Method and device for determining an air gap breakdown voltage, computer program product

Abstract

The application provides a method and device for determining an air gap breakdown voltage, and a computer program product, comprising: obtaining air gap data, the air gap data comprising an electric field feature set formed by a plurality of electric field characteristics; inputting the air gap data into a breakdown voltage determination model, and outputting a target breakdown voltage, the breakdown voltage determination model being a converged model obtained by training a predetermined model using a plurality of sets of training data, each set of training data comprising a sample electric field feature set and a sample breakdown voltage; and determining that the insulation characteristics of the air gap do not meet the relay protection requirements in the case where the target breakdown voltage is less than a set voltage. The method solves the problem of low prediction accuracy caused by long cycle and high cost of experimental research on air gap breakdown characteristics in the prior art, which mainly relies on experimental data.

Description

Technical Field

[0001] This invention relates to the field of research on the breakdown characteristics of air gaps, and more specifically, to a method, apparatus, computer-readable storage medium, and computer program product for determining the breakdown voltage of an air gap. Background Technology

[0002] The breakdown characteristics of air gaps are a key factor in the external insulation design of UHV transmission and transformation projects, but current theoretical research cannot fully explain their complex behavior. Therefore, engineering design relies heavily on experimental data, derived from long-gap discharge tests on typical structural gaps such as rod-plate gaps and actual engineering gaps. This method has significant limitations: 1. The types of gap structures used in actual transmission and transformation projects are diverse, and their discharge characteristics vary accordingly. Standard gap models cannot fully cover all possible engineering application scenarios, thus limiting the model's universality and application scope. 2. Voltage waveforms and atmospheric conditions have a significant impact on the breakdown characteristics of air gaps. To ensure the broad applicability of experimental data, tests need to be conducted under various voltage waveforms and environmental conditions, increasing the complexity of the experiments. 3. When gap coefficients are used to predict the breakdown voltage of actual engineering gaps, they often deviate from experimental values, increasing the uncertainty of the prediction results. Experimental research on the breakdown characteristics of air gaps is time-consuming and costly. Summary of the Invention

[0003] The main objective of this application is to provide a method, apparatus, computer-readable storage medium, and computer program product for determining air gap breakdown voltage, so as to at least solve the problem that the prior art mainly relies on experimental data, and the experimental study of air gap breakdown characteristics is characterized by long cycle, high cost, and low accuracy of prediction results.

[0004] To achieve the above objectives, according to one aspect of this application, a method for determining the breakdown voltage of an air gap is provided, comprising: acquiring air gap data, the air gap data including an electric field feature set formed by multiple electric field feature quantities, the electric field feature set including at least electric field feature quantities of a discharge channel region, electric field feature quantities of an electrode surface region, and electric field feature quantities of an external non-breakdown region, the electric field feature quantities including at least electric field strength, electric field energy, energy density, surface area of ​​an electric field strength unit, electric field gradient, path length of an electric field strength unit, and scaling parameter; inputting the air gap data into a breakdown voltage determination model, and outputting a target breakdown voltage, the breakdown voltage determination model being a convergent model obtained by training a predetermined model using multiple sets of training data, each set of training data including a sample electric field feature set and a sample breakdown voltage; and determining that the insulation characteristics of the air gap do not meet the relay protection requirements when the target breakdown voltage is less than a set voltage.

[0005] Optionally, before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, the method further includes: conducting breakdown tests on the air gaps of different typical electrodes to obtain breakdown voltage data corresponding to each typical electrode; determining the breakdown range and non-breakdown range corresponding to each typical electrode based on the breakdown voltage data; performing electric field calculations on the corresponding breakdown range and non-breakdown range for different typical electrodes to obtain multiple electric field characteristic quantities; and performing preprocessing operations on the multiple electric field characteristic quantities to obtain the sample electric field characteristic set.

[0006] Optionally, preprocessing operations are performed on multiple electric field feature quantities to obtain the sample electric field feature set, including: normalizing each electric field feature quantity to obtain normalized electric field feature quantities; and performing dimensionality reduction processing on the normalized electric field feature quantities using principal component analysis to obtain the sample electric field feature set.

[0007] Optionally, before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, the method further includes: constructing an initial breakdown voltage determination model and initializing the model parameters of the initial breakdown voltage determination model, wherein the initial breakdown voltage determination model is an XGBoost model to be trained, and the model parameters include at least the classifier type, the weight of positive samples, the number of decision trees, the maximum depth of the decision trees, the learning rate, and the penalty coefficient; and a training step, in which multiple sets of sample electric field feature sets are input into the initial breakdown voltage determination model for training to obtain an intermediate breakdown voltage determination model, wherein the intermediate breakdown voltage determination model is the initial breakdown voltage determination model. The model obtained after model training is determined; the testing step involves inputting multiple sets of test electric field feature sets into the initial breakdown voltage determination model for testing to obtain the accuracy of the intermediate breakdown voltage determination model. Each set of test electric field feature sets includes test electric field feature quantities and test breakdown voltage. If the accuracy is less than a set accuracy, the training step and the testing step are re-executed at least once until the accuracy is greater than or equal to the set accuracy or the number of iterations reaches the maximum set number of iterations. The intermediate breakdown voltage determination model is then determined as the breakdown voltage determination model. The number of iterations is the number of times the initial breakdown voltage determination model is iteratively trained.

[0008] Optionally, inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage includes: inputting the air gap data into the breakdown voltage determination model to obtain interval characteristic values, wherein the interval characteristic values ​​are intermediate variables in the calculation process of the breakdown voltage determination model, and the interval characteristic values ​​are -1 or 1. When the interval characteristic value is 1, it indicates that the air gap has broken down, and when the interval characteristic value is -1, it indicates that the air gap has not broken down.

[0009] The first calculation step involves calculating the applied voltage according to a first formula when the characteristic value of the interval is -1. The first formula is U. L =U0+dU L U L The voltage dU represents the applied voltage, where U0 represents the initial voltage of the typical electrode under initial loading. L The input step involves setting an adjustment value, where the applied voltage is the voltage applied to the typical electrode of the power equipment. The first input step involves calculating the electric field of the air gap of the typical electrode to extract a current first electric field feature set, and then re-inputting this current first electric field feature set into the breakdown voltage determination model to obtain a first current interval feature value. This first current interval feature value is the interval feature value obtained by re-inputting the current first electric field feature set into the breakdown voltage determination model. The first calculation step and the first input step are repeated sequentially until the current interval feature value is 1, at which point the current applied voltage is determined as the target breakdown voltage.

[0010] Optionally, after inputting the air gap data into the breakdown voltage determination model to obtain the interval characteristic value, the method further includes: a second calculation step, whereby, when the interval characteristic value is 1, the applied voltage is calculated according to a second formula, where the second formula is U L =U0-dU L The second input step involves calculating the electric field of the air gap of the typical electrode to extract the current second electric field feature set, and re-inputting the current second electric field feature set into the breakdown voltage determination model to obtain the current second interval feature value. The current second interval feature value is the interval feature value obtained by re-inputting the current second electric field feature set into the breakdown voltage determination model. The second calculation step and the second input step are repeated sequentially until the current interval feature value is -1. The sum of the current applied voltage and the set adjustment value is determined as the target breakdown voltage.

[0011] Optionally, after inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, the method further includes: if the target breakdown voltage is greater than or equal to the set voltage, determining that the insulation characteristics of the air gap meet the relay protection requirements.

[0012] According to another aspect of this application, an apparatus for determining the breakdown voltage of an air gap is provided. The apparatus includes: an acquisition unit for acquiring air gap data, the air gap data including an electric field feature set formed by multiple electric field feature quantities, the electric field feature set including at least electric field feature quantities of a discharge channel region, electric field feature quantities of an electrode surface region, and electric field feature quantities of an external non-breakdown region, the electric field feature quantities including at least electric field strength, electric field energy, energy density, surface area of ​​an electric field strength unit, electric field gradient, path length of an electric field strength unit, and scale parameter; an input / output unit for inputting the air gap data into a breakdown voltage determination model and outputting a target breakdown voltage, the breakdown voltage determination model being a convergent model obtained by training a predetermined model using multiple sets of training data, each set of training data including a sample electric field feature set and a sample breakdown voltage; and a first determination unit for determining that the insulation characteristics of the air gap do not meet the relay protection requirements when the target breakdown voltage is less than a set voltage.

