High-voltage switch cabinet characteristic gas inversion partial discharge detection method and device

Abstract

The application provides a high-voltage switch cabinet characteristic gas inversion partial discharge detection method and device, and belongs to the technical field of high-voltage switch cabinet partial discharge monitoring. The steps comprise: obtaining field fault data representing the composition of mixed gas in the high-voltage switch cabinet when an insulation defect occurs in the cabinet, obtaining simulated fault data representing the composition of the mixed gas through simulation, and generating an original data set based on the field fault data and the simulated fault data; wherein the insulation defect types comprise N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects; the original data set is expanded by using a piecewise cubic spline interpolation (PSCI) method to obtain an expanded data set; a Markov chain Monte Carlo (MCMC) algorithm is introduced into a Gaussian process (GP) to resample the expanded data set to obtain a sample set F used for estimating a posterior distribution; a Bagging-DF model is used for characteristic gas inversion, the model is trained by using the sample set F, the model is tested by using the original data set, and the insulation defect type is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of partial discharge monitoring of high-voltage switchgear, in particular to a method and device for detecting partial discharge of high-voltage switchgear by characteristic gas inversion. BACKGROUND

[0002] High-voltage switchgear is mainly used to ensure the safe and stable operation of power transmission and distribution. It is usually composed of high-voltage circuit breakers, high-voltage load switches, high-voltage contactors, high-voltage fuses, high-voltage disconnectors, high-voltage grounding switches, high-voltage transformers, and station transformers, as well as related control, measurement, protection, regulation devices, internal connections, auxiliary components, housings, and support components. It is a widely used power transmission and distribution equipment in power systems.

[0003] With the rapid economic development in China, the power consumption and installed capacity of the power system are growing rapidly. When the power system experiences short-circuit, open-circuit failure, or needs to be powered off for maintenance, high-voltage switchgear can safely disconnect power through related devices, protecting downstream power equipment and ensuring the safety of power operators. However, improper operation during manufacturing, transportation, and maintenance of high-voltage switchgear can cause insulation defects inside the cabinet. Typical defect types can be divided into three categories: (1) metal protrusion defects: metal burrs protruding from the inner wall of the high-voltage switchgear cavity or high-voltage (HV) conductors; (2) metal particle defects: conductive particles left in the switchgear cavity during manufacturing and installation, as well as metal powder generated by friction of metal components during operation; (3) insulation layer air gap defects: internal air gaps in solid insulation layers due to aging, production process, and other factors. Air gaps may also occur due to differences in material expansion coefficients, plastic aging, and other issues. Partial discharge caused by insulation defects is one of the common faults of high-voltage switchgear. Early detection and treatment of partial discharge is crucial for preventing equipment failure.

[0004] In recent years, researchers have found that partial discharge inside high-voltage switchgear can cause changes in the composition of air around insulation defects, resulting in the generation of new gases. The type, content and generation rate of these characteristic gases are closely related to the type of defect and energy density. Therefore, by effectively extracting the characteristic information of the characteristic gas decomposition component, the internal insulation fault of the high-voltage switchgear can be effectively diagnosed. However, the concentration and composition of the characteristic gas may be affected by the internal environmental factors of the high-voltage switchgear (such as temperature, humidity, gas mixing, partial discharge mode, etc.), as well as maintenance state factors (operation method of maintenance personnel, inspection means and equipment condition), gas collection method, equipment condition and other factors, which may affect the fault diagnosis results; in addition, there may be limitations on the amount of effective decomposition data: insufficient data: due to the complexity and dynamics of the gas generation process caused by partial discharge, multiple characteristic gases may be generated, and in actual application, the amount of effective gas decomposition data obtained is limited. This means that when fault diagnosis is based on limited data, the model may not be able to fully capture all possible fault patterns, resulting in limited accuracy of fault diagnosis. Data missing or incomplete: in some cases, due to limitations of gas collection technology or insufficient gas sampling when a fault occurs, the data available for diagnosis is incomplete. Due to the above environmental and data limitations, misdiagnosis and missed diagnosis are prone to occur in the existing technology diagnosis process. Misdiagnosis may occur because the gases generated by different types of faults are similar, causing the model to misclassify them; missed diagnosis may occur when the amount of data is insufficient or the characteristic gas changes are relatively weak, and the fault is not accurately detected. SUMMARY

