A method and device for evaluating the insulation strength of combined gaps in live working

Abstract

The present invention discloses a method and device for evaluating the insulation strength of a combination gap for live working, the method comprising: constructing a finite element model for a preset data set; the preset data set comprises a preset number of combination gaps of total gap lengths, wherein the combination gaps that meet a first condition are classified into a first subset, otherwise they are classified into a second subset; the first condition comprises that the first gap length in the combination gap is greater than a first gap length critical value corresponding to 50% of the breakdown voltage; respectively obtaining parameter sets of the first subset and the second subset to obtain a first parameter set and a second parameter set; respectively training an LSSVR model according to the first parameter set and the second parameter set to obtain a first model and a second model; obtaining an actual value of the first gap length and an actual value of the parameter set of the target combination gap; if the actual value of the first gap length meets the first condition, performing insulation strength evaluation according to the first model and the actual value of the parameter set, otherwise performing insulation strength evaluation according to the second model and the actual value of the parameter set.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation and maintenance, and in particular to a method and device for evaluating the insulation strength of a combined gap in live working. Background Art

[0002] In modern power system operation and maintenance, live working technology has become a key measure to ensure continuous and reliable operation. Live working allows workers to perform maintenance and repairs on high-voltage transmission lines without shutting down power, which is crucial for reducing outages and improving power supply reliability and economic benefits. With the continuous expansion of power grids and the growing demand for electricity, the importance of live working has become increasingly prominent.

[0003] The combined gap for live working is a key factor in live working, and the accurate assessment of its insulation strength is crucial to ensuring the safety of workers and the reliable operation of power systems. The combined gap for live working refers to the combined air gap formed by workers and live conductors, towers, etc. during equipotential live working. When workers enter and exit the equipotential area, the insulation strength of the combined gap is directly related to the safety of the work. The breakdown voltage of the combined gap reflects its insulation strength. In actual work, only by accurately assessing the breakdown voltage and calculating the risk rate of live working can we ensure that workers maintain a sufficient safe distance from live objects and avoid the occurrence of discharge accidents. However, the traditional method for assessing the breakdown voltage of the combined gap is mainly based on the safety distance and the relevant empirical formula. This method is difficult to generalize to situations with different geometric structures and electric field distributions, and cannot adapt well to various combined gap types, making it difficult to achieve accurate assessment of the breakdown voltage of various combined gaps.

[0004] Therefore, a new technical solution is urgently needed to solve the technical problem of how to accurately and efficiently evaluate the insulation strength of the combined gap of live working. Summary of the Invention

[0005] The present invention provides a method and device for evaluating the insulation strength of a live working combination gap, so as to solve the technical problem of how to accurately and efficiently evaluate the insulation strength of a live working combination gap.

[0006] To achieve the above object, the present invention provides a method for evaluating the insulation strength of a combined gap during live working, comprising:

[0007] A finite element model is constructed for a preset data set; the preset data set includes a preset number of combined gaps with total gap lengths, wherein the combined gaps that meet a first condition are classified into a first subset, and the combined gaps that do not meet the first condition are classified into a second subset; the first condition includes that the first gap length in the combined gap is greater than a first gap length critical value corresponding to 50% of the breakdown voltage.

[0008] The finite element model includes a high-voltage electrode, a ground electrode, and a suspended conductor electrode; the suspended conductor electrode is suspended between the high-voltage electrode and the ground electrode; the gap between the high-voltage electrode and the ground electrode is defined as the total gap, the gap between the suspended conductor electrode and the high-voltage electrode is defined as the first gap, and the gap between the suspended conductor electrode and the ground electrode is defined as the second gap.

[0009] The parameter sets of the first subset and the second subset are respectively obtained to obtain the first parameter set and the second parameter set; the parameter sets include the preset electric field strength feature quantities of the first gap and the second gap in the combined gap; and the LSSVR model is trained according to the first parameter set and the second parameter set to obtain the first model and the second model.

[0010] Obtain the first gap length actual value and parameter set actual value of the target combination gap; if the first gap length actual value meets the first condition, perform insulation strength evaluation according to the first model and parameter set actual value, otherwise perform insulation strength evaluation according to the second model and parameter set actual value.

[0011] Preferably, respectively obtaining the parameter sets of the first subset and the second subset, to obtain the first parameter set and the second parameter set comprises:

[0012] The first subset and the second subset are respectively subjected to first processing to obtain a first parameter set and a second parameter set.

[0013] The first processing includes: selecting a preset number of sampling points at equal intervals in each first gap and second gap, and extracting the electric field strength of the sampling points through finite element simulation; obtaining preset electric field strength characteristic quantities of the sampling points of the first gap and the second gap based on the electric field strength of the sampling points of the first gap and the second gap, respectively; obtaining a parameter set based on the preset electric field strength characteristic quantities of the sampling points of the first gap and the second gap; and performing standardization processing on the parameter sets of the first gap and the second gap to obtain the first parameter set and the second parameter set.

[0014] The preset electric field strength characteristic quantities include the maximum electric field strength, the minimum electric field strength, the average electric field strength, the median electric field strength, the sum of squares of the electric field strength, the average of the sum of squares of the electric field strength, the standard deviation of the electric field strength, the ratio of the sum of squares of the electric field strength on the line segment where the electric field strength is greater than 90% of the maximum electric field strength to the sum of squares of the electric field strength, the electric field strength uniformity coefficient, the electric field strength distortion rate, the electric field strength variation coefficient, the maximum electric field gradient, the minimum electric field gradient, the average electric field gradient and the median electric field gradient.