[0013] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform any of the methods described.

[0014] According to another aspect of this application, a computer program product is provided, including computer instructions that, when executed by a processor, implement any of the methods described.

[0015] Applying the technical solution of this application, in the method for determining the breakdown voltage of an air gap, firstly, air gap data is acquired. This air gap data includes an electric field feature set formed by multiple electric field characteristic quantities. The electric field feature set includes at least the electric field characteristic quantities of the discharge channel region, the electric field characteristic quantities of the electrode surface region, and the electric field characteristic quantities of the external non-breakdown region. These electric field characteristic quantities include at least electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scaling parameters. Then, the air gap data is input into a breakdown voltage determination model, and the target breakdown voltage is output. This breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage. Finally, if the target breakdown voltage is less than a set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements. This application uses typical electrode air gap data obtained from experiments as the training and testing sets for training a model to predict breakdown voltage. This allows for the creation of a breakdown voltage determination model for air gaps. By acquiring the air gap data to be predicted in real time and inputting it into the breakdown voltage determination model, the predicted result, i.e., the target breakdown voltage, is obtained, thus determining whether the insulation characteristics of the air gap meet the requirements of relay protection. This further provides guidance for the insulation protection of power system equipment and the safe and economical operation of the power grid. This application addresses the problem in existing technologies that primarily rely on experimental data, where experimental research on air gap breakdown characteristics is time-consuming, costly, and leads to low accuracy in prediction results. Attached Figure Description

[0016] Figure 1 A hardware block diagram of a mobile terminal for performing a method for determining air gap breakdown voltage according to an embodiment of this application is shown.

[0017] Figure 2 A flowchart illustrating a method for determining air gap breakdown voltage according to an embodiment of this application is shown.

[0018] Figure 3 A flowchart illustrating the prediction process of a breakdown voltage determination model provided according to an embodiment of this application is shown.

[0019] Figure 4 A structural block diagram of an air gap breakdown voltage determination device provided according to an embodiment of this application is shown.

[0020] The above figures include the following reference numerals:

[0021] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. Detailed Implementation

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] As described in the background section, existing experimental studies on air gap breakdown characteristics suffer from long cycles and high costs. To address the issue that relying primarily on experimental data and the long cycles and high costs of experimental studies on air gap breakdown characteristics lead to low accuracy in prediction results, embodiments of this application provide a method, apparatus, computer-readable storage medium, and computer program product for determining air gap breakdown voltage.

[0026] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0027] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a method of determining air gap breakdown voltage according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0028] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the method for determining the air gap breakdown voltage in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0029] This embodiment provides a method for determining the air gap breakdown voltage of a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] Figure 2 This is a flowchart of a method for determining the air gap breakdown voltage according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0031] Step S201: Obtain air gap data. The air gap data includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter.

[0032] Specifically, experiments were conducted on the air gaps of typical electrodes with different structures, and electric field calculations were performed to obtain air gap data.

[0033] It should also be noted that the air gap breakdown discharge path has a high degree of randomness and is difficult to determine accurately. Therefore, the electric field gradient characteristic class on the shortest path is used to characterize the gap structure and does not represent the discharge path in the actual physical process. Thus, the specific discharge path does not affect the prediction results of the method of this invention. The following electric field characteristic quantity classes are defined in terms of physical quantities and dimensions: electric field strength, electric field energy, energy density, surface area, electric field gradient, path length, and scaling parameters. The specific definitions of the electric field characteristic quantities are as follows: Maximum electric field strength: E m =maxE i (i = 1, 2, ..., n), where E i Let be the average electric field intensity of the i-th spatial element, where n is the total number of elements. Average electric field intensity: Electric field distortion rate: Total electric field energy W: In the formula, ε0 is the vacuum permittivity, V i Let V be the volume of the i-th spatial unit. Total volume V: Energy density W d : The proportion of the volume occupied by the region with an electric field strength exceeding x% of the maximum electric field strength, V rx : In the formula, V xi Let xn be the volume of the i-th cell exceeding x% of the maximum electric field intensity, and let xn be the total number of cells exceeding x% of the maximum electric field intensity. The percentage of energy W in the region exceeding x% of the maximum electric field intensity. rx : In the formula, W xi Let be the energy of the i-th cell exceeding x% of the maximum electric field strength. Total surface area S: In the formula, S i Let S be the surface area of ​​the i-th element, and n be the total number of elements. S represents the proportion of the surface area occupied by regions exceeding x% of the maximum electric field intensity. rx : In the formula, S xiLet L be the surface area of ​​the i-th unit exceeding x% of the maximum electric field intensity. Electric field intensity magnitude E′: E′=|-gradE|, where E is the electric field intensity at a point, a scalar. The proportion L represents the path length of the region exceeding x% of the maximum electric field intensity. rx : In the formula, L xi Let V be the shortest discharge path length for the i-th element exceeding x% of the maximum electric field strength, and L be the total shortest discharge path length. The aforementioned proportional parameters include V. rx W rx S rx and L rx .

[0034] Step S202: Input the above air gap data into the breakdown voltage determination model and output the target breakdown voltage. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage.

[0035] Specifically, an XGBOOST model is used to establish a breakdown voltage prediction model, which is then trained to obtain a breakdown voltage determination model. This breakdown voltage determination model takes an electric field feature set as input and the breakdown voltage as output. Therefore, by inputting the aforementioned air gap data into the model and using the breakdown voltage determination model for calculation, the target breakdown voltage can be output.

[0036] Step S203: If the target breakdown voltage is less than the set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements.

[0037] Specifically, based on the target breakdown voltage, it is determined whether the insulation characteristics of the air gap meet the protection requirements. If the predicted breakdown voltage is lower than the design voltage, it indicates that the insulation characteristics of the air gap may be insufficient, posing a safety risk, and improvement measures need to be taken.

[0038] In this embodiment, firstly, air gap data is acquired. This air gap data includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scaling parameter. Then, the air gap data is input into the breakdown voltage determination model, and the target breakdown voltage is output. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage. Finally, if the target breakdown voltage is less than a set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements. This application uses typical electrode air gap data obtained from experiments as the training and testing sets for training a model to predict breakdown voltage. This allows for the creation of a breakdown voltage determination model for air gaps. By acquiring the air gap data to be predicted in real time and inputting it into the breakdown voltage determination model, the predicted result, i.e., the target breakdown voltage, is obtained, thus determining whether the insulation characteristics of the air gap meet the requirements of relay protection. This further provides guidance for the insulation protection of power system equipment and the safe and economical operation of the power grid. This application addresses the problem in existing technologies that primarily rely on experimental data, where experimental research on air gap breakdown characteristics is time-consuming, costly, and leads to low accuracy in prediction results.

[0039] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the method for determining the air gap breakdown voltage of this application will be described in detail below with reference to specific embodiments.

[0040] To reduce the number of tests required for external insulation design to obtain the electric field feature set, in an optional embodiment, prior to step S202 above, the method further includes:

[0041] Step S301: Perform a breakdown test on the air gap of different typical electrodes to obtain the breakdown voltage data corresponding to each typical electrode.

[0042] Step S302: Determine the breakdown range and non-breakdown range corresponding to each of the above typical electrodes based on the breakdown voltage data.

[0043] Step S303: Apply the above-mentioned breakdown range and the above-mentioned non-breakdown range to different typical electrodes respectively, and perform electric field calculation to obtain multiple electric field characteristic quantities;

[0044] Step S304: Perform preprocessing operations on the above-mentioned electric field feature quantities to obtain the above-mentioned sample electric field feature set.

[0045] In the above embodiments, breakdown voltage data is obtained by conducting breakdown tests on air gaps of typical electrodes with different structures. For example, the measured breakdown voltage is U, and the range [90%U, U] is defined as the non-breakdown interval with a characteristic value of -1; the range [U, 110%U] is defined as the breakdown interval with a characteristic value of 1. For the initial applied voltage U0 of typical electrodes with different structures, electric field calculations are performed on their corresponding non-breakdown intervals and breakdown voltage intervals to obtain electric field characteristic quantities. After preprocessing, the electric field characteristic sets corresponding to each applied voltage are obtained. By exploring the use of algorithmic models to analyze the influence of different gap structures, voltage waveforms, and environmental factors on breakdown characteristics in actual power transmission and transformation projects, a dataset is obtained. The breakdown characteristics of air gaps are calculated or predicted and used to guide the external insulation design of power transmission and transformation projects, thereby achieving the goal of reducing the number of tests required for external insulation design.

