic factor to generate a detection quality evaluation result;

[0027] The test quality evaluation results are:

[0028] ;

[0029] Among them, H is the detection quality evaluation result, a1 is the preset detection quality evaluation value of the preset first characteristic factor difference with the greatest similarity to the first characteristic factor difference, c1 is the weight coefficient of the first characteristic factor, a2 is the preset detection quality evaluation value of the preset second characteristic factor difference with the greatest similarity to the second characteristic factor difference, c2 is the weight coefficient of the second characteristic factor, a3 is the preset detection quality evaluation value of the preset third characteristic factor difference with the greatest similarity to the third characteristic factor difference, and c3 is the weight coefficient of the third characteristic factor.

[0030] In some embodiments of the present application, before determining the optimal detection type and the corresponding preferred detection data based on the detection quality evaluation results, the method further includes:

[0031] When the integrity is less than a preset integrity threshold, the corresponding feature detection data reference image is removed, and the integrity difference of the removed feature detection data reference image is calculated;

[0032] Obtain the number of rejections of the feature detection data reference image in each detection type and the corresponding integrity difference to generate a compensation coefficient for the corresponding detection type;

[0033] The calculation formula of the compensation coefficient is:

[0034] ;

[0035] Among them, r is the compensation coefficient, z is the compensation conversion coefficient, n is the total number of all feature detection data reference images in the detection type, m is the eliminated feature detection data reference image, W0 is the preset integrity threshold, and W1i is the integrity of the i-th eliminated feature detection data reference image.

[0036] The detection quality evaluation result of the characteristic detection data of the corresponding detection type is corrected according to the compensation coefficient, and the corrected detection quality evaluation result is H*r.

[0037] In some embodiments of the present application, determining the optimal detection type and corresponding preferred detection data based on the detection quality evaluation results includes:

[0038] Generate a comprehensive test quality evaluation result of the corresponding test type based on multiple revised test quality evaluation results of the same test type;

[0039] Ranking the comprehensive test quality evaluation results of multiple test types, and setting the test type ranked first as the optimal test type;

[0040] Compare the multiple corrected detection quality evaluation results of the optimal detection type with the preset detection quality evaluation result threshold, and set the feature detection data corresponding to the detection quality evaluation results greater than the preset detection quality evaluation result threshold as the preferred detection data.

[0041] In some embodiments of the present application, inputting into a prediction detection model to obtain a prediction detection result includes:

[0042] Generate a training data set based on the preferred detection data and the corresponding historical detection results, train a neural network model based on the training data set, and obtain a predictive detection model;

[0043] The detection tester involved in the optimal detection type obtains real-time feature detection data of the lightning arrester, and the real-time feature detection data is input into the prediction detection model to obtain the prediction detection result.

[0044] In some embodiments of the present application, calculating the reliability of the predicted test result includes:

[0045] Get the historical test results in the previous historical period before the current predicted test result;

[0046] Determining adjacent historical states and historical change characteristics of adjacent historical states within the previous historical period based on adjacent historical detection results within the previous historical period, wherein the historical change characteristics include a historical state change trend, a historical state fluctuation degree, and a historical state change rate;

[0047] Generate a first state evaluation value according to a historical state change trend;

[0048] Generate a second state evaluation value according to the historical state fluctuation degree;

[0049] generating a third state evaluation value according to the historical state change rate;

[0050] generating a plurality of comprehensive state evaluation values ​​of adjacent historical detection results according to the first state evaluation value, the second state evaluation value, and the third state evaluation value;

[0051] Setting a normal variation interval of the comprehensive state evaluation value according to multiple comprehensive state evaluation values ​​of adjacent historical detection results;

[0052] Generate a predicted comprehensive state evaluation value based on the predicted detection result and the previous adjacent historical detection result;

[0053] Determine whether the predicted comprehensive state evaluation value is within the normal variation range of the comprehensive state evaluation value. If so, the credibility is greater than the preset credibility threshold; if not, the credibility is less than the preset credibility threshold.

[0054] In some embodiments of the present application, determining whether to send an alarm signal based on credibility includes:

[0055] When the credibility is greater than the preset credibility threshold, the real-time status of the arrester is generated according to the predicted detection results;

[0056] When the real-time status is a fault state, an alarm signal is sent; when the real-time status is a normal state, an early warning signal is sent and the next preset detection time interval is set; when the real-time status is a good state, no signal is sent;

[0057] When the credibility is less than the preset credibility threshold, the prediction detection model is rebuilt and the prediction detection result is reacquired until the credibility is greater than the preset credibility threshold.