[0015] Preferably, obtaining the preset electric field intensity characteristic quantities of the sampling points of the first gap and the second gap according to the electric field intensity of the sampling points of the first gap and the second gap respectively includes:

[0016] Performing a second process on the first gap and the second gap respectively to obtain a preset electric field intensity characteristic value of the sampling points of the first gap and the second gap; the second process includes:

[0017] Calculate the maximum electric field strength , minimum electric field strength , average electric field strength and the median electric field strength ,include:

[0018] ;

[0019] ;

[0020] ;

[0021] ;

[0022] in, Indicates the sampling points; Indicates the The electric field strength at each sampling point; Indicates the total number of sampling points in the first gap or the second gap.

[0023] Calculate the sum of squares of the electric field strength , the average value of the sum of squares of the electric field strength , standard deviation of electric field strength And the sum of the squares of the electric field strength on the line segment where the electric field strength is greater than 90% of the maximum electric field strength and the sum of the squares of the electric field strength Ratio ,include:

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] Calculation of electric field strength non-uniformity coefficient and , electric field intensity distortion rate and the coefficient of variation of electric field strength ,include:

[0029] ;

[0030] ;

[0031] ;

[0032] ;

[0033] Calculate the maximum value of the electric field gradient , minimum electric field gradient , average electric field gradient and the median electric field gradient ,include:

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] in, Indicates the The electric field gradient at each sampling point.

[0039] Preferably, performing the normalization process according to the parameter sets of the first gap and the second gap includes:

[0040] Standardization processing includes using z-score standardization to process data so that the data conforms to a standard normal distribution with a mean of 0 and a standard deviation of 1, including:

[0041] ;

[0042] in, represents the original data points; and represent the mean and standard deviation of the features respectively; Represents the normalized data points.

[0043] Preferably, training the LSSVR model according to the first parameter set and the second parameter set respectively to obtain the first model and the second model includes:

[0044] The kernel function parameters and regularization parameters of the LSSVR model are optimized using a six-fold cross validation method according to the first parameter set and the second parameter set, respectively, to obtain a first intermediate model and a second intermediate model.

[0045] Model training is performed according to the first intermediate model in combination with the first parameter set, and verified by a preset validation set to obtain a first model; model training is performed according to the second intermediate model in combination with the second parameter set, and verified by a preset validation set to obtain a second model.

[0046] Preferably, the kernel function parameters and regularization parameters of the LSSVR model are optimized using a six-fold cross validation method according to the first parameter set and the second parameter set, respectively, to obtain the first intermediate model and the second intermediate model, including:

[0047] The value ranges of the kernel function parameters and the regularization parameters are set to a preset parameter value set; the preset parameter value set includes a preset number of parameter values; all value combinations of the kernel function parameters and the regularization parameters are exhaustively enumerated based on the preset parameter value set, and a parameter optimization training set is obtained based on all value combinations.

[0048] The first parameter set and the second parameter set are randomly divided into 6 parameter subsets respectively; the LSSVR model is initialized according to the combination of kernel function parameters and regularization parameter values ​​in the parameter optimization training set.

[0049] According to the parameter subset of the first parameter set, the LSSVR model with each combination of kernel function parameters and regularization parameter values ​​is iteratively optimized for 6 times, in which 5 parameter subsets are selected for training in each iteration, 1 parameter subset is retained for verification, and the mean absolute error (MAE) of the verification results is calculated; the mean absolute error (MAE) in all iterative optimizations of the LSSVR model with all combinations of kernel function parameters and regularization parameter values ​​is statistically calculated, and the kernel function parameter and regularization parameter value combination corresponding to the minimum mean absolute error (MAE) is selected as the model parameter to obtain the first intermediate model.

[0050] According to the parameter subset of the second parameter set, the LSSVR model with each combination of kernel function parameters and regularization parameter values ​​is iteratively optimized for 6 times. In each iteration, 5 parameter subsets are selected for training, 1 parameter subset is retained for verification, and the mean absolute error (MAE) of the verification results is calculated. The mean absolute error (MAE) in all iterative optimizations of the LSSVR model with all combinations of kernel function parameters and regularization parameter values ​​is statistically calculated, and the kernel function parameter and regularization parameter value combination corresponding to the minimum mean absolute error (MAE) is selected as the model parameters to obtain the second intermediate model.

[0051] Calculating the mean absolute error (MAE) of the validation results involves:

[0052] ;

[0053] in, Indicates actual value; represents the predicted value; Indicates the sample size.

[0054] Preferably, performing model training according to the first intermediate model in combination with the first parameter set and verifying it through a preset validation set to obtain the first model; performing model training according to the second intermediate model in combination with the second parameter set and verifying it through a preset validation set to obtain the second model includes:

[0055] The optimization problems of the LSSVR model include:

[0056] ;

[0057] Constraint statistics include:

[0058] ;

[0059] in, represents the weight vector; represents the bias term; Represents a mapping function that maps input data to a high-dimensional feature space; represents the error term, represents the regularization parameter used to control the complexity of the model; Indicates transpose.

[0060] Via Lagrange multipliers With the introduction of , the optimization problem of the LSSVR model is transformed into a dual problem:

[0061] ;

[0062] in, represents the kernel function; Represents the sample point to be predicted; Indicates the first sample points;

[0063] The kernel function matrix is ​​constructed based on the Gaussian kernel function, which includes the Gaussian kernel function values ​​between all training samples. , is the kernel width parameter; substitute the kernel function matrix into the LSSVR model to obtain a linear equation system; solve the linear equation system to obtain The trained model is obtained by taking the values ​​of a and b. The trained model is verified based on the preset validation set, and the training effect of the model is determined by calculating the mean absolute error (MAE) and the coefficient of determination (R²) on the preset validation set. The model training is completed when the required training effect is achieved.