[0046] To reduce computational complexity and accelerate model convergence, in one optional implementation, step S304 includes:

[0047] Step S3041: Normalize each of the above electric field characteristic quantities to obtain normalized electric field characteristic quantities.

[0048] Step S3042: Principal component analysis is used to reduce the dimensionality of the normalized electric field features to obtain the sample electric field feature set.

[0049] In the above embodiments, to reduce computational complexity and improve computational efficiency, the electric field feature set can be dimensionality reduced. To eliminate the influence of different orders of magnitude and dimensions, and to accelerate the training and convergence speed of the model, the electric field feature set needs to be normalized. Before dimensionality reduction of the electric field feature set, each feature quantity needs to be normalized to eliminate the influence of different orders of magnitude and dimensions, reduce the impact of improper data processing on the prediction effect, and accelerate the training and convergence speed of the prediction model. The normalization method is as follows: In the formula, x′ i Let x be a certain characteristic quantity in the electric field characteristic set. i The normalized value, x max x is the maximum value of this characteristic quantity. min This is the minimum value of the characteristic quantity. At this point, the characteristic quantity x... i By [x max x min Normalize to [0,1].

[0050] Principal component analysis is a commonly used feature dimension reduction method. Based on the mathematical idea of dimension reduction, it attempts to recombine multiple highly correlated indicators to obtain a few uncorrelated comprehensive indicators and replace the original multiple indicators. This is a multivariate statistical method. Usually, the recombined comprehensive indicators are called principal components. The principal component analysis mainly includes the following steps: (1) Calculate the correlation coefficient matrix according to the normalized data table: R = (r ij ) p×p , where p is the feature dimension. (2) Calculate the eigenvalues and eigenvectors of R. Solve the characteristic equation |R - λI| = 0 to obtain the characteristic roots λ, and arrange them from largest to smallest as λ1, λ2,..., λ p , and at the same time, the corresponding eigenvectors u1, u2,..., u p can be obtained. I in the characteristic equation represents the identity matrix. (3) Calculate the variance contribution rate e i and the cumulative variance contribution rate P. (4) Select the first k principal components corresponding to the eigenvalues with the cumulative variance contribution rate reaching the predetermined value P (0.85 < P < 0.95) to obtain the above-mentioned sample electric field feature set. The calculation formula for the cumulative variance contribution rate of the principal components is: where k is the number of principal components, p is the number of original features, k ≤ p, and λ i is the eigenvalue of the i-th principal component.

[0051] In order to improve the accuracy of classification and recognition, in an optional implementation manner, before the above step S202, the method further includes:

[0052] Step S401, construct an initial breakdown voltage determination model and initialize the model parameters of the above initial breakdown voltage determination model. The above initial breakdown voltage determination model is an XGBoost model to be trained. The above model parameters at least include the classifier type, the weight of positive samples, the number of decision trees, the maximum depth of the above decision trees, the learning rate, and the penalty coefficient;

[0053] Step S402, training step, input multiple groups of the above sample electric field feature sets into the above initial breakdown voltage determination model for training to obtain an intermediate breakdown voltage determination model. The above intermediate breakdown voltage determination model is the model obtained after training the above initial breakdown voltage determination model;

[0054] Step S403, testing step, input multiple groups of test electric field feature sets into the above initial breakdown voltage determination model for testing to obtain the accuracy rate of the above intermediate breakdown voltage determination model. Each group of the above test electric field feature sets includes test electric field feature quantities and test breakdown voltages;

[0055] Step S404: If the accuracy is less than the set accuracy, repeat the training step and the testing step at least once until the accuracy is greater than or equal to the set accuracy or the number of iterations reaches the maximum set number of iterations. Then, determine the intermediate breakdown voltage determination model as the breakdown voltage determination model. The number of iterations is the number of times the initial breakdown voltage determination model is iterated and trained.

[0056] In the above embodiments, the main purpose of training the XGBOOST model is to determine the best tree structure. The core issue is finding an optimal splitting node. A greedy algorithm is used to calculate the gain before and after the split to determine whether the node needs to be split. Then, the various parameters of the model are iteratively optimized based on the objective function. The XGBOOST model is finally obtained through training according to the above process. The training process is as follows: Figure 3 As shown.

[0057] The XGBoost model contains a large number of parameters, which can be broadly categorized into three types: The first type consists of conventional parameters that control the macroscopic functions, such as the classifier type and the weights of positive samples; the second type comprises model parameters related to the tree structure, such as the number of decision trees and the maximum tree depth; and the third type consists of target parameters for model learning, such as the learning rate and penalty coefficient. When building the XGBoost model, these parameters can be set appropriately to determine the optimal parameters, thereby improving the accuracy of classification and recognition.

[0058] The maximum number of trees, n_estimator, determines the model's learning ability. If the number of trees is too small, the model's learning capacity may be insufficient, reducing its predictive ability. Conversely, if the number of trees is too large, while it provides significant learning capacity, it may lead to overfitting. To quantitatively evaluate the quality of the classification prediction model, accuracy is used as the evaluation metric, calculated using the following formula: In the formula, n is the total number of test samples; n p This represents the number of correctly classified cases. Accuracy is positively correlated with the model's classification performance; that is, the higher the accuracy, the better the model's classification performance.

[0059] The maximum tree depth, `max_depth`, directly impacts prediction results. If this value is too small, the tree lacks sufficient depth, leading to imprecise classification and reduced prediction accuracy. Conversely, if the value is too large, overlearning can occur, resulting in only a slight improvement in prediction accuracy but a significant decrease in computational speed. A controlled variable method can be used to determine this parameter. Setting `max_depth` to a range of 1 to 10, comparing the classification accuracy at each depth, and selecting the `max_depth` with the highest classification accuracy as the maximum depth of the model tree.

[0060] The learning rate eta is based on the principle that taking small steps to gradually approach the result is more effective than taking large leaps directly. In other words, it doesn't rely on the entire tree but only learns a portion of it. The learning rate is used to control the weights of leaf nodes, thus adjusting the tree's influence on the final prediction. When the learning rate is small, the weight of the current leaf node decreases, and the tree's influence on the final prediction also decreases. Typically, the learning rate is not too large to prevent overfitting. Learning rate values ​​between 0 and 1, with intervals of 0.05, are selected to obtain the prediction accuracy under different learning rates. The classification accuracy values ​​are then compared to determine the optimal learning rate.

[0061] The minimum child weight (min_child_weight) is used to determine whether the current leaf node needs to be split. If the current weight of a leaf node is less than the set minimum weight, the splitting stops. The smaller this value, the more likely overfitting will occur. Based on the determined maximum number of trees, maximum depth, and learning rate, the value of min_child_weight is set to vary within the range of 0 to 10, and the accuracy under different minimum weights is observed to determine the minimum weight of the leaf.