[0058] Compared with the prior art, the method for detecting a charged lightning arrester according to the embodiment of the present application has the following advantages:

[0059] By analyzing the characteristic detection data and characteristic factors of each detection type, determining the detection quality evaluation results of the characteristic detection data, obtaining the optimal detection type and preferred detection data, and constructing a predictive detection model, the detection accuracy is improved, the detection cost and the interference of the surrounding environment on the detection are reduced, the detection accuracy of the lightning arrester status is guaranteed, and the operation reliability of the lightning arrester is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a flow chart of a method for detecting a charged lightning arrester in a preferred embodiment of the present application. DETAILED DESCRIPTION

[0061] The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0062] In the description of this application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application.

[0063] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.

[0064] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances.

[0065] like Figure 1 As shown, a lightning arrester charging detection method according to a preferred embodiment of the present application includes:

[0066] Step S101: obtaining a historical detection log of live detection of a lightning arrester and historical detection data in the historical detection log, and preprocessing the historical detection data to obtain initial detection data;

[0067] Step S102: classifying the initial detection data according to detection type, and extracting characteristic detection data and characteristic factors of the characteristic detection data from different historical detection logs of each detection type;

[0068] Step S103: Performing a detection quality evaluation on the characteristic factors of the characteristic detection data, and determining the optimal detection type and the corresponding preferred detection data according to the detection quality evaluation result;

[0069] Step S104: Generate a prediction detection model based on the preferred detection data and the corresponding detection results, obtain real-time detection data of the current lightning arrester according to the optimal detection type, and input it into the prediction detection model to obtain a prediction detection result;

[0070] Step S105: Calculate the credibility of the predicted detection result and determine whether to send an alarm signal based on the credibility.

[0071] In this embodiment, the historical detection log of the lightning arrester's energized detection is obtained based on different detection schemes of multiple different detection types, including third harmonic detection, self-compensation detection, secondary voltage detection, etc. The detection scheme in each detection type is determined based on a comprehensive analysis of the actual environment and actual parameters of the lightning arrester. The historical detection data in multiple historical detection logs are analyzed and the optimal detection type of the current lightning arrester and the corresponding preferred detection data are determined. The corresponding predictive detection model is constructed to improve the accurate judgment of the lightning arrester status and ensure its normal operation and safe use.

[0072] In some embodiments of the present application, the preprocessing includes normalizing time units, removing outliers, filling missing values, and data cleaning.

[0073] In some embodiments of the present application, the initial detection data is classified according to the detection type, and characteristic detection data and characteristic factors of the characteristic detection data in different historical detection logs of each detection type are extracted, including:

[0074] Obtain the initial detection data of the same historical detection log, analyze the data source of the initial detection data, determine the sensor type of the initial detection data on the same historical detection day, and determine the detection type of the initial detection data in the corresponding historical detection log based on the sensor type;

[0075] Preset feature detection data for each detection type;

[0076] Analyze the initial detection data in different historical detection logs of the corresponding detection type according to the preset feature detection data to obtain feature detection data in different historical detection logs of the corresponding detection type;

[0077] Preset a preset collection time interval for each detection type, obtain feature detection data from the same historical detection log of the corresponding detection type according to the preset collection time interval, and map it to the time reference line constructed by the historical detection duration in the corresponding historical detection log, and generate a feature detection data reference graph of multiple historical detection logs of the current detection type;

[0078] The characteristic detection data reference graphs of multiple historical detection logs of each detection type are analyzed to determine the characteristic factors of the corresponding characteristic detection data.

[0079] In this embodiment, the preset feature detection data refers to the data in the detection data of each detection type that is strongly correlated with the detection results. For example, when the detection type is third harmonic detection, the preset feature detection data is the third harmonic component in the leakage current. By determining the feature detection data in multiple historical detection logs in each detection type and calculating the characteristic factors of the corresponding feature detection data, the corresponding detection type and the detection quality evaluation results of the corresponding feature detection data are generated according to the characteristic factors, laying the foundation for the subsequent construction of a predictive detection model, accelerating the detection efficiency of the lightning arrester and improving the detection accuracy.