[0064] Preferably, performing the insulation strength assessment includes:

[0065] After the actual value of the parameter set is input into the first model or the second model, the model outputs the predicted value of the 50% breakdown voltage of the target combined gap; the maximum operating overvoltage of the system is obtained based on the operating status and historical data of the power system.

[0066] The hazard rate of live working of the target combination gap is obtained based on the predicted value of 50% breakdown voltage of the target combination gap and the maximum operating overvoltage of the system. The safety of live working is evaluated based on the hazard rate, and the insulation strength evaluation is completed.

[0067] Preferably, the hazard rate of live working of the target combination gap is obtained based on the predicted value of the 50% breakdown voltage of the target combination gap and the maximum operating overvoltage of the system. The safety of live working is evaluated based on the hazard rate, including:

[0068] Assuming that the probability distribution of system operation overvoltage and the probability distribution of air gap breakdown both obey normal distribution, the hazard rates according to probability statistics include:

[0069] ;

[0070] ;

[0071] ;

[0072] ;

[0073] in, represents the hazard rate; The probability density function representing the overvoltage amplitude; represents the probability distribution function of air gap breakdown under switching overvoltage with amplitude U; Indicates the standard deviation of switching overvoltage; Indicates the predicted value of 50% breakdown voltage of target combination gap; Indicates the standard deviation of the gap discharge voltage during live working; Indicates the average value of switching overvoltage; Indicates the relative standard deviation of overvoltage; Indicates the system's maximum operating overvoltage.

[0074] If the hazard rate is less than the preset hazard threshold, the live working is determined to be safe; otherwise, the live working is determined to be unsafe.

[0075] The present invention also provides an insulation strength assessment device for a live working combined gap, which is used in the method of the present invention. The device comprises a first module, a second module and a third module.

[0076] The first module is used to construct a finite element model for a preset data set; the preset data set includes a preset number of combined gaps with total gap lengths, where the combined gaps that meet the first condition are classified into a first subset, and the combined gaps that do not meet the first condition are classified into a second subset; the first condition includes that the first gap length in the combined gap is greater than the first gap length critical value corresponding to 50% of the breakdown voltage.

[0077] The finite element model includes a high-voltage electrode, a ground electrode, and a suspended conductor electrode; the suspended conductor electrode is suspended between the high-voltage electrode and the ground electrode; the gap between the high-voltage electrode and the ground electrode is defined as the total gap, the gap between the suspended conductor electrode and the high-voltage electrode is defined as the first gap, and the gap between the suspended conductor electrode and the ground electrode is defined as the second gap.

[0078] The second module is used to obtain the parameter sets of the first subset and the second subset respectively, to obtain the first parameter set and the second parameter set; the parameter set includes the preset electric field strength characteristic quantities of the first gap and the second gap in the combined gap; the LSSVR model is trained according to the first parameter set and the second parameter set respectively, to obtain the first model and the second model.

[0079] The third module is used to obtain the actual value of the first gap length and the actual value of the parameter set of the target combination gap; if the actual value of the first gap length meets the first condition, the insulation strength evaluation is performed according to the first model and the actual value of the parameter set; otherwise, the insulation strength evaluation is performed according to the second model and the actual value of the parameter set.

[0080] The present invention has the following beneficial effects:

[0081] The insulation strength assessment method for live working combination gaps of the present invention can perform safety assessment by constructing a finite element model for the combination gap to obtain electric field characteristics, so that the method of the present invention has good assessment efficiency and is convenient for practical application. The LSSVR model is trained based on training data of two cases to obtain a first model and a second model, so that the method of the present invention can better capture the complex patterns and nonlinear relationships in the corresponding data. The first model or the second model is dynamically selected for assessment according to the actual situation of the target combination gap, so that the method of the present invention can cover the combination gaps of each case, avoid the performance degradation problem that may occur in a single model when processing diversified data, and improve the accuracy of the prediction assessment, that is, the method of the present invention is flexible, targeted and versatile. The method of the present invention can accurately and efficiently assess the insulation strength of live working combination gaps.

[0082] The device for evaluating the insulation strength of a combined gap during live working according to the present invention is used in the method according to the present invention and has the same beneficial effects as the method according to the present invention.

[0083] In addition to the above-described objects, features and advantages, the present invention has other objects, features and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The accompanying drawings, which constitute part of the present invention, are provided to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are provided to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings:

[0085] Figure 1 It is a schematic diagram of a method flow of a preferred embodiment of the present invention.

[0086] Figure 2 Schematic diagram of the combined gap of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0087] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. However, the present invention can be implemented in many different ways as defined and covered by the claims.

[0088] In a preferred embodiment of the present invention, high-voltage electrodes are used to simplify the simulation of high-voltage conductors in live working, and suspended conductor electrodes are used to simplify the simulation of live workers at suspended potential during live working. The gap between the high-voltage electrodes and the suspended conductor electrodes is used to simulate the potential transfer gap between the high-voltage conductors and the live workers during live working. Figure 1 In a preferred embodiment of the present invention, a method for evaluating the insulation strength of a combined gap during live working is provided, comprising:

[0089] S1. Construct a finite element model for a preset data set; the preset data set includes a preset number of combined gaps of total gap lengths, wherein the combined gaps satisfying a first condition are classified into a first subset, and the combined gaps not satisfying the first condition are classified into a second subset; the first condition includes a first gap length within the combined gaps being greater than a first gap length critical value corresponding to 50% breakdown voltage. The first gap length critical value corresponding to 50% breakdown voltage is obtained based on historical data.