[0062] It should also be noted that, as an ensemble tree model, the XGBOOST model contains multiple CART trees, and its output is the sum of the sum of these trees. This sum is then used as the predicted value of the XGBOOST model, and its mathematical expression is as follows: In the formula, K is the number of CART trees; F is the number of possible CART trees; and fk represents a specific CART tree. This is the output of the XGBOOST model. Given a sample length of n and the number of features of m, then: D={(x i Y i )}(|D|=n,x i ∈R m y i ∈R), where x i y represents the i-th sample input; i This represents the output corresponding to the i-th sample input; F is the space of the CART tree, which can be represented as: F = {f(x) = w} q (x)}(q:R m →T,w∈R T In the equation, q represents the structure of the CART tree; T is the number of child nodes in the CART tree; f(x) represents the CART tree structure q and the weights w of the child nodes; the XGBOOST model is built by learning the CART tree and determining its structure and weights. Substituting the regularization term into the objective function, we have: In the formula, and y iγ and λ represent the predicted and actual label values ​​of the i-th sample, respectively; γ and λ are weight coefficients; the first part of F(Φ) is the loss error, typically including mean squared error and logistic regression, while the second part is the regularization term, usually used to constrain the depth of the CART tree, reduce its complexity, and prevent overfitting. When building an XGBoost model to solve practical problems, the goal is to obtain the optimal coefficients so that the predicted values ​​approximate the true values ​​as closely as possible, and to give the model good generalization ability and computational speed, thus ensuring good versatility. During model training, if the process is additive, then: in, For the t-th round XGBOOST model; for The previous prediction model; f t (x i The function is newly added. During training, this new function is added to the t-th round. At this point, a constant term C is added to the cumulative difference between the predicted and actual values. The objective function can then be expressed as: Perform a Taylor expansion on the above equation:

[0063]

[0064] Since the residual of a CART tree is obtained by subtracting the previous tree from the t-th tree, the predicted value of the (t-1)-th tree is known. Since the part in the Taylor expansion is a fixed value that does not affect the optimization calculation of the objective function, it can be incorporated into the constant term or removed directly. Therefore, the objective function can be expressed as:

[0065]

[0066] The preceding section explained the relationship between error and the objective function. The following section explains how the regularization term added in XGBoost defines the complexity of the CART tree. The objective function is then transformed as follows:

[0067]

[0068] At this point, two regularization terms, the number of leaf nodes T and the leaf node score w, are added to the objective function. Therefore, the objective function part of the regularization term also adds two cumulative parts: the number of samples and the number of leaf nodes. The two terms in the above formula are then summed using G. j and H j If we substitute, the above equation can be simplified to: Replacing the above equation with w j Taking the derivative and setting it to 0, we get: Finally, w j Substituting the optimal solution into the objective function, we get: As shown in the above formula, once the structure of a tree is determined, the score of the structure is only related to its first and second derivatives, and the score is negatively correlated with the structure. That is, the smaller the score, the better the structure, and vice versa. In addition, to determine the structure of a tree, it is also necessary to pay attention to the node partitioning of the tree and the conditions under which the tree stops growing.

[0069] 1) Node Partitioning: Regarding node partitioning, XGBoost does not use the inefficient enumeration method, but instead employs a greedy algorithm to iteratively split the tree by adding nodes, starting from a single child node. After node partitioning, the loss function is as follows: The three terms in the above formula represent the scores of the left and right trees, and the score when no split is performed. These three terms are used to determine the value of the loss function. λ is used not only to control the complexity of the tree but also as a threshold to determine whether a split should be performed. If the gain after the split is greater than λ, then the split is performed; otherwise, it is not.

[0070] 2) Stopping Growth: A tree cannot grow indefinitely, as this would increase its complexity. Therefore, certain constraints are needed to limit the tree's growth, causing it to automatically stop growing. If the objective function value F of a child node after a new split... loss If the value is less than 0, the split will be abandoned, which is a balance between the loss value and the complexity of the model structure. In addition, when the depth of the tree reaches the set maximum depth, the splitting of child nodes will stop and no more trees will be built. The depth of the tree is related to whether the model will have underfitting, so the depth of the tree needs to be optimized to determine an appropriate depth. When a child node splits, the weight coefficients of the left and right leaf nodes will be updated. If the weight coefficient of any child node is less than a set threshold, the split will also be abandoned. Here, it is necessary to optimize another parameter of the model, the weight of the smallest child node, and thus determine the depth of the tree and the weight of the child nodes.

[0071] In order to accurately determine the breakdown voltage of the air gap, in one optional embodiment, step S202 includes:

[0072] Step S2021: Input the above air gap data into the breakdown voltage determination model to obtain the interval characteristic value. The above interval characteristic value is an intermediate variable in the calculation process of the above breakdown voltage determination model. The above interval characteristic value is -1 or 1. When the above interval characteristic value is 1, it indicates that the air gap has broken down. When the above interval characteristic value is -1, it indicates that the air gap has not broken down.

[0073] Step S2022, the first calculation step, when the characteristic value of the above interval is -1, calculate the applied voltage according to the first formula, where the first formula is U. L =U0+dUL U L The above-mentioned applied voltage is represented by U0, which represents the initial voltage of the typical electrode under initial loading. dU L This indicates the set adjustment value, where the above-mentioned applied voltage is the voltage applied to the above-mentioned typical electrodes of the power equipment;

[0074] Step S2023, first input step, perform electric field calculation on the air gap of the above typical electrode to extract the current first electric field feature set, and re-input the current first electric field feature set into the breakdown voltage determination model to obtain the first current interval feature value. The first current interval feature value is the interval feature value obtained by re-inputting the current first electric field feature set into the breakdown voltage determination model.

[0075] Step S2024: Repeat the first calculation step and the first input step in sequence until the current interval characteristic value is 1, and determine the current applied voltage as the target breakdown voltage.

[0076] In the above embodiments, if the interval characteristic value of the breakdown voltage determination model is -1, it indicates that the air gap has not broken down at this time. Therefore, the loading voltage is increased, i.e., the loading voltage U... L =U0+dU L Then, the electric field is recalculated for the air gap of the typical electrode, and the electric field feature set is extracted and input into the breakdown voltage determination model until the characteristic value of the breakdown voltage determination model interval is 1. At this time, the applied voltage U is the breakdown voltage of the electrode to be predicted.

[0077] To accurately determine the breakdown voltage of the air gap, in one optional embodiment, after step S2021, the method further includes:

[0078] Step S501, the second calculation step, where the characteristic value of the above-mentioned interval is 1, calculates the above-mentioned applied voltage according to the second formula, where the second formula is U. L =U0-dU L ;

[0079] Step S502, the second input step, calculates the electric field of the air gap of the typical electrode to extract the current second electric field feature set, and re-inputs the current second electric field feature set into the breakdown voltage determination model to obtain the current second interval feature value. The current second interval feature value is the interval feature value obtained by re-inputting the current second electric field feature set into the breakdown voltage determination model.

[0080] Step S503: Repeat the second calculation step and the second input step in sequence until the current interval characteristic value is -1, and determine the sum of the current applied voltage and the set adjustment value as the target breakdown voltage.

[0081] In the above embodiments, if the interval characteristic value of the breakdown voltage determination model is 1 when U0 is applied, then the applied voltage is reduced, i.e., the applied voltage U... L =U0-dU L The electric field of the air gap of a typical electrode is recalculated, and the electric field feature set is extracted and input into the breakdown voltage determination model until the interval feature value of the breakdown voltage determination model is -1. The applied voltage corresponding to the lower limit of the voltage interval with an interval feature value of 1 is the breakdown voltage. That is, the breakdown voltage of the electrode to be predicted is the sum of the current applied voltage and dU. L The sum of dU L It can be set and adjusted by the user.

[0082] To ensure that the insulation characteristics of the air gap meet the requirements, in an optional embodiment, after step S202 above, the method further includes:

[0083] Step S601: If the target breakdown voltage is greater than or equal to the set voltage, determine that the insulation characteristics of the air gap meet the relay protection requirements.

[0084] In the above embodiment, the breakdown voltage predicted by the breakdown voltage determination model (i.e., the target breakdown voltage) is compared with the system's set voltage. If the target breakdown voltage is higher than or equal to the design voltage, the insulation characteristics of the opening gap can be considered sufficient to meet the protection requirements.

[0085] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0086] This application also provides an apparatus for determining air gap breakdown voltage. It should be noted that this apparatus can be used to execute the method for determining air gap breakdown voltage provided in this application. This apparatus is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0087] The following describes the apparatus for determining the air gap breakdown voltage provided in the embodiments of this application.

[0088] Figure 4 This is a structural block diagram of a device for determining the air gap breakdown voltage according to an embodiment of this application. Figure 4 As shown, the device includes:

[0089] The acquisition unit 10 is used to acquire air gap data, which includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter.

[0090] Specifically, experiments were conducted on the air gaps of typical electrodes with different structures, and electric field calculations were performed to obtain air gap data.