[0005] Therefore, the present application provides a high-voltage switchgear characteristic gas inversion partial discharge detection method and device, which can quickly and accurately identify the fault defect type inside the switchgear based on gas component data affected by internal environmental factors of the high-voltage switchgear, maintenance state factors, gas collection method, equipment condition and other factors, avoiding misdiagnosis and missed diagnosis, and improving diagnosis efficiency.

[0006] The technical solution adopted by the embodiment of the present application to solve the technical problems is:

[0007] A high-voltage switchgear characteristic gas inversion partial discharge detection method, comprising:

[0008] In step S1, field fault data representing a mixed gas component in a high-voltage switch cabinet when an insulation defect occurs in the cabinet is acquired, simulation fault data representing the mixed gas component is obtained by simulating the insulation defect in the high-voltage switch cabinet, and an original data set is generated based on the field fault data and the simulation fault data; wherein the insulation defect types include N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects; and the mixed gas is composed of SO2F2, SOF4, SO2, CF4, and CO2.

[0009] In step S2, piecewise cubic spline interpolation (PSCI) is used to perform piecewise cubic spline interpolation fitting on all data in the original data set, a fitting curve is obtained, and data points are extracted from the fitting curve to expand the original data set, thereby obtaining an expanded data set.

[0010] In step S3, a Monte Carlo Markov chain algorithm (MCMC) is introduced into a Gaussian process (GP) to resample the expanded data set, thereby obtaining a sample set F={f1, f2,..., f N} for estimating a posterior distribution.

[0011] In step S4, Bagging bootstrap sampling ensemble learning and decision fusion are introduced into a generalized prediction model to become a Bagging-DF model, feature gas inversion is performed, the Bagging-DF model is trained using the sample set F, the original data set is input into the trained Bagging-DF model for testing, and the insulation defect type in the switch cabinet is obtained.

[0012] In step S41, a Bagging method is used to generate M base classifiers C m .

[0013] In step S42, a Bootstrap sampling method is used to sample the sample set F N1 times with replacement to obtain a training subset Fm with the same length as X, and M training subsets F m , m∈[1,M] are obtained by performing M times of Bootstrap sampling.

[0014] In step S43, the M training subsets F m are used to train M base classifiers C m respectively, thereby obtaining M base classifiers C m with differences.

[0015] In step S44, the M trained base classifiers C m are used to perform prediction analysis on the data in the original data set respectively, thereby obtaining a basic probability assignment set m m of each base classifier C m .

[0016] Step S45, the output result is synthesized by using the Dp synthesis rule to obtain a synthesis result m(A):

[0017]

[0018]

[0019] wherein, m M (B M ) represents the probability of m M subset B M ; K is a conflict coefficient;

[0020] Step S46, the trust degree Bel(A) is calculated:

[0021]

[0022] The label corresponding to the A value with the maximum Bel(A) value is selected as the final output classification, that is, the insulation defect type in the switch cabinet.

[0023] Preferably, the step S1 obtains simulated fault data representing mixed gas components by simulating insulation defects occurring in the high-voltage switch cabinet, including:

[0024] A high-voltage switch cabinet model with three types of internal insulation defects in the cavity is established, and sensors are installed at N2 collection positions in the high-voltage switch cabinet model;

[0025] A single insulation defect type is set for fault simulation, and the mixed gas component concentrations in N directions are collected as a single set of simulated fault data;

[0026] N3, N4, and N5 times of fault simulation are performed for N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects, respectively, to obtain the simulated fault data of each insulation defect type.

[0027] Preferably, the field fault data and the simulated fault data are both gas component data affected by high-voltage switch cabinet internal environmental factors, maintenance state factors, gas collection methods, and equipment factory parameters.