[0090] The finite element model includes a high-voltage electrode, a ground electrode, and a suspended conductor electrode; the suspended conductor electrode is suspended between the high-voltage electrode and the ground electrode; the gap between the high-voltage electrode and the ground electrode is defined as the total gap, the gap between the suspended conductor electrode and the high-voltage electrode is defined as the first gap, and the gap between the suspended conductor electrode and the ground electrode is defined as the second gap.

[0091] S2. Obtain parameter sets for the first subset and the second subset, respectively, to obtain first parameter sets and second parameter sets; the parameter sets include preset electric field strength characteristics of the first gap and the second gap within the combined gap; and train LSSVR models based on the first parameter set and the second parameter set, respectively, to obtain first models and second models. The LSSVR model stands for Least Squares Support Vector Regression model.

[0092] In a preferred embodiment of the present invention, respectively obtaining parameter sets of the first subset and the second subset, and obtaining the first parameter set and the second parameter set includes:

[0093] The first subset and the second subset are respectively subjected to first processing to obtain a first parameter set and a second parameter set.

[0094] The first processing includes: selecting a preset number of sampling points at equal intervals in each first gap and second gap, and extracting the electric field strength of the sampling points through finite element simulation; obtaining preset electric field strength characteristic quantities of the sampling points of the first gap and the second gap based on the electric field strength of the sampling points of the first gap and the second gap, respectively; obtaining a parameter set based on the preset electric field strength characteristic quantities of the sampling points of the first gap and the second gap; and performing standardization processing on the parameter sets of the first gap and the second gap to obtain the first parameter set and the second parameter set.

[0095] The preset electric field strength characteristic quantities include the maximum electric field strength, the minimum electric field strength, the average electric field strength, the median electric field strength, the sum of squares of the electric field strength, the average of the sum of squares of the electric field strength, the standard deviation of the electric field strength, the ratio of the sum of squares of the electric field strength on the line segment where the electric field strength is greater than 90% of the maximum electric field strength to the sum of squares of the electric field strength, the electric field strength uniformity coefficient, the electric field strength distortion rate, the electric field strength variation coefficient, the maximum electric field gradient, the minimum electric field gradient, the average electric field gradient and the median electric field gradient.

[0096] In a preferred embodiment of the present invention, obtaining the preset electric field intensity characteristic quantities of the sampling points of the first gap and the second gap according to the electric field intensity of the sampling points of the first gap and the second gap respectively includes:

[0097] Performing a second process on the first gap and the second gap respectively to obtain a preset electric field intensity characteristic value of the sampling points of the first gap and the second gap; the second process includes:

[0098] Calculate the maximum electric field strength , minimum electric field strength , average electric field strength and the median electric field strength ,include:

[0099] ;

[0100] ;

[0101] ;

[0102] ;

[0103] in, Indicates the sampling points; Indicates the The electric field strength at each sampling point; Indicates the total number of sampling points in the first gap or the second gap.

[0104] Calculate the sum of squares of the electric field strength , the average value of the sum of squares of the electric field strength , standard deviation of electric field strength And the sum of the squares of the electric field strength on the line segment where the electric field strength is greater than 90% of the maximum electric field strength and the sum of the squares of the electric field strength Ratio ,include:

[0105] ;

[0106] ;

[0107] ;

[0108] ;

[0109] Calculation of electric field strength non-uniformity coefficient and , electric field intensity distortion rate and the coefficient of variation of electric field strength ,include:

[0110] ;

[0111] ;

[0112] ;

[0113] ;

[0114] Calculate the maximum value of the electric field gradient , minimum electric field gradient , average electric field gradient and the median electric field gradient ,include:

[0115] ;

[0116] ;

[0117] ;

[0118] ;

[0119] in, Indicates the The electric field gradient at each sampling point.

[0120] In a preferred embodiment of the present invention, performing normalization processing according to the parameter sets of the first gap and the second gap includes:

[0121] Standardization processing includes using z-score standardization to process data so that the data conforms to a standard normal distribution with a mean of 0 and a standard deviation of 1, including:

[0122] ;

[0123] in, represents the original data points; and represent the mean and standard deviation of the features respectively; Represents the normalized data points.

[0124] In a preferred embodiment of the present invention, the LSSVR model is trained according to the first parameter set and the second parameter set, respectively, to obtain the first model and the second model, including:

[0125] The kernel function parameters and regularization parameters of the LSSVR model are optimized using a six-fold cross validation method according to the first parameter set and the second parameter set, respectively, to obtain a first intermediate model and a second intermediate model.

[0126] The kernel function parameter controls the width of the Gaussian kernel function, affecting the model's ability to capture local features of the data; the regularization parameter is used to balance the model's fit and complexity. In a preferred embodiment of the present invention, to fully explore the parameter space, the kernel function parameter and the regularization parameter are both set to the range of [0.001, 0.01, 0.1, 1, 10, 100, 1000]. This range covers multiple orders of magnitude, from extremely small to extremely large, and aims to find the parameter combination that optimizes the model's performance on the current dataset.

[0127] Model training is performed according to the first intermediate model in combination with the first parameter set, and verified by a preset validation set to obtain a first model; model training is performed according to the second intermediate model in combination with the second parameter set, and verified by a preset validation set to obtain a second model.