[0080] In some embodiments of the present application, determining a characteristic factor corresponding to the characteristic detection data includes:

[0081] Obtaining the integrity of the feature detection data in the feature detection data reference graph, and when the integrity is greater than a preset integrity threshold, generating a feature detection data change curve for the corresponding feature detection data reference graph;

[0082] Acquire multiple fluctuation values ​​of adjacent preset acquisition time intervals of each characteristic detection data change curve, generate a stability evaluation value based on the multiple fluctuation values, and set the stability evaluation value as a first characteristic factor of the corresponding characteristic detection data;

[0083] Obtain the historical detection accuracy, historical detection cost, and historical detection timeliness of each feature detection data change curve, generate a detection evaluation value for the corresponding feature detection data change curve based on the historical detection accuracy, historical detection cost, and historical detection timeliness, and set the detection evaluation value as the second characteristic factor of the corresponding feature detection data;

[0084] Obtain historical environmental information within the historical detection period of each feature detection data change curve, determine the environmental information interval in which the historical environmental information is located, and the mutation time node of the historical environmental information within the historical detection period;

[0085] The mutation time node is marked in the corresponding feature detection data change curve, and the change value of the feature detection data at the corresponding mutation time node is obtained. An anti-interference evaluation value is generated according to the change values ​​at multiple mutation time nodes in the same feature detection data change curve, and the anti-interference evaluation value is set as the third characteristic factor of the corresponding feature detection data.

[0086] In this embodiment, the preset integrity threshold refers to the minimum data continuity of the feature detection data obtained for each detection type in the corresponding detection time that has detection significance. When the data continuity is strong, it means that the integrity is high, and when the data continuity is low, it means that the integrity is low. According to the relationship between the integrity and the preset integrity threshold, unqualified detection data in the historical detection log of each detection type is screened out to avoid affecting the accuracy of the detection quality evaluation results of the detection type and feature detection data.

[0087] In this embodiment, the fluctuation value refers to the data difference between the feature detection data of adjacent preset collection time intervals. The larger the fluctuation value and the more fluctuation values ​​that are greater than the preset fluctuation value threshold, the smaller the stability evaluation value, which means that the stability of the feature detection data of the current detection type is low, and the accuracy of the corresponding detection results in the later period is reduced. The smaller the fluctuation value and the more fluctuation values ​​that are less than the preset fluctuation value threshold, the larger the stability evaluation value, which means that the stability of the feature detection data of the current detection type is high, and the accuracy of the corresponding detection results in the later period is increased.

[0088] In this embodiment, historical detection accuracy refers to the difference between the historical detection results corresponding to the current feature detection data and the standard detection results. When the difference in the detection results is small, the historical detection accuracy is high, and when the difference in the detection results is large, the historical detection accuracy is low. Historical detection timeliness refers to the timeliness of the historical detection results corresponding to the feature detection data of the current detection type. Historical detection cost refers to the cost of obtaining the feature detection data of the corresponding detection type. The detection evaluation value is calculated based on the historical detection accuracy, historical detection cost and historical detection timeliness combined with the corresponding weight coefficient. When the historical detection accuracy is high, the historical detection cost is low and the historical detection timeliness is fast, the detection evaluation value is larger. Otherwise, the detection evaluation value is smaller.

[0089] In this embodiment, historical environmental information includes the ambient temperature and humidity of the lightning arrester during the detection process, that is, environmental information that affects the detection effect of the lightning arrester. The mutation time node refers to the time point when the change in historical environmental information in the historical detection period is greater than the preset change threshold. According to the change value of the feature detection data at the mutation time node, the interference of the environmental information change on the feature detection data of the current detection type can be judged, that is, the impact evaluation value. When the change value at multiple mutation time nodes is larger, the anti-interference evaluation value is smaller, and vice versa.

[0090] In this embodiment, by determining the first characteristic factor, second characteristic factor and third characteristic factor of each feature detection data, the foundation is laid for the subsequent generation of the detection quality evaluation value result, thereby screening out the detection types, preferred detection data and corresponding detection results that are stable, accurate, low-cost and low-interference under real-time environmental information, constructing predicted detection results, reducing the analysis and processing volume of real-time detection data, and speeding up the judgment of detection results.