[0028] Preferably, the step S3 includes:

[0029] Step S31, defining a posterior distribution: for a given = {x1, x2,..., x N} and a class label set Y = {y1, y2,..., y N}, n ∈ [1, N], wherein y n represents x nThe posterior distribution p(f|X,y) of the latent function f is defined as:

[0030]

[0031] where p(y|X) approximates the posterior distribution by MCMC method; p(f|X) is the prior distribution of the Gaussian process; p(y|f) is the likelihood of the class label given the latent function value f, which is expressed as:

[0032]

[0033] where σ(f(x)) is a Sigmoid function;

[0034] The MCMC sampling process of p(y|X) for a single element is as follows:

[0035] Initialization: select an element from the extended dataset X as the initial latent function value f (0) ;

[0036] Iterative update: let t represent the current iteration number, t∈[1,T], f (t) represent the new sample obtained after the tth iteration based on the previous sample f (t-1) ; the iteration process is as follows:

[0037] Starting from the current sample f (t-1) , generate a new candidate sample f' ~ N(f (t-1) ,∑) based on the jump distribution, where ∑ is a covariance matrix;

[0038] Calculate the acceptance probability a(f (t) ,f′):

[0039]

[0040] Generate a random number A∈[0,1] following a uniform distribution, and compare a(f (t) ,f′) and A: p(f'|X) is the target distribution probability density of the candidate sample f' under the extended dataset X; q(f (t) |X) represents the probability distribution of f (t) under the condition of given data X;

[0041] If a(f (t) ,f′)≥A, then the candidate sample f' is taken as the output result f (t) of the tth iteration.

[0042] If a(f (t) ,f′)<A, then the candidate sample f' is rejected, and the sample f (t-1)Output result f of the tth iteration (t) ;

[0043] When the maximum iteration number T is reached, the iteration is stopped, and f is obtained (0) Final iteration result f (T) ;

[0044] In step S33, each element in the extended data set X is processed based on the process in step S32, and F = {f1, f2,..., f N} is obtained, where f n is the processing result of x n .

[0045] The application provides a high-voltage switch cabinet characteristic gas inversion partial discharge detection device, which is characterized by implementing the above method.

[0046] According to the technical solution, the high-voltage switch cabinet characteristic gas inversion partial discharge detection method and device are provided. First, an original database is established, field fault data representing the composition of mixed gas in the high-voltage switch cabinet when an insulation defect occurs in the cabinet is obtained, simulation fault data representing the composition of mixed gas is obtained by simulating the insulation defect in the high-voltage switch cabinet, and an original data set is generated based on the field fault data and the simulation fault data. The insulation defect types include N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects. The mixed gas is composed of SO2F2, SOF4, SO2, CF4, and CO2. The piecewise cubic spline interpolation PSCI method is used to perform piecewise cubic spline interpolation fitting on all data in the original data set, a fitting curve is obtained, and data points are extracted from the fitting curve to extend the original data set, and an extended data set is obtained. The Monte Carlo Markov chain algorithm MCMC is introduced into the Gaussian process GP to resample the extended data set, and a sample set F = {f1, f2,..., f N} for estimating the posterior distribution is obtained. The Bagging bootstrap sampling set learning and decision fusion are introduced into the generalized prediction model to become a Bagging-DF model, the characteristic gas inversion is performed, the Bagging-DF model is trained using the sample set F, the original data set is input into the trained Bagging-DF model for testing, and the insulation defect type in the switch cabinet is obtained. The application can quickly and accurately identify the fault defect type in the switch cabinet based on the gas composition data affected by the internal environmental factors, maintenance state factors, collection methods, equipment conditions, and other factors, and improve the diagnosis efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 It is the general technical roadmap of the high-voltage switch cabinet characteristic gas inversion partial discharge detection method of the application.

[0048] Figure 2 Flow chart of the extended data scheme of piecewise cubic spline interpolation of the present application;

[0049] Figure 3 Block diagram for establishing the Bagging-DF model in the present application; DETAILED DESCRIPTION

[0050] The technical solutions and technical effects of the present application are further described in detail below in combination with the accompanying drawings of the present application.