[0128] In a preferred embodiment of the present invention, the kernel function parameters and regularization parameters of the LSSVR model are optimized using a six-fold cross validation method according to the first parameter set and the second parameter set, respectively, to obtain the first intermediate model and the second intermediate model including:

[0129] The value ranges of the kernel function parameters and the regularization parameters are set to a preset parameter value set; the preset parameter value set includes a preset number of parameter values; all value combinations of the kernel function parameters and the regularization parameters are exhaustively enumerated based on the preset parameter value set, and a parameter optimization training set is obtained based on all value combinations.

[0130] The first parameter set and the second parameter set are randomly divided into 6 parameter subsets respectively; the LSSVR model is initialized according to the combination of kernel function parameters and regularization parameter values ​​in the parameter optimization training set.

[0131] According to the parameter subset of the first parameter set, the LSSVR model with each combination of kernel function parameters and regularization parameter values ​​is iteratively optimized for 6 times, in which 5 parameter subsets are selected for training in each iteration, 1 parameter subset is retained for verification, and the mean absolute error (MAE) of the verification results is calculated; the mean absolute error (MAE) in all iterative optimizations of the LSSVR model with all combinations of kernel function parameters and regularization parameter values ​​is statistically calculated, and the kernel function parameter and regularization parameter value combination corresponding to the minimum mean absolute error (MAE) is selected as the model parameter to obtain the first intermediate model.

[0132] According to the parameter subset of the second parameter set, the LSSVR model with each combination of kernel function parameters and regularization parameter values ​​is iteratively optimized for 6 times. In each iteration, 5 parameter subsets are selected for training, 1 parameter subset is retained for verification, and the mean absolute error (MAE) of the verification results is calculated. The mean absolute error (MAE) in all iterative optimizations of the LSSVR model with all combinations of kernel function parameters and regularization parameter values ​​is statistically calculated, and the kernel function parameter and regularization parameter value combination corresponding to the minimum mean absolute error (MAE) is selected as the model parameters to obtain the second intermediate model.

[0133] Calculating the mean absolute error (MAE) of the validation results involves:

[0134] ;

[0135] in, Indicates actual value; represents the predicted value; Indicates the sample size.

[0136] In a preferred embodiment of the present invention, the kernel function parameters and regularization parameters of the LSSVR model are optimized using a six-fold cross-validation method, which can effectively prevent the model from overfitting, thereby significantly improving the generalization performance of the model when facing new data, making it more applicable and reliable.

[0137] In a preferred embodiment of the present invention, model training is performed according to the first intermediate model in combination with the first parameter set, and the model is verified by a preset validation set to obtain the first model; model training is performed according to the second intermediate model in combination with the second parameter set, and the model is verified by a preset validation set to obtain the second model, including:

[0138] The optimization problems of the LSSVR model include:

[0139] ;

[0140] Constraint statistics include:

[0141] ;

[0142] in, represents the weight vector; represents the bias term; Represents a mapping function that maps input data to a high-dimensional feature space; represents the error term, represents the regularization parameter used to control the complexity of the model; Indicates transpose.

[0143] Via Lagrange multipliers With the introduction of , the optimization problem of the LSSVR model is transformed into a dual problem:

[0144] ;

[0145] in, represents the kernel function; Represents the sample point to be predicted; Indicates the first sample points.

[0146] The kernel function matrix is ​​constructed based on the Gaussian kernel function, which includes the Gaussian kernel function values ​​between all training samples. , is the kernel width parameter; substitute the kernel function matrix into the LSSVR model to obtain a linear equation system; solve the linear equation system to obtain The trained model is obtained by adjusting the values ​​of a and b. The trained model is then validated using a pre-set validation set, and the training effect of the model is determined by calculating the mean absolute error (MAE) and the coefficient of determination (R²) on the pre-set validation set. Model training is completed when the required training effect is achieved. The smaller the MAE and the closer the R² is to 1, the higher the model's predictive accuracy and better explanatory power.

[0147] S3. Obtain the actual value of the first gap length and the actual value of the parameter set of the target combination gap; if the actual value of the first gap length meets the first condition, perform insulation strength evaluation according to the first model and the actual value of the parameter set; otherwise, perform insulation strength evaluation according to the second model and the actual value of the parameter set.

[0148] A finite element model is established based on the target combined gap to obtain the actual value of the first gap length and the actual value of the parameter set of the target combined gap. In a preferred embodiment of the present invention, performing the insulation strength evaluation includes:

[0149] After the actual value of the parameter set is input into the first model or the second model, the model outputs the predicted value of the 50% breakdown voltage of the target combined gap; the maximum operating overvoltage of the system is obtained based on the operating status and historical data of the power system;

[0150] The hazard rate of live working of the target combination gap is obtained based on the predicted value of 50% breakdown voltage of the target combination gap and the maximum operating overvoltage of the system. The safety of live working is evaluated based on the hazard rate, and the insulation strength evaluation is completed.