[0091] In some embodiments of the present application, the detection quality evaluation of the characteristic factors of the characteristic detection data includes:

[0092] Comparing the first characteristic factor, the second characteristic factor, and the third characteristic factor of each feature detection data with the first standard characteristic factor, the second standard characteristic factor, and the third standard characteristic factor of the preset feature detection data of the corresponding detection type, respectively, to obtain a first characteristic factor difference, a second characteristic factor difference, and a third characteristic factor difference;

[0093] Determine a corresponding detection quality evaluation library according to the detection type of each feature detection data, the detection quality evaluation library including a plurality of preset first feature factor difference values, preset second feature factor difference values, and preset third feature factor difference values ​​of the preset feature detection data, and each of the preset first feature factor difference values, the preset second feature factor difference values, and the preset third feature factor difference values ​​is associated with a specific preset detection quality evaluation value;

[0094] Perform similarity analysis on the first characteristic factor difference, the second characteristic factor difference, and the third characteristic factor difference of each characteristic detection data with the preset first characteristic factor difference, the preset second characteristic factor difference, and the preset third characteristic factor difference in the corresponding detection quality evaluation library, respectively, and combine the preset detection quality evaluation values ​​of the preset first characteristic factor difference, the preset second characteristic factor difference, and the preset third characteristic factor difference with the greatest similarity with the weight coefficients of the first characteristic factor, the second characteristic factor, and the third characteristic factor to generate a detection quality evaluation result;

[0095] The test quality evaluation results are:

[0096] ;

[0097] Among them, H is the detection quality evaluation result, a1 is the preset detection quality evaluation value of the preset first characteristic factor difference with the greatest similarity to the first characteristic factor difference, c1 is the weight coefficient of the first characteristic factor, a2 is the preset detection quality evaluation value of the preset second characteristic factor difference with the greatest similarity to the second characteristic factor difference, c2 is the weight coefficient of the second characteristic factor, a3 is the preset detection quality evaluation value of the preset third characteristic factor difference with the greatest similarity to the third characteristic factor difference, and c3 is the weight coefficient of the third characteristic factor.

[0098] In this embodiment, the detection quality evaluation library for each detection type is set in advance according to the detection quality standard, and the corresponding detection quality evaluation results are obtained, which lays the foundation for the subsequent screening of the optimal detection type and the optimization of detection data, improves the detection quality of the lightning arrester, and ensures the accuracy of its status diagnosis.

[0099] In some embodiments of the present application, before determining the optimal detection type and the corresponding preferred detection data based on the detection quality evaluation results, the method further includes:

[0100] When the integrity is less than a preset integrity threshold, the corresponding feature detection data reference image is removed, and the integrity difference of the removed feature detection data reference image is calculated;

[0101] Obtain the number of rejections of the feature detection data reference image in each detection type and the corresponding integrity difference to generate a compensation coefficient for the corresponding detection type;

[0102] The calculation formula of the compensation coefficient is:

[0103] ;

[0104] Among them, r is the compensation coefficient, z is the compensation conversion coefficient, n is the total number of all feature detection data reference images in the detection type, m is the eliminated feature detection data reference image, W0 is the preset integrity threshold, and W1i is the integrity of the i-th eliminated feature detection data reference image.

[0105] The detection quality evaluation result of the characteristic detection data of the corresponding detection type is corrected according to the compensation coefficient, and the corrected detection quality evaluation result is H*r.

[0106] In this embodiment, the detection quality evaluation results of the detection type are corrected based on the data of the reference graph of the eliminated feature detection data of each detection type, so as to improve the accuracy of the detection quality evaluation results of the feature detection data within the detection type, and lay the foundation for the subsequent determination of the optimal detection type and the preferred detection data.

[0107] In some embodiments of the present application, determining the optimal detection type and corresponding preferred detection data based on the detection quality evaluation results includes:

[0108] Generate a comprehensive test quality evaluation result of the corresponding test type based on multiple revised test quality evaluation results of the same test type;

[0109] Ranking the comprehensive test quality evaluation results of multiple test types, and setting the test type ranked first as the optimal test type;

[0110] Compare the multiple corrected detection quality evaluation results of the optimal detection type with the preset detection quality evaluation result threshold, and set the feature detection data corresponding to the detection quality evaluation results greater than the preset detection quality evaluation result threshold as the preferred detection data.