[0051] The purpose of the present application is to design a feature gas inversion partial discharge detection method for high-voltage switchgear based on Gaussian process classification (GPC) and decision fusion (DF), which can effectively improve the generalization ability of diagnosis of different insulation defects (metallic protrusion defects, metallic particle defects, and insulation layer air gap defects) inside the high-voltage switchgear. A piecewise cubic spline interpolation method is used to expand a large amount of laboratory simulation and field fault data accumulated in the early stage, and the expanded data is used as the training data of the fault diagnosis model. On this basis, a Bagging ensemble learning model is introduced in combination with the Gaussian process classification (GPC) method as a base classifier. In the generalized predictive control model, Markov chain Monte Carlo (MCMC) is introduced for sampling, and the Bagging ensemble model is further improved. A variety of evidence theory combination rules are used for integration.

[0052] The field fault data used in the scheme of the present application is fused with the influence of internal environmental factors (such as temperature, humidity, gas mixing, and partial discharge mode) of the high-voltage switchgear, as well as maintenance state factors (operation method of maintenance personnel, inspection means, and equipment condition), gas collection method, equipment condition, and other factors on the measurement results, and the limitation of effective data volume (limited data). Compared with using only simulation data to create original data, it is more realistic, can improve the accuracy of classification and prediction when the model is applied to actual scenes, and effectively reduces the occurrence of misdiagnosis and missed diagnosis.

[0053] REFERENCE Figure 1 As shown in the figure, the present application provides a feature gas inversion partial discharge detection method for a high-voltage switchgear, which comprises the following steps:

[0054] In step S1, field fault data representing the composition of mixed gas in the high-voltage switchgear when an insulation defect occurs in the cabinet is obtained. Simulation fault data representing the composition of mixed gas is obtained by simulating the occurrence of an insulation defect in the high-voltage switchgear. An original data set is generated based on the field fault data and the simulation fault data. The types of insulation defects include N-type metallic protrusion defects, P-type metallic particle defects, and G-type insulation layer air gap defects. The mixed gas is composed of SO2F2, SOF4, SO2, CF4, and CO2.

[0055] Step S2, using piecewise cubic spline interpolation PSCI method, all data in the original data set are fitted by piecewise cubic spline interpolation, and a fitting curve is obtained, and data points are extracted from the fitting curve to expand the original data set, and an expanded data set is obtained;

[0056] Step S3, introducing Monte Carlo Markov chain algorithm MCMC into Gaussian process GP, resampling the expanded data set to obtain a sample set F={f1,f2,...,f N} for estimating the posterior distribution;

[0057] Step S4, introducing Bagging bootstrap sampling ensemble learning and decision fusion into the generalized prediction model to become Bagging-DF model, performing feature gas inversion, training the Bagging-DF model using the sample set F, inputting the original data set into the trained Bagging-DF model for testing, and obtaining the insulation defect type in the switch cabinet.

[0058] Preferably, step S1 obtains simulated fault data representing mixed gas components by simulating insulation defects in the high-voltage switch cabinet, including:

[0059] A high-voltage switch cabinet model with three types of internal insulation defects is established, and N2 sensors are installed at selected collection positions in the high-voltage switch cabinet model, wherein the sensors are not all of the same type and model, and the sensitivity can also differ.

[0060] A single insulation defect type is set for fault simulation, and the mixed gas component concentrations in N directions are collected as a single set of simulated fault data.

[0061] N3, N4, and N5 fault simulations are performed for N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects, respectively, to obtain simulated fault data for each insulation defect type.

[0062] As an optional implementation, the data proportion of the three types of insulation defects in the original data set is 1:1:1.

[0063] The piecewise cubic spline interpolation PSCI method is used as Figure 2 Xnew represents the abscissa of the sampling point on the fitting curve, i is the abscissa of the original data, j represents the jth type of insulation defect, w is the growth feature, and Ynew(w) represents new data generated by the growth feature in the fitting curve of the jth type of insulation defect, to ensure effective training of the model.