[0151] In a preferred embodiment of the present invention, the hazard rate of live working at the target combination gap is obtained based on the predicted value of the 50% breakdown voltage of the target combination gap and the maximum operating overvoltage of the system. The safety of live working is evaluated based on the hazard rate, including:

[0152] Assuming that the probability distribution of system operation overvoltage and the probability distribution of air gap breakdown both obey normal distribution, the hazard rates according to probability statistics include:

[0153] ;

[0154] ;

[0155] ;

[0156] ;

[0157] in, represents the hazard rate; The probability density function representing the overvoltage amplitude; represents the probability distribution function of air gap breakdown under switching overvoltage with amplitude U; Indicates the standard deviation of the switching overvoltage (kV), which is generally taken as 12%~15% of the switching overvoltage; Indicates the predicted value of 50% breakdown voltage of target combination gap (kV); Indicates the standard deviation of the gap discharge voltage during live working (kV), which is usually taken as 6%; Indicates the average value of switching overvoltage (kV); Indicates the relative standard deviation of overvoltage, generally 12%~15%; Indicates the system's maximum operating overvoltage.

[0158] If the risk rate is less than the preset risk threshold, then the live working is determined to be safe, otherwise it is determined that the live working is unsafe. In a preferred embodiment of the present invention, the preset risk threshold includes .

[0159] The insulation strength assessment method for live working combination gaps of the present invention can perform safety assessment by constructing a finite element model for the combination gap to obtain electric field characteristics, so that the method of the present invention has good assessment efficiency and is convenient for practical application. The LSSVR model is trained based on training data of two cases to obtain a first model and a second model, so that the method of the present invention can better capture the complex patterns and nonlinear relationships in the corresponding data. The first model or the second model is dynamically selected for assessment according to the actual situation of the target combination gap, so that the method of the present invention can cover the combination gaps of each case, avoid the performance degradation problem that may occur in a single model when processing diversified data, and improve the accuracy of the prediction assessment, that is, the method of the present invention is flexible, targeted and versatile. The method of the present invention can accurately and efficiently assess the insulation strength of live working combination gaps.

[0160] In a preferred embodiment of the present invention, there is also provided an insulation strength assessment device for a combined gap during live working, which is used in the method of the present invention. The device includes a first module, a second module and a third module.

[0161] The first module is used to construct a finite element model for a preset data set; the preset data set includes a preset number of combined gaps with total gap lengths, where the combined gaps that meet the first condition are classified into a first subset, and the combined gaps that do not meet the first condition are classified into a second subset; the first condition includes that the first gap length in the combined gap is greater than the first gap length critical value corresponding to 50% of the breakdown voltage.

[0162] The finite element model includes a high-voltage electrode, a ground electrode, and a suspended conductor electrode; the suspended conductor electrode is suspended between the high-voltage electrode and the ground electrode; the gap between the high-voltage electrode and the ground electrode is defined as the total gap, the gap between the suspended conductor electrode and the high-voltage electrode is defined as the first gap, and the gap between the suspended conductor electrode and the ground electrode is defined as the second gap.

[0163] The second module is used to obtain the parameter sets of the first subset and the second subset respectively, to obtain the first parameter set and the second parameter set; the parameter set includes the preset electric field strength characteristic quantities of the first gap and the second gap in the combined gap; the LSSVR model is trained according to the first parameter set and the second parameter set respectively, to obtain the first model and the second model.

[0164] The third module is used to obtain the actual value of the first gap length and the actual value of the parameter set of the target combination gap; if the actual value of the first gap length meets the first condition, the insulation strength evaluation is performed according to the first model and the actual value of the parameter set; otherwise, the insulation strength evaluation is performed according to the second model and the actual value of the parameter set.

[0165] The device for evaluating the insulation strength of a combined gap during live working according to the present invention is used in the method according to the present invention and has the same beneficial effects as the method according to the present invention.

[0166] Verification part:

[0167] See also Figure 2 According to the on-site working conditions of live working on a 220kV transmission line, a live working combined gap model was constructed in COMSOL Multiphysics, in which the total gap length is 2.0m, the length of sub-gap 1, i.e. the first gap, is 0.7m, the length of sub-gap 2, i.e. the second gap, is 1.1m, and the length of the suspended conductor electrode is 0.5m.

[0168] Eleven sampling points were set in each of the first and second gaps. The electric field strength values ​​of each sampling point were obtained through finite element simulation, and the electric field characteristic values ​​of each sampling point were calculated accordingly. According to historical data statistics, for a total gap length of 2.0m, the critical value of the first gap length corresponding to a 50% breakdown voltage is 0.6m. Since the length of the first gap in the current configuration is 0.7m, the first model is selected. The input characteristic values ​​are normalized by z-score, and then the normalized characteristic vector is input into the first model to obtain the prediction result: the 50% breakdown voltage is 578.6kV. Therefore, the discharge voltage standard deviation of the combined gap parameters for live working is = 578.6 × 0.06 = 34.7 kV. Based on the historical overvoltage level information obtained for this transmission line, the system's maximum operating overvoltage is 420 kV. Assuming the relative standard deviation of the overvoltage is 13%, the average operating overvoltage is: =420 / (1+2.05×0.13) = 331.6kV, switching overvoltage standard deviation = 331.6×0.13 = 43.1 kV. The hazard rate of live working is calculated according to the formula and obtained by numerical integration. Since the calculated risk rate of live working is , meeting safety standards. Assessment conclusion: Live working on this 220kV transmission line is safe under the current configuration and can be carried out.