[0111] In this embodiment, the optimal detection type is determined by multiple revised detection quality evaluation results of each detection type, and the preferred detection data is determined based on the multiple revised detection quality evaluation results of the optimal detection type, that is, detection types and detection data with high data stability, high result accuracy, low cost and low interference are screened out, and a predictive detection model is constructed to reduce the amount of detection data analysis and processing, speed up the judgment of detection results, and promptly determine the operating status of the lightning arrester to ensure its normal operation.

[0112] In some embodiments of the present application, inputting into a prediction detection model to obtain a prediction detection result includes:

[0113] Generate a training data set based on the preferred detection data and the corresponding historical detection results, train a neural network model based on the training data set, and obtain a predictive detection model;

[0114] The detection tester involved in the optimal detection type obtains real-time feature detection data of the lightning arrester, and the real-time feature detection data is input into the prediction detection model to obtain the prediction detection result.

[0115] In some embodiments of the present application, calculating the reliability of the predicted test result includes:

[0116] Get the historical test results in the previous historical period before the current predicted test result;

[0117] Determining adjacent historical states and historical change characteristics of adjacent historical states within the previous historical period based on adjacent historical detection results within the previous historical period, wherein the historical change characteristics include a historical state change trend, a historical state fluctuation degree, and a historical state change rate;

[0118] Generate a first state evaluation value according to a historical state change trend;

[0119] Generate a second state evaluation value according to the historical state fluctuation degree;

[0120] generating a third state evaluation value according to the historical state change rate;

[0121] generating a plurality of comprehensive state evaluation values ​​of adjacent historical detection results according to the first state evaluation value, the second state evaluation value, and the third state evaluation value;

[0122] Setting a normal variation interval of the comprehensive state evaluation value according to multiple comprehensive state evaluation values ​​of adjacent historical detection results;

[0123] Generate a predicted comprehensive state evaluation value based on the predicted detection result and the previous adjacent historical detection result;

[0124] Determine whether the predicted comprehensive state evaluation value is within the normal variation range of the comprehensive state evaluation value. If so, the credibility is greater than the preset credibility threshold; if not, the credibility is less than the preset credibility threshold.

[0125] In this embodiment, the historical states determined by the historical detection results include a good state, a general state, and a fault state. The historical state change trend refers to an upward trend and a downward trend. For example, when the adjacent historical state changes from a good state to a fault state, it is a downward trend. The historical state fluctuation degree refers to the fluctuation level of the historical state. For example, when the adjacent historical state changes from a good state to a fault state, it is a first-level fluctuation level, and when the adjacent historical state changes from a good state to a fault state, it is a second-level fluctuation level. The historical state change rate refers to the fluctuation degree of the adjacent historical state / the detection interval period of the adjacent detection results.

[0126] In this embodiment, the accuracy of the predicted detection result is determined by the credibility, thereby ensuring the accuracy of the diagnosis of the arrester status.

[0127] In some embodiments of the present application, determining whether to send an alarm signal based on credibility includes:

[0128] When the credibility is greater than the preset credibility threshold, the real-time status of the arrester is generated according to the predicted detection results;

[0129] When the real-time status is a fault state, an alarm signal is sent; when the real-time status is a normal state, an early warning signal is sent and the next preset detection time interval is set; when the real-time status is a good state, no signal is sent;

[0130] When the credibility is less than the preset credibility threshold, the prediction detection model is rebuilt and the prediction detection result is reacquired until the credibility is greater than the preset credibility threshold.

[0131] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and replacements can be made without departing from the technical principles of the present application. These improvements and replacements should also be regarded as the scope of protection of the present application.