[0091] It should also be noted that the air gap breakdown discharge path has a high degree of randomness and is difficult to determine accurately. Therefore, the electric field gradient characteristic class on the shortest path is used to characterize the gap structure and does not represent the discharge path in the actual physical process. Thus, the specific discharge path does not affect the prediction results of the method of this invention. The following electric field characteristic quantity classes are defined in terms of physical quantities and dimensions: electric field strength, electric field energy, energy density, surface area, electric field gradient, path length, and scaling parameters. The specific definitions of the electric field characteristic quantities are as follows: Maximum electric field strength: E m =maxE i (i = 1, 2, ..., n), where E i Let be the average electric field intensity of the i-th spatial element, where n is the total number of elements. Average electric field intensity: Electric field distortion rate: Total electric field energy W: In the formula, ε0 is the vacuum permittivity, V i Let V be the volume of the i-th spatial unit. Total volume V: Energy density W d : The proportion of the volume occupied by the region with an electric field strength exceeding x% of the maximum electric field strength, V rx : In the formula, V xi Let xn be the volume of the i-th cell exceeding x% of the maximum electric field intensity, and let xn be the total number of cells exceeding x% of the maximum electric field intensity. The percentage of energy W in the region exceeding x% of the maximum electric field intensity. rx : In the formula, W xi Let be the energy of the i-th cell exceeding x% of the maximum electric field strength. Total surface area S: In the formula, S i Let S be the surface area of ​​the i-th element, and n be the total number of elements. S represents the proportion of the surface area occupied by regions exceeding x% of the maximum electric field intensity. rx : In the formula, S xi Let L be the surface area of ​​the i-th unit exceeding x% of the maximum electric field intensity. Electric field intensity magnitude E′: E′=|-gradE|, where E is the electric field intensity at a point, a scalar. The proportion L represents the path length of the region exceeding x% of the maximum electric field intensity. rx : In the formula, L xi Let V be the shortest discharge path length for the i-th element exceeding x% of the maximum electric field strength, and L be the total shortest discharge path length. The aforementioned proportional parameters include V. rx W rx S rx and L rx .

[0092] The input / output unit 20 is used to input the air gap data into the breakdown voltage determination model and output the target breakdown voltage. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage.

[0093] Specifically, an XGBOOST model is used to establish a breakdown voltage prediction model, which is then trained to obtain a breakdown voltage determination model. This breakdown voltage determination model takes an electric field feature set as input and the breakdown voltage as output. Therefore, by inputting the aforementioned air gap data into the model and using the breakdown voltage determination model for calculation, the target breakdown voltage can be output.

[0094] The first determination unit 30 determines that the insulation characteristics of the air gap do not meet the relay protection requirements when the target breakdown voltage is less than the set voltage.

[0095] Specifically, based on the target breakdown voltage, it is determined whether the insulation characteristics of the air gap meet the protection requirements. If the predicted breakdown voltage is lower than the design voltage, it indicates that the insulation characteristics of the air gap may be insufficient, posing a safety risk, and improvement measures need to be taken.

[0096] In this embodiment, the acquisition unit is used to acquire air gap data, which includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least electric field strength, electric field energy, energy density, surface area of ​​electric field strength unit, electric field gradient, path length of electric field strength unit, and scaling parameter. The input / output unit is used to input the air gap data into the breakdown voltage determination model and output the target breakdown voltage. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage. The first determination unit is used to determine that the insulation characteristics of the air gap do not meet the relay protection requirements when the target breakdown voltage is less than a set voltage. This application uses typical electrode air gap data obtained from experiments as the training and testing sets for training a model to predict breakdown voltage. This allows for the creation of a breakdown voltage determination model for air gaps. By acquiring the air gap data to be predicted in real time and inputting it into the breakdown voltage determination model, the predicted result, i.e., the target breakdown voltage, is obtained, thus determining whether the insulation characteristics of the air gap meet the requirements of relay protection. This further provides guidance for the insulation protection of power system equipment and the safe and economical operation of the power grid. This application addresses the problem in existing technologies that primarily rely on experimental data, where experimental research on air gap breakdown characteristics is time-consuming, costly, and leads to low accuracy in prediction results.

[0097] To reduce the number of tests required for external insulation design to obtain the electric field characteristic set, in one optional embodiment, the device further includes:

[0098] The test unit is used to perform breakdown tests on the air gaps of different typical electrodes before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, so as to obtain the breakdown voltage data corresponding to each of the above-mentioned typical electrodes.

[0099] The determining unit is used to determine the breakdown range and non-breakdown range corresponding to each of the above-mentioned typical electrodes based on the breakdown voltage data.

[0100] The first calculation unit is used to perform electric field calculations on the corresponding breakdown interval and non-breakdown interval of different typical electrodes to obtain multiple electric field characteristic quantities.

[0101] The preprocessing unit is used to perform preprocessing operations on multiple of the above-mentioned electric field feature quantities to obtain the above-mentioned sample electric field feature set.

[0102] In the above embodiments, breakdown voltage data is obtained by conducting breakdown tests on air gaps of typical electrodes with different structures. For example, the measured breakdown voltage is U, and the range [90%U, U] is defined as the non-breakdown interval with a characteristic value of -1; the range [U, 110%U] is defined as the breakdown interval with a characteristic value of 1. For the initial applied voltage U0 of typical electrodes with different structures, electric field calculations are performed on their corresponding non-breakdown intervals and breakdown voltage intervals to obtain electric field characteristic quantities. After preprocessing, the electric field characteristic sets corresponding to each applied voltage are obtained. By exploring the influence of different gap structures, voltage waveforms, and environmental factors on breakdown characteristics in actual power transmission and transformation projects using algorithmic models, a dataset is obtained. The breakdown characteristics of air gaps are calculated or predicted and used to guide the external insulation design of power transmission and transformation projects, thereby achieving the goal of reducing the number of tests required for external insulation design.

[0103] To reduce computational complexity and accelerate model convergence, in one optional implementation, the preprocessing unit includes:

[0104] The normalization module normalizes each of the above electric field characteristics to obtain normalized electric field characteristics.

[0105] The dimensionality reduction module uses principal component analysis to reduce the dimensionality of the normalized electric field features to obtain the sample electric field feature set.

[0106] In the above embodiments, to reduce computational complexity and improve computational efficiency, the electric field feature set can be dimensionality reduced. To eliminate the influence of different orders of magnitude and dimensions, and to accelerate the training and convergence speed of the model, the electric field feature set needs to be normalized. Before dimensionality reduction of the electric field feature set, each feature quantity needs to be normalized to eliminate the influence of different orders of magnitude and dimensions, reduce the impact of improper data processing on the prediction effect, and accelerate the training and convergence speed of the prediction model. The normalization method is as follows: In the formula, x′ i Let x be a certain characteristic quantity in the electric field characteristic set. i The normalized value, x max x is the maximum value of this characteristic quantity. min This is the minimum value of the characteristic quantity. At this point, the characteristic quantity x... i By [x max x min Normalize to [0,1].

[0107] Principal component analysis (PCA) is a commonly used feature reduction method. Based on the mathematical concept of dimensionality reduction, it recombines multiple highly correlated indicators to obtain a few independent composite indicators, which then replace the original multiple indicators. The recombined composite indicators are usually called principal components. PCA mainly includes the following steps: (1) Calculate the correlation coefficient matrix based on the normalized data table: R = (r... ij ) p×p In the formula, p is the feature dimension. (2) Calculate the eigenvalues ​​and eigenvectors of R. Solve the characteristic equation |R-λI|=0 to obtain the characteristic roots λ, and arrange them from largest to smallest as λ1, λ2, …, λ p At the same time, the corresponding feature vectors u1, u2, ..., u can be obtained. p In the characteristic equation, I represents the identity matrix. (3) Calculate the variance contribution rate e i And the cumulative variance contribution rate P. (4) Take the first k principal components corresponding to the eigenvalues ​​where the cumulative variance contribution rate reaches the predetermined value P (0.85 < P < 0.95) to obtain the above sample electric field feature set. The formula for calculating the cumulative variance contribution rate of the principal components is: In the formula, k is the principal component number, p is the original characteristic number, k≤p, and λ i Let be the eigenvalue of the i-th principal component.

[0108] To improve the accuracy of classification and recognition, in one optional embodiment, the device further includes:

[0109] The construction unit is used to construct an initial breakdown voltage determination model and initialize the model parameters of the initial breakdown voltage determination model before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage. The initial breakdown voltage determination model is an XGBoost model to be trained. The model parameters include at least the classifier type, the weight of positive samples, the number of decision trees, the maximum depth of the decision trees, the learning rate, and the penalty coefficient.