[0169] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for evaluating the insulation strength of a combined gap during live working, characterized in that: include: Construct finite element models for preset data sets; The preset data set includes a preset number of combined gaps of total gap lengths, wherein the combined gaps satisfying a first condition are classified into a first subset, and the combined gaps not satisfying the first condition are classified into a second subset; the first condition includes that a first gap length in the combined gaps is greater than a first gap length critical value corresponding to 50% of the breakdown voltage; The finite element model includes a high-voltage electrode, a ground electrode, and a suspended conductor electrode; the suspended conductor electrode is suspended between the high-voltage electrode and the ground electrode; the gap between the high-voltage electrode and the ground electrode is defined as the total gap, the gap between the suspended conductor electrode and the high-voltage electrode is defined as the first gap, and the gap between the suspended conductor electrode and the ground electrode is defined as the second gap; Acquiring parameter sets of the first subset and the second subset respectively to obtain a first parameter set and a second parameter set; the parameter sets include preset electric field strength characteristic quantities of the first gap and the second gap in the combined gap; Train the LSSVR model according to the first parameter set and the second parameter set respectively to obtain a first model and a second model; Obtain the first gap length actual value and parameter set actual value of the target combination gap; if the first gap length actual value meets the first condition, perform insulation strength evaluation according to the first model and the actual value of the parameter set, otherwise perform insulation strength evaluation according to the second model and the actual value of the parameter set.

2. The method for evaluating the insulation strength of a live working gap according to claim 1, wherein: Obtaining parameter sets of the first subset and the second subset respectively, to obtain the first parameter set and the second parameter set includes: performing first processing on the first subset and the second subset respectively to obtain the first parameter set and the second parameter set; The first processing includes: selecting a preset number of sampling points at equal intervals in each of the first gap and the second gap, and extracting electric field strengths at the sampling points through finite element simulation; obtaining preset electric field strength feature quantities of the sampling points of the first gap and the second gap based on the electric field strengths of the sampling points of the first gap and the second gap, respectively; obtaining the parameter set based on the preset electric field strength feature quantities of the sampling points of the first gap and the second gap; and performing standardization processing on the parameter sets of the first gap and the second gap to obtain the first parameter set and the second parameter set. The preset electric field strength characteristic quantities include the maximum electric field strength, the minimum electric field strength, the average electric field strength, the median electric field strength, the sum of squares of the electric field strength, the average of the sum of squares of the electric field strength, the standard deviation of the electric field strength, the ratio of the sum of squares of the electric field strength on the line segment where the electric field strength is greater than 90% of the maximum electric field strength to the sum of squares of the electric field strength, the electric field strength uniformity coefficient, the electric field strength distortion rate, the electric field strength variation coefficient, the maximum electric field gradient, the minimum electric field gradient, the average electric field gradient and the median electric field gradient.

3. The method for evaluating the insulation strength of a live working gap according to claim 2, wherein: The step of obtaining the preset electric field intensity characteristic quantities of the sampling points of the first gap and the second gap according to the electric field intensity of the sampling points of the first gap and the second gap respectively comprises: Performing a second processing on the first gap and the second gap respectively to obtain preset electric field intensity characteristic quantities of the sampling points of the first gap and the second gap; the second processing includes: Calculate the maximum electric field strength , minimum electric field strength , average electric field strength and the median electric field strength ,include: ； ； ； ； in, Indicates the sampling points; Indicates the The electric field strength at each sampling point; Indicates the total number of sampling points in the first gap or the second gap; Calculate the sum of squares of the electric field strength , the average value of the sum of squares of the electric field strength , standard deviation of electric field strength And the sum of the squares of the electric field strength on the line segment where the electric field strength is greater than 90% of the maximum electric field strength and the sum of the squares of the electric field strength Ratio ,include: ； ； ； ； Calculation of electric field strength non-uniformity coefficient and , electric field intensity distortion rate and the coefficient of variation of electric field strength ,include: ； ； ； ； Calculate the maximum value of the electric field gradient , minimum electric field gradient , average electric field gradient and the median electric field gradient ,include: ； ； ； ； in, Indicates the The electric field gradient at each sampling point.

4. The method for evaluating the insulation strength of a live working gap according to claim 3, wherein: The performing normalization processing according to the parameter sets of the first gap and the second gap includes: The standardization process includes data processing using a z-score standardization method, including: ； in, represents the original data points; and represent the mean and standard deviation of the features respectively; Represents the normalized data points.

5. The method for evaluating the insulation strength of a live working gap according to claim 4, wherein: The LSSVR models are trained according to the first parameter set and the second parameter set respectively, to obtain the first model and the second model, including: Optimizing kernel function parameters and regularization parameters of the LSSVR model using a six-fold cross-validation method according to the first parameter set and the second parameter set, respectively, to obtain a first intermediate model and a second intermediate model; Model training is performed based on the first intermediate model in combination with the first parameter set, and verified by a preset validation set to obtain the first model; model training is performed based on the second intermediate model in combination with the second parameter set, and verified by a preset validation set to obtain the second model.