Claims

1. A lightning arrester live detection method, characterized in that: include: Obtain historical detection logs of live detection of lightning arresters and historical detection data in the historical detection logs, preprocess the historical detection data, and obtain initial detection data; Classify the initial detection data according to the detection type, and extract the characteristic detection data and characteristic factors of the characteristic detection data from different historical detection logs of each detection type; Performing a quality evaluation on the characteristic factors of the characteristic detection data, and determining the optimal detection type and the corresponding preferred detection data based on the detection quality evaluation results; Generate a prediction detection model based on the preferred detection data and the corresponding detection results, obtain the real-time detection data of the current lightning arrester according to the optimal detection type, and input it into the prediction detection model to obtain the prediction detection results; Calculate the credibility of the predicted detection results and determine whether to send an alarm signal based on the credibility; Perform detection quality evaluation on the characteristic factors of feature detection data, including: Comparing the first characteristic factor, the second characteristic factor, and the third characteristic factor of each feature detection data with the first standard characteristic factor, the second standard characteristic factor, and the third standard characteristic factor of the preset feature detection data of the corresponding detection type, respectively, to obtain a first characteristic factor difference, a second characteristic factor difference, and a third characteristic factor difference; Determine a corresponding detection quality evaluation library according to the detection type of each feature detection data, the detection quality evaluation library including a plurality of preset first feature factor difference values, preset second feature factor difference values, and preset third feature factor difference values ​​of the preset feature detection data, and each of the preset first feature factor difference values, the preset second feature factor difference values, and the preset third feature factor difference values ​​is associated with a specific preset detection quality evaluation value; Perform similarity analysis on the first characteristic factor difference, the second characteristic factor difference, and the third characteristic factor difference of each characteristic detection data with the preset first characteristic factor difference, the preset second characteristic factor difference, and the preset third characteristic factor difference in the corresponding detection quality evaluation library, respectively, and combine the preset detection quality evaluation values ​​of the preset first characteristic factor difference, the preset second characteristic factor difference, and the preset third characteristic factor difference with the greatest similarity with the weight coefficients of the first characteristic factor, the second characteristic factor, and the third characteristic factor to generate a detection quality evaluation result; The test quality evaluation results are: ； Among them, H is the detection quality evaluation result, a1 is the preset detection quality evaluation value of the preset first characteristic factor difference with the greatest similarity to the first characteristic factor difference, c1 is the weight coefficient of the first characteristic factor, a2 is the preset detection quality evaluation value of the preset second characteristic factor difference with the greatest similarity to the second characteristic factor difference, c2 is the weight coefficient of the second characteristic factor, a3 is the preset detection quality evaluation value of the preset third characteristic factor difference with the greatest similarity to the third characteristic factor difference, and c3 is the weight coefficient of the third characteristic factor.

2. The method for detecting a lightning arrester being charged according to claim 1, wherein: The preprocessing includes time unit normalization, outlier removal, missing value filling and data cleaning.

3. The method for detecting a lightning arrester being charged according to claim 2, wherein: The initial detection data is classified according to the detection type, and the characteristic detection data and characteristic factors of the characteristic detection data in the different historical detection logs of each detection type are extracted, including: Obtain the initial detection data of the same historical detection log, analyze the data source of the initial detection data, determine the sensor type of the initial detection data on the same historical detection day, and determine the detection type of the initial detection data in the corresponding historical detection log based on the sensor type; Preset feature detection data for each detection type; Analyze the initial detection data in different historical detection logs of the corresponding detection type according to the preset feature detection data to obtain feature detection data in different historical detection logs of the corresponding detection type; Preset a preset collection time interval for each detection type, obtain feature detection data from the same historical detection log of the corresponding detection type according to the preset collection time interval, and map it to the time reference line constructed by the historical detection duration in the corresponding historical detection log, and generate a feature detection data reference graph of multiple historical detection logs of the current detection type; The characteristic detection data reference graphs of multiple historical detection logs of each detection type are analyzed to determine the characteristic factors of the corresponding characteristic detection data.

4. The method for detecting a lightning arrester being charged according to claim 3, wherein: Determine the characteristic factors corresponding to the feature detection data, including: Obtaining the integrity of the feature detection data in the feature detection data reference graph, and when the integrity is greater than a preset integrity threshold, generating a feature detection data change curve for the corresponding feature detection data reference graph; Acquire multiple fluctuation values ​​of adjacent preset acquisition time intervals of each characteristic detection data change curve, generate a stability evaluation value based on the multiple fluctuation values, and set the stability evaluation value as a first characteristic factor of the corresponding characteristic detection data; Obtain the historical detection accuracy, historical detection cost, and historical detection timeliness of each feature detection data change curve, generate a detection evaluation value for the corresponding feature detection data change curve based on the historical detection accuracy, historical detection cost, and historical detection timeliness, and set the detection evaluation value as the second characteristic factor of the corresponding feature detection data; Obtain historical environmental information within the historical detection period of each feature detection data change curve, determine the environmental information interval in which the historical environmental information is located, and the mutation time node of the historical environmental information within the historical detection period; The mutation time node is marked in the corresponding feature detection data change curve, and the change value of the feature detection data at the corresponding mutation time node is obtained. An anti-interference evaluation value is generated according to the change values ​​at multiple mutation time nodes in the same feature detection data change curve, and the anti-interference evaluation value is set as the third characteristic factor of the corresponding feature detection data.