[0110] The training unit is used to perform the training steps, inputting multiple sets of the above-mentioned sample electric field feature sets into the above-mentioned initial breakdown voltage determination model for training, to obtain the intermediate breakdown voltage determination model, which is the model obtained after training the above-mentioned initial breakdown voltage determination model.

[0111] The test unit is used to perform test steps, input multiple sets of test electric field feature sets into the above-mentioned initial breakdown voltage determination model for testing, so as to obtain the accuracy of the above-mentioned intermediate breakdown voltage determination model. Each set of the above-mentioned test electric field feature sets includes test electric field feature quantities and test breakdown voltage.

[0112] The first repeating unit is used to re-execute the training steps and the testing steps at least once when the accuracy is less than the set accuracy, until the accuracy is greater than or equal to the set accuracy or the number of iterations reaches the maximum set number of iterations, and to determine the intermediate breakdown voltage determination model as the breakdown voltage determination model. The number of iterations is the number of times the initial breakdown voltage determination model is iteratively trained.

[0113] In the above embodiments, the main purpose of training the XGBOOST model is to determine the best tree structure. The core issue is finding an optimal splitting node. A greedy algorithm is used to calculate the gain before and after the split to determine whether the node needs to be split. Then, the various parameters of the model are iteratively optimized based on the objective function. The XGBOOST model is finally obtained through training according to the above process. The training process is as follows: Figure 3 As shown.

[0114] The XGBoost model contains a large number of parameters, which can be broadly categorized into three types: The first type consists of conventional parameters that control the macroscopic functions, such as the classifier type and the weights of positive samples; the second type comprises model parameters related to the tree structure, such as the number of decision trees and the maximum tree depth; and the third type consists of target parameters for model learning, such as the learning rate and penalty coefficient. When building the XGBoost model, these parameters can be set appropriately to determine the optimal parameters, thereby improving the accuracy of classification and recognition.

[0115] The maximum number of trees, n_estimator, determines the model's learning ability. If the number of trees is too small, the model's learning capacity may be insufficient, reducing its predictive ability. Conversely, if the number of trees is too large, while it provides significant learning capacity, it may lead to overfitting. To quantitatively evaluate the quality of the classification prediction model, accuracy is used as the evaluation metric, calculated using the following formula: In the formula, n is the total number of test samples; n p This represents the number of correctly classified cases. Accuracy is positively correlated with the model's classification performance; that is, the higher the accuracy, the better the model's classification performance.

[0116] The maximum tree depth, `max_depth`, directly impacts prediction results. If this value is too small, the tree lacks sufficient depth, leading to imprecise classification and reduced prediction accuracy. Conversely, if the value is too large, overlearning can occur, resulting in only a slight improvement in prediction accuracy but a significant decrease in computational speed. A controlled variable method can be used to determine this parameter. Setting `max_depth` to a range of 1 to 10, comparing the classification accuracy at each depth, and selecting the `max_depth` with the highest classification accuracy as the maximum depth of the model tree.

[0117] The learning rate eta is based on the principle that taking small steps to gradually approach the result is more effective than taking large leaps directly. In other words, it doesn't rely on the entire tree but only learns a portion of it. The learning rate is used to control the weights of leaf nodes, thus adjusting the tree's influence on the final prediction. When the learning rate is small, the weight of the current leaf node decreases, and the tree's influence on the final prediction also decreases. Typically, the learning rate is not too large to prevent overfitting. Learning rate values ​​between 0 and 1, with intervals of 0.05, are selected to obtain the prediction accuracy under different learning rates. The classification accuracy values ​​are then compared to determine the optimal learning rate.

[0118] The minimum child weight (min_child_weight) is used to determine whether the current leaf node needs to be split. If the current weight of a leaf node is less than the set minimum weight, the splitting stops. The smaller this value, the more likely overfitting will occur. Based on the determined maximum number of trees, maximum depth, and learning rate, the value of min_child_weight is set to vary within the range of 0 to 10, and the accuracy under different minimum weights is observed to determine the minimum weight of the leaf.

[0119] It should also be noted that, as an ensemble tree model, the XGBOOST model contains multiple CART trees, and its output is the sum of the sum of these trees. This sum is then used as the predicted value of the XGBOOST model, and its mathematical expression is as follows: In the formula, K is the number of CART trees; F is the number of possible CART trees; and fk represents a specific CART tree. This is the output of the XGBOOST model. Given a sample length of n and the number of features of m, then: D={(x i y i )}(|D|=n,x i ∈R m y i ∈R), where x i y represents the i-th sample input; i This represents the output corresponding to the i-th sample input; F is the space of the CART tree, which can be represented as: F = {f(x) = w} q (x)}(q:R m →T,w∈R T In the equation, q represents the structure of the CART tree; T is the number of child nodes in the CART tree; f(x) represents the CART tree structure q and the weights w of the child nodes; the XGBOOST model is built by learning the CART tree and determining its structure and weights. Substituting the regularization term into the objective function, we have: In the formula, Let yi and yi represent the predicted value and actual label value of the i-th sample, respectively; γ and λ are weight coefficients; the first part of F(Φ) is the loss error, typically including mean squared error and logistic regression, while the second part is the regularization term, usually used to constrain the depth of the CART tree, reduce its complexity, and prevent overfitting. When building an XGBoost model to solve practical problems, the goal is to obtain the optimal coefficients so that the predicted value approximates the true value as closely as possible, and to give the model good generalization ability and computational speed, thus ensuring good versatility. During model training, if the process is additive, then: in, For the t-th round XGBOOST model; for The previous prediction model; f t (x i The function is newly added. During training, this new function is added to the t-th round. At this point, a constant term C is added to the cumulative difference between the predicted and actual values. The objective function can then be expressed as: Perform a Taylor expansion on the above equation:

[0120]

[0121] Since the residual of a CART tree is obtained by subtracting the previous tree from the t-th tree, the predicted value of the (t-1)-th tree is known. Since the part in the Taylor expansion is a fixed value that does not affect the optimization calculation of the objective function, it can be incorporated into the constant term or removed directly. Therefore, the objective function can be expressed as:

[0122]

[0123] The preceding section explained the relationship between error and the objective function. The following section explains how the regularization term added in XGBoost defines the complexity of the CART tree. The objective function is then transformed as follows:

[0124]

[0125]

[0126] At this point, two regularization terms, the number of leaf nodes T and the leaf node score w, are added to the objective function. Therefore, the objective function part of the regularization term also adds two cumulative parts: the number of samples and the number of leaf nodes. The two terms in the above formula are then summed using G. j and H j If we substitute, the above equation can be simplified to: Replacing the above equation with w j Taking the derivative and setting it to 0, we get: Finally, w j Substituting the optimal solution into the objective function, we get: As shown in the above formula, once the structure of a tree is determined, the score of the structure is only related to its first and second derivatives, and the score is negatively correlated with the structure. That is, the smaller the score, the better the structure, and vice versa. In addition, to determine the structure of a tree, it is also necessary to pay attention to the node partitioning of the tree and the conditions under which the tree stops growing.

[0127] 1) Node Partitioning: Regarding node partitioning, XGBoost does not use the inefficient enumeration method, but instead employs a greedy algorithm to iteratively split the tree by adding nodes, starting from a single child node. After node partitioning, the loss function is as follows: The three terms in the above formula represent the scores of the left and right trees, and the score when no split is performed. These three terms are used to determine the value of the loss function. λ is used not only to control the complexity of the tree but also as a threshold to determine whether a split should be performed. If the gain after the split is greater than λ, then the split is performed; otherwise, it is not.

[0128] 2) Stopping Growth: A tree cannot grow indefinitely, as this would increase its complexity. Therefore, certain constraints are needed to limit the tree's growth, causing it to automatically stop growing. If the objective function value F of a child node after a new split... loss If the value is less than 0, the split will be abandoned, which is a balance between the loss value and the complexity of the model structure. In addition, when the depth of the tree reaches the set maximum depth, the splitting of child nodes will stop and no more trees will be built. The depth of the tree is related to whether the model will have underfitting, so the depth of the tree needs to be optimized to determine an appropriate depth. When a child node splits, the weight coefficients of the left and right leaf nodes will be updated. If the weight coefficient of any child node is less than a set threshold, the split will also be abandoned. Here, it is necessary to optimize another parameter of the model, the weight of the smallest child node, and thus determine the depth of the tree and the weight of the child nodes.