6. The method for evaluating the insulation strength of a live working gap according to claim 5, wherein: The kernel function parameters and regularization parameters of the LSSVR model are optimized using a six-fold cross validation method according to the first parameter set and the second parameter set, respectively, to obtain a first intermediate model and a second intermediate model including: Setting the value ranges of the kernel function parameters and the regularization parameters to be a preset parameter value set; the preset parameter value set includes a preset number of parameter values; exhaustively enumerating all value combinations of the kernel function parameters and the regularization parameters according to the preset parameter value set, and obtaining a parameter optimization training set according to all value combinations; The first parameter set and the second parameter set are randomly divided into 6 parameter subsets respectively; the LSSVR model is initialized according to the kernel function parameter and regularization parameter value combination in the parameter optimization training set; Performing six iterative optimizations on the LSSVR model for each combination of kernel function parameters and regularization parameter values ​​based on the parameter subset of the first parameter set, wherein five parameter subsets are selected for training in each iteration, one parameter subset is retained for verification, and the mean absolute error (MAE) of the verification results is calculated; statistically calculating the mean absolute error (MAE) in all iterative optimizations of the LSSVR model for all combinations of kernel function parameters and regularization parameter values, selecting the kernel function parameter and regularization parameter value combination corresponding to the minimum MAE as the model parameter, and obtaining a first intermediate model; The LSSVR model of each combination of kernel function parameters and regularization parameter values ​​is optimized iteratively six times according to the parameter subset of the second parameter set, wherein five parameter subsets are selected for training in each iteration, one parameter subset is retained for verification, and the mean absolute error (MAE) of the verification results is calculated; the mean absolute error (MAE) of all iterative optimizations of the LSSVR model of all combinations of kernel function parameters and regularization parameter values ​​is calculated, and the kernel function parameter and regularization parameter value combination corresponding to the minimum MAE is selected as the model parameter to obtain the second intermediate model; The mean absolute error (MAE) of the calculation verification results includes: ； in, Indicates actual value; represents the predicted value; Indicates the sample size.

7. The method for evaluating the insulation strength of a live working gap according to claim 6, wherein: Performing model training based on the first intermediate model in combination with the first parameter set, and validating the model using a preset validation set, to obtain the first model; Performing model training according to the second intermediate model in combination with the second parameter set and verifying it using a preset validation set to obtain the second model includes: The optimization problems of the LSSVR model include: ； Constraint statistics include: ； in, represents the weight vector; represents the bias term; Represents a mapping function that maps input data to a high-dimensional feature space; represents the error term, represents the regularization parameter used to control the complexity of the model; Indicates transposition; Via Lagrange multipliers With the introduction of , the optimization problem of the LSSVR model is transformed into a dual problem: ； in, represents the kernel function; Represents the sample point to be predicted; Indicates the first sample points; A kernel function matrix is ​​constructed based on the Gaussian kernel function, and the kernel function matrix includes the Gaussian kernel function values ​​between all training samples. , is the kernel width parameter; the kernel function matrix is ​​substituted into the LSSVR model to obtain a linear equation system; the linear equation system is solved to obtain The trained model is obtained by taking the values ​​of a and b. The trained model is verified based on the preset validation set, and the training effect of the model is determined by calculating the mean absolute error (MAE) and the coefficient of determination (R²) on the preset validation set. The model training is completed when the required training effect is achieved.

8. The method for evaluating the insulation strength of a live working gap according to claim 7, wherein: The insulation strength assessment includes: After the actual value of the parameter set is input into the first model or the second model, the model outputs a predicted value of the target combined gap 50% breakdown voltage; the maximum operating overvoltage of the system is obtained according to the operating status and historical data of the power system; The live working hazard rate of the target combination gap is obtained based on the predicted value of the 50% breakdown voltage of the target combination gap and the maximum operating overvoltage of the system. The safety of the live working is evaluated based on the live working hazard rate to complete the insulation strength evaluation.

9. The method for evaluating the insulation strength of a live working gap according to claim 8, wherein: The live working hazard rate of the target combined gap is obtained based on the predicted value of the 50% breakdown voltage of the target combined gap and the maximum operating overvoltage of the system. The safety of the live working is evaluated based on the hazard rate, including: Assuming that the probability distribution of system operation overvoltage and the probability distribution of air gap breakdown both obey normal distribution, the hazard rates according to probability statistics include: ； ； ； ； in, represents the hazard rate; The probability density function representing the overvoltage amplitude; represents the probability distribution function of air gap breakdown under switching overvoltage with amplitude U; Indicates the standard deviation of switching overvoltage; Indicates the predicted value of 50% breakdown voltage of target combination gap; Indicates the standard deviation of the gap discharge voltage during live working; Indicates the average value of switching overvoltage; Indicates the relative standard deviation of overvoltage; Indicates the maximum operating overvoltage of the system; If the hazard rate is less than a preset hazard threshold, the live working is determined to be safe; otherwise, the live working is determined to be unsafe.

10. An insulation strength assessment device for a combined gap during live working, used in the method according to any one of claims 1 to 9, characterized in that: The device includes a first module, a second module and a third module; The first module is used to construct a finite element model for a preset data set; the preset data set includes a preset number of combined gaps of total gap lengths, wherein the combined gaps that meet a first condition are classified into a first subset, and the combined gaps that do not meet the first condition are classified into a second subset; the first condition includes that the first gap length in the combined gaps is greater than a first gap length critical value corresponding to 50% of the breakdown voltage; The finite element model includes a high-voltage electrode, a ground electrode, and a suspended conductor electrode; the suspended conductor electrode is suspended between the high-voltage electrode and the ground electrode; the gap between the high-voltage electrode and the ground electrode is defined as the total gap, the gap between the suspended conductor electrode and the high-voltage electrode is defined as the first gap, and the gap between the suspended conductor electrode and the ground electrode is defined as the second gap; The second module is used to obtain parameter sets of the first subset and the second subset respectively, to obtain a first parameter set and a second parameter set; the parameter set includes preset electric field strength characteristic quantities of the first gap and the second gap in the combined gap; Train the LSSVR model according to the first parameter set and the second parameter set respectively to obtain a first model and a second model; The third module is used to obtain the actual value of the first gap length and the actual value of the parameter set of the target combination gap; if the actual value of the first gap length meets the first condition, the insulation strength evaluation is performed according to the first model and the actual value of the parameter set; otherwise, the insulation strength evaluation is performed according to the second model and the actual value of the parameter set.

Images