5. The method for detecting a charged lightning arrester according to claim 4, wherein: Before determining the optimal test type and the corresponding preferred test data based on the test quality evaluation results, the following steps are also required: When the integrity is less than a preset integrity threshold, the corresponding feature detection data reference image is removed, and the integrity difference of the removed feature detection data reference image is calculated; Obtain the number of rejections of the feature detection data reference image in each detection type and the corresponding integrity difference to generate a compensation coefficient for the corresponding detection type; The calculation formula of the compensation coefficient is: ； Where r is the compensation coefficient, z is the compensation conversion coefficient, n is the total number of all feature detection data reference maps in the detection type, m is the feature detection data reference map that is eliminated, W0 is the preset integrity threshold, and W1i is the integrity of the i-th eliminated feature detection data reference map; The detection quality evaluation result of the characteristic detection data of the corresponding detection type is corrected according to the compensation coefficient, and the corrected detection quality evaluation result is H*r.

6. The method for detecting a charged lightning arrester according to claim 5, wherein: Determine the optimal test type and corresponding preferred test data based on the test quality evaluation results, including: Generate a comprehensive test quality evaluation result of the corresponding test type based on multiple revised test quality evaluation results of the same test type; Ranking the comprehensive test quality evaluation results of multiple test types, and setting the test type ranked first as the optimal test type; Compare the multiple corrected detection quality evaluation results of the optimal detection type with the preset detection quality evaluation result threshold, and set the feature detection data corresponding to the detection quality evaluation results greater than the preset detection quality evaluation result threshold as the preferred detection data.

7. The method for detecting a charged lightning arrester according to claim 6, wherein: Input into the prediction detection model to obtain the prediction detection results, including: Generate a training data set based on the preferred detection data and the corresponding historical detection results, train a neural network model based on the training data set, and obtain a predictive detection model; The detection tester involved in the optimal detection type obtains real-time feature detection data of the lightning arrester, and the real-time feature detection data is input into the prediction detection model to obtain the prediction detection result.

8. The method for detecting a charged lightning arrester according to claim 7, wherein: Calculate the confidence level of the predicted test results, including: Get the historical test results in the previous historical period before the current predicted test result; Determining adjacent historical states and historical change characteristics of adjacent historical states within the previous historical period based on adjacent historical detection results within the previous historical period, wherein the historical change characteristics include a historical state change trend, a historical state fluctuation degree, and a historical state change rate; Generate a first state evaluation value according to a historical state change trend; Generate a second state evaluation value according to the historical state fluctuation degree; generating a third state evaluation value according to the historical state change rate; generating a plurality of comprehensive state evaluation values ​​of adjacent historical detection results according to the first state evaluation value, the second state evaluation value, and the third state evaluation value; Setting a normal variation interval of the comprehensive state evaluation value according to multiple comprehensive state evaluation values ​​of adjacent historical detection results; Generate a predicted comprehensive state evaluation value based on the predicted detection result and the previous adjacent historical detection result; Determine whether the predicted comprehensive state evaluation value is within the normal variation range of the comprehensive state evaluation value. If so, the credibility is greater than the preset credibility threshold; if not, the credibility is less than the preset credibility threshold.

9. The method for detecting a lightning arrester being charged according to claim 8, wherein: Determine whether to send an alarm signal based on the credibility, including: When the credibility is greater than the preset credibility threshold, the real-time status of the arrester is generated according to the predicted detection results; When the real-time status is a fault state, an alarm signal is sent; when the real-time status is a normal state, an early warning signal is sent and the next preset detection time interval is set; when the real-time status is a good state, no signal is sent; When the credibility is less than the preset credibility threshold, the prediction detection model is rebuilt and the prediction detection result is reacquired until the credibility is greater than the preset credibility threshold.

Images