[0129] In order to accurately determine the breakdown voltage of the air gap, in one optional embodiment, the above-mentioned input / output unit includes:

[0130] The first input module inputs the air gap data into the breakdown voltage determination model to obtain the interval characteristic value. The interval characteristic value is an intermediate variable in the calculation process of the breakdown voltage determination model. The interval characteristic value is -1 or 1. When the interval characteristic value is 1, it indicates that the air gap has broken down. When the interval characteristic value is -1, it indicates that the air gap has not broken down.

[0131] The first calculation module is used to perform the first calculation step, which calculates the applied voltage according to the first formula when the characteristic value of the above-mentioned interval is -1. The first formula is U. L =U0+dU L U L The above-mentioned applied voltage is represented by U0, which represents the initial voltage of the typical electrode under initial loading. dU L This indicates the set adjustment value, where the above-mentioned applied voltage is the voltage applied to the above-mentioned typical electrodes of the power equipment;

[0132] The second input module is used to execute the first input step, to perform electric field calculation on the air gap of the typical electrode to extract the current first electric field feature set, and to re-input the current first electric field feature set into the breakdown voltage determination model to obtain the first current interval feature value. The first current interval feature value is the interval feature value obtained by re-inputting the current first electric field feature set into the breakdown voltage determination model.

[0133] The repeating module repeats the first calculation step and the first input step in sequence until the current interval characteristic value is 1, and determines the current applied voltage as the target breakdown voltage.

[0134] In the above embodiments, if the interval characteristic value of the breakdown voltage determination model is -1, it indicates that the air gap has not broken down at this time. Therefore, the loading voltage is increased, i.e., the loading voltage U... L =U0+dU L Then, the electric field is recalculated for the air gap of the typical electrode, and the electric field feature set is extracted and input into the breakdown voltage determination model until the characteristic value of the breakdown voltage determination model interval is 1. At this time, the applied voltage U is the breakdown voltage of the electrode to be predicted.

[0135] In an optional embodiment, to accurately determine the breakdown voltage of the air gap, the device further includes:

[0136] The second calculation unit performs the second calculation step. After inputting the air gap data into the breakdown voltage determination model and obtaining the interval characteristic value, if the interval characteristic value is 1, it calculates the applied voltage according to the second formula, where the second formula is U. L =U0-dU L ;

[0137] The input unit is used to perform the second input step, to calculate the electric field of the air gap of the typical electrode to extract the current second electric field feature set, and to re-input the current second electric field feature set into the breakdown voltage determination model to obtain the current second interval feature value. The current second interval feature value is the interval feature value obtained by re-inputting the current second electric field feature set into the breakdown voltage determination model.

[0138] The second repeating unit is used to repeat the second calculation step and the second input step in sequence until the current interval characteristic value is -1, and to determine the sum of the current loading voltage and the set adjustment value as the target breakdown voltage.

[0139] In the above embodiments, if the interval characteristic value of the breakdown voltage determination model is 1 when U0 is applied, then the applied voltage is reduced, i.e., the applied voltage U... L =U0-dU L The electric field of the air gap of a typical electrode is recalculated, and the electric field feature set is extracted and input into the breakdown voltage determination model until the interval feature value of the breakdown voltage determination model is -1. The applied voltage corresponding to the lower limit of the voltage interval with an interval feature value of 1 is the breakdown voltage. That is, the breakdown voltage of the electrode to be predicted is the sum of the current applied voltage and dU. L The sum of dU L It can be set and adjusted by the user.

[0140] To ensure that the insulation characteristics of the air gap meet the requirements, in one optional embodiment, the device further includes:

[0141] The second determination unit is used to determine that the insulation characteristics of the air gap meet the relay protection requirements when the target breakdown voltage is greater than or equal to the set voltage after the air gap data is input into the breakdown voltage determination model and the target breakdown voltage is output.

[0142] In the above embodiment, the breakdown voltage predicted by the breakdown voltage determination model (i.e., the target breakdown voltage) is compared with the system's set voltage. If the target breakdown voltage is higher than or equal to the design voltage, the insulation characteristics of the opening gap can be considered sufficient to meet the protection requirements.

[0143] The aforementioned device for determining the air gap breakdown voltage includes a processor and a memory. The acquisition unit, input / output unit, and first determination unit are all stored as program units in the memory, and the processor executes these program units to achieve the corresponding functions. All of the aforementioned modules are located in the same processor; alternatively, the modules may be located in different processors in any combination.

[0144] The processor contains a kernel, which retrieves the corresponding program unit from memory. One or more kernels can be configured. By adjusting kernel parameters, the problems of existing technologies that primarily rely on experimental data, and the long, costly, and inaccurate experimental studies of air gap breakdown characteristics, leading to poor prediction accuracy, can be addressed.

[0145] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0146] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the method for determining the air gap breakdown voltage.

[0147] Specifically, the methods for determining the air gap breakdown voltage include:

[0148] Step S201: Obtain air gap data. The air gap data includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter.

[0149] Step S202: Input the above air gap data into the breakdown voltage determination model and output the target breakdown voltage. The above breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of the above training data includes a sample electric field feature set and a sample breakdown voltage.

[0150] Step S203: If the target breakdown voltage is less than the set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements.

[0151] This invention provides a processor for running a program, wherein the program executes the method for determining the air gap breakdown voltage.

[0152] Specifically, the methods for determining the air gap breakdown voltage include:

[0153] Step S201: Obtain air gap data. The air gap data includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter.

[0154] Step S202: Input the above air gap data into the breakdown voltage determination model and output the target breakdown voltage. The above breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of the above training data includes a sample electric field feature set and a sample breakdown voltage.

[0155] Step S203: If the target breakdown voltage is less than the set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements.

[0156] This invention provides an air gap breakdown voltage management system, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps:

[0157] Step S201: Obtain air gap data. The air gap data includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter.

[0158] Step S202: Input the above air gap data into the breakdown voltage determination model and output the target breakdown voltage. The above breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of the above training data includes a sample electric field feature set and a sample breakdown voltage.

[0159] Step S203: If the target breakdown voltage is less than the set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements.

[0160] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0161] Step S201: Obtain air gap data. The air gap data includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter.

[0162] Step S202: Input the above air gap data into the breakdown voltage determination model and output the target breakdown voltage. The above breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of the above training data includes a sample electric field feature set and a sample breakdown voltage.

[0163] Step S203: If the target breakdown voltage is less than the set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements.

[0164] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0165] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0166] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0167] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0168] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0169] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0170] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0171] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0172] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0173] As can be seen from the above description, the embodiments of this application achieve the following technical effects:

[0174] 1) The method for determining the air gap breakdown voltage of this application firstly involves acquiring air gap data, which includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter. Then, the air gap data is input into a breakdown voltage determination model, and a target breakdown voltage is output. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage. Finally, if the target breakdown voltage is less than a set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements. This application uses typical electrode air gap data obtained from experiments as the training and testing sets for training a model to predict breakdown voltage. This allows for the creation of a breakdown voltage determination model for air gaps. By acquiring the air gap data to be predicted in real time and inputting it into the breakdown voltage determination model, the predicted result, i.e., the target breakdown voltage, is obtained, thus determining whether the insulation characteristics of the air gap meet the requirements of relay protection. This further provides guidance for the insulation protection of power system equipment and the safe and economical operation of the power grid. This application addresses the problem in existing technologies that primarily rely on experimental data, where experimental research on air gap breakdown characteristics is time-consuming, costly, and leads to low accuracy in prediction results.