[0064] Preferably, step S3, which uses the Gaussian process (GP) to introduce the Monte Carlo Romanov chain algorithm (MCMC) to process the data, is as follows:

[0065] Step S31, define the posterior distribution: for a given x1, x2, ..., x... N} and category label set Y = {y1, y2, ..., y N}, n∈[1,N], where y n x represents n For the corresponding internal insulation defect type, the posterior distribution p(f|X,y) of the latent function f is defined as:

[0066]

[0067] Where p(y|X) approximates the posterior distribution using the MCMC method; p(f|X) is the prior distribution of the Gaussian process, typically assumed to follow a Gaussian process GP(0,k(x,x′)), i.e., zero mean and covariance defined by the kernel function k(x,x′); p(y|f) is the likelihood of the latent function value f given the class label, expressed as:

[0068]

[0069] In the formula, σ(f(x)) is the Sigmoid function;

[0070] Step S32, the MCMC sampling process for a single element of p(y|X) is as follows:

[0071] Initialization: Select an element from the extended dataset X as the initial latent function value f. (0) ;

[0072] Iterative update: Let t represent the current iteration number, t∈[1,T], f (t) Indicates based on the previous sample f (t-1) The new sample obtained after the t-th iteration; the iteration process is as follows:

[0073] From the current sample f (t-1) Starting from this point, new candidate samples f′~N(f) are generated based on the jump distribution. (t-1) ,∑), where Σ is a covariance matrix representing the distribution width of the generated new samples;

[0074] Calculate the acceptance probability α(f) (t) ,f′):

[0075]

[0076] Generate a uniformly distributed random number A∈[0,1], and compare the acceptance probability α(f) with the random number A∈[0,1]. (t), f') and A: p(f'|X) is the target distribution probability density of the candidate sample f' under the extended dataset X; q(f (t) |X) represents the probability distribution of f (t) under the condition of given data X;

[0077] If α(f (t) , f') ≥ A, then take the candidate sample f' as the output result f of the t-th iteration (t) ;

[0078] If α(f (t) , f') < A, then reject the candidate sample f', and take the sample f (t-1) as the output result f of the t-th iteration (t) ;

[0079] When the maximum number of iterations T is reached, stop the iteration and obtain f​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​m The basic probability assignment set m m of each base classifier C m is obtained by performing prediction analysis on the data in the original data set respectively.

[0086] In step S45, the output result is synthesized by using the Dp synthesis rule to obtain a synthesis result m(A):

[0087]

[0088] In the formula, m M (B M ) represents the probability of the subset B M in m M ; K is a conflict coefficient.

[0089] In step S46, the trust degree Bel(A) is calculated:

[0090]

[0091] The label corresponding to the A value with the maximum Bel(A) value is selected as the final output classification, that is, the type of insulation defect in the switch cabinet.

[0092] The application provides a high-voltage switch cabinet feature gas inversion partial discharge detection device, characterized by being used for implementing the method.

[0093] The application has the following technical features:

[0094] 1. The generalized predictive control based on MCMC sampling introduces a piecewise cubic spline interpolation method, and 1200 new data are obtained by extending 352 groups of data accumulated in the early stage. The model training method taking the new data as a training set and the original data as a test set improves the robustness and generalization ability of the air decomposition component fault diagnosis.

[0095] 2、The Monte Carlo Markov chain algorithm (MCMC) is introduced into the Gaussian process (GP) sampling scheme to quantify the error introduced by approximation. For support vector machines, the hinge loss function is used to calculate the empirical risk, and a regularization term is added to the solution system to optimize the structural risk. In the deep belief network (DBN), three RBM units are set and stacked in turn. The optimization algorithm of DBN is the small batch gradient descent method, and the learning rate is 0.8. For the diagnosis problem of defect type in partial discharge fault, the extended data is used as the training set, and it is proved that GPC is the best fault recognition model. Gaussian process classification (GPC): GPC can provide flexible modeling capability when dealing with fault diagnosis problems with uncertainty and complexity. It models the feature gas data through the Gaussian process, which helps to overcome the problem of insufficient data and improve the accuracy of classification. Decision fusion (DF): The decision fusion method can effectively combine the results of multiple classifiers to reduce the misdiagnosis and missed diagnosis problems that may be caused by a single classifier, and enhance the stability and reliability of diagnosis