[0175] 2) The air gap breakdown voltage determination device of this application includes an acquisition unit for acquiring air gap data, wherein the air gap data includes an electric field feature set formed by multiple electric field feature quantities, wherein the electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region, wherein the electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter; an input / output unit for inputting the air gap data into a breakdown voltage determination model and outputting a target breakdown voltage, wherein the breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data, wherein each set of training data includes a sample electric field feature set and a sample breakdown voltage; and a first determination unit for determining that the insulation characteristics of the air gap do not meet the relay protection requirements when the target breakdown voltage is less than a set voltage. This application uses typical electrode air gap data obtained from experiments as the training and testing sets for training a model to predict breakdown voltage. This allows for the creation of a breakdown voltage determination model for air gaps. By acquiring the air gap data to be predicted in real time and inputting it into the breakdown voltage determination model, the predicted result, i.e., the target breakdown voltage, is obtained, thus determining whether the insulation characteristics of the air gap meet the requirements of relay protection. This further provides guidance for the insulation protection of power system equipment and the safe and economical operation of the power grid. This application addresses the problem in existing technologies that primarily rely on experimental data, where experimental research on air gap breakdown characteristics is time-consuming, costly, and leads to low accuracy in prediction results.

[0176] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for determining the breakdown voltage of an air gap, characterized in that, include: Acquire air gap data, which includes an electric field feature set formed by multiple electric field feature quantities. The electric field feature set includes at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region. The electric field feature quantities include at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter. The air gap data is input into the breakdown voltage determination model, and the target breakdown voltage is output. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage. If the target breakdown voltage is less than the set voltage, it is determined that the insulation characteristics of the air gap do not meet the relay protection requirements. Before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, the method further includes: conducting breakdown tests on the air gaps of different typical electrodes to obtain breakdown voltage data corresponding to each typical electrode; determining the breakdown range and non-breakdown range corresponding to each typical electrode based on the breakdown voltage data; performing electric field calculations on the corresponding breakdown range and non-breakdown range for different typical electrodes to obtain multiple electric field characteristic quantities; and performing preprocessing operations on the multiple electric field characteristic quantities to obtain the sample electric field characteristic set. The process of inputting the air gap data into a breakdown voltage determination model and outputting the target breakdown voltage includes: inputting the air gap data into the breakdown voltage determination model to obtain interval characteristic values, wherein the interval characteristic values ​​are intermediate variables in the calculation process of the breakdown voltage determination model, and the interval characteristic values ​​are -1 or 1. An interval characteristic value of 1 indicates air gap breakdown, and an interval characteristic value of -1 indicates air gap non-breakdown; a first calculation step, wherein when the interval characteristic value is -1, the applied voltage is calculated according to a first formula, wherein the first formula is... , This indicates the applied voltage. This represents the initial voltage of the typical electrode under initial loading. The input step involves setting an adjustment value, where the applied voltage is the voltage applied to the typical electrode of the power equipment. The first input step involves calculating the electric field of the air gap of the typical electrode to extract a current first electric field feature set, and then re-inputting this current first electric field feature set into the breakdown voltage determination model to obtain a first current interval feature value. This first current interval feature value is the interval feature value obtained by re-inputting the current first electric field feature set into the breakdown voltage determination model. The first calculation step and the first input step are repeated sequentially until the current interval feature value is 1, at which point the current applied voltage is determined as the target breakdown voltage.

2. The method according to claim 1, characterized in that, Preprocessing operations are performed on multiple electric field feature quantities to obtain the sample electric field feature set, including: The electric field characteristic quantities are normalized to obtain normalized electric field characteristic quantities; Principal component analysis was used to reduce the dimensionality of the normalized electric field features to obtain the sample electric field feature set.

3. The method according to claim 1, characterized in that, Before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, the method further includes: Construct an initial breakdown voltage determination model and initialize the model parameters of the initial breakdown voltage determination model. The initial breakdown voltage determination model is an XGBoost model to be trained. The model parameters include at least the classifier type, the weight of positive samples, the number of decision trees, the maximum depth of the decision trees, the learning rate, and the penalty coefficient. The training step involves inputting multiple sets of the sample electric field feature sets into the initial breakdown voltage determination model for training, thereby obtaining an intermediate breakdown voltage determination model, which is the model obtained after training the initial breakdown voltage determination model. The testing steps involve inputting multiple sets of test electric field feature sets into the initial breakdown voltage determination model for testing, in order to obtain the accuracy of the intermediate breakdown voltage determination model. Each set of test electric field feature sets includes test electric field feature quantities and test breakdown voltage. If the accuracy is less than the set accuracy, the training step and the testing step are re-executed at least once until the accuracy is greater than or equal to the set accuracy or the number of iterations reaches the maximum set number of iterations. The intermediate breakdown voltage determination model is then determined as the breakdown voltage determination model, and the number of iterations is the number of times the initial breakdown voltage determination model is iterated and trained.

4. The method according to claim 1, characterized in that, After inputting the air gap data into the breakdown voltage determination model to obtain the interval characteristic values, the method further includes: The second calculation step involves calculating the applied voltage according to a second formula when the characteristic value of the interval is 1. The second formula is: ; The second input step involves calculating the electric field of the air gap of the typical electrode to extract the current second electric field feature set, and re-inputting the current second electric field feature set into the breakdown voltage determination model to obtain the current second interval feature value. The current second interval feature value is the interval feature value obtained by re-inputting the current second electric field feature set into the breakdown voltage determination model. Repeat the second calculation step and the second input step in sequence until the current interval characteristic value is -1, and determine the target breakdown voltage by the sum of the current applied voltage and the set adjustment value.

5. The method according to claim 1, characterized in that, After inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, the method further includes: If the target breakdown voltage is greater than or equal to the set voltage, the insulation characteristics of the air gap are determined to meet the relay protection requirements.

6. A device for determining the breakdown voltage of an air gap, characterized in that, The device includes: An acquisition unit is used to acquire air gap data, the air gap data including an electric field feature set formed by multiple electric field feature quantities, the electric field feature set including at least the electric field feature quantities of the discharge channel region, the electric field feature quantities of the electrode surface region, and the electric field feature quantities of the external non-breakdown region, the electric field feature quantities including at least the electric field strength, electric field energy, energy density, surface area of ​​the electric field strength unit, electric field gradient, path length of the electric field strength unit, and scale parameter; The input / output unit is used to input the air gap data into the breakdown voltage determination model and output the target breakdown voltage. The breakdown voltage determination model is a convergent model obtained by training a predetermined model using multiple sets of training data. Each set of training data includes a sample electric field feature set and a sample breakdown voltage. The first determination unit is used to determine that the insulation characteristics of the air gap do not meet the relay protection requirements when the target breakdown voltage is less than a set voltage. The device further includes: a testing unit, used to perform breakdown tests on the air gaps of different typical electrodes before inputting the air gap data into the breakdown voltage determination model and outputting the target breakdown voltage, to obtain breakdown voltage data corresponding to each typical electrode; a determination unit, used to determine the breakdown interval and non-breakdown interval corresponding to each typical electrode based on the breakdown voltage data; a first calculation unit, used to perform electric field calculations on the corresponding breakdown interval and non-breakdown interval for different typical electrodes respectively, to obtain multiple electric field feature quantities; and a preprocessing unit, used to perform preprocessing operations on the multiple electric field feature quantities to obtain the sample electric field feature set. The input / output unit includes: a first input module, which inputs the air gap data into the breakdown voltage determination model to obtain interval characteristic values, wherein the interval characteristic values ​​are intermediate variables in the calculation process of the breakdown voltage determination model, and the interval characteristic values ​​are -1 or 1, where an interval characteristic value of 1 indicates air gap breakdown, and an interval characteristic value of -1 indicates air gap non-breakdown; and a first calculation module, which performs a first calculation step, calculating the applied voltage according to a first formula when the interval characteristic value is -1. The first formula is: , This indicates the applied voltage. This represents the initial voltage of the typical electrode under initial loading. The first input module is used to execute the first input step, which calculates the electric field of the air gap of the typical electrode to obtain the current first electric field feature set, and re-inputs the current first electric field feature set into the breakdown voltage determination model to obtain the first current interval feature value. The first current interval feature value is the interval feature value obtained by re-inputting the current first electric field feature set into the breakdown voltage determination model. The second input module repeats the first calculation step and the first input step in sequence until the current interval feature value is 1, and determines the current applied voltage as the target breakdown voltage.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 5.

8. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the method described in any one of claims 1 to 5.

Images