[0096] 3、Bagging uses Bootstrap sampling to make each base classifier have different prediction performance, solving the overfitting problem that may occur in a single classifier. In the Bagging framework, DF fusion complementary information is introduced to reduce the uncertainty of the model, and redundant information fusion is used to improve the stability of the model, and a Bagging-DF model is constructed. In the Bagging-DF model, the base classifier is a GPC model with MCMC sampling.

[0097] According to the embodiments disclosed by the application, the application also provides an electronic device, a readable storage medium and a computer program product. The electronic device is intended to represent various forms of digital computers, including a computing unit, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) or a computer program loaded into a random access memory (RAM) from a storage unit. In the RAM, various programs and data required for device operation can also be stored. The computing unit, the ROM and the RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0098] The plurality of components in the device are connected to the I / O interface, including an input unit, an output unit, a storage unit, and a communication unit. The communication unit allows the device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunications networks.

[0099] The computing unit can be various general-purpose and / or special-purpose processing components with processing and computing capabilities, which perform various methods and processes described above, such as the GIS sub-millimeter microparticle defect accumulation morphology and development stage recognition method.

[0100] The above-described embodiments are merely preferred embodiments of the present application, and thus are not intended to limit the scope of the present application. Those skilled in the art can understand that all or part of the above-described embodiments can be implemented, and equivalent changes made thereto, without departing from the scope of the present application as defined by the following claims.

Claims

1. A high-voltage switchgear characteristic gas inversion partial discharge detection method, characterized in that, Comprise: Step S1, obtaining field fault data representing the composition of mixed gas in the high-voltage switch cabinet when an insulation defect occurs in the cabinet, obtaining simulated fault data representing the composition of mixed gas by simulating the occurrence of insulation defects in the high-voltage switch cabinet, and generating an original data set based on the field fault data and the simulated fault data; wherein the insulation defect types include N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects; the mixed gas is composed of SO2F2, SOF4, SO2, CF4, and CO2; Step S2, using piecewise cubic spline interpolation PSCI method to perform piecewise cubic spline interpolation fitting on all data in the original data set, obtaining a fitting curve, and extracting data points from the fitting curve to expand the original data set to obtain an expanded data set; Step S3, introducing Monte Carlo Markov chain algorithm MCMC into Gaussian process GP to resample the expanded data set to obtain a sample set F for estimating the posterior distribution; Step S4, introducing Bagging bootstrap sampling ensemble learning and decision fusion into a generalized prediction model to become a Bagging-DF model, performing feature gas inversion, training the Bagging-DF model using the sample set F, inputting the original data set into the trained Bagging-DF model for testing, and obtaining the insulation defect type in the switch cabinet; Step S41, generating M base classifiers C using Bagging method m ; Step S42, using Bootstrap sampling method to sample the sample set F to get a training subset Fm with the same length as X, N1 times with replacement, a total of M times Bootstrap sampling to get M training subsets F m , m ∈ [1, M] Step S43, training each of the training subsets F m Respectively training M base classifiers C m , obtaining M base classifiers C m with differences Step S44, using the trained M base classifiers C m Respectively, the original data set data for predictive analysis, each base classifier C m The basic probability assignment set m m ; Step S45, performing output result synthesis using Dp synthesis rule to obtain a synthesis result m(A): where m M (B M ) denotes the probability of a subset B M of m M subsets; K is a collision factor; Step S46, calculating the trust degree Bel(A): Selecting the label corresponding to the A value with the maximum Bel(A) value as the final output classification, i.e. the insulation defect type in the switch cabinet.

2. The partial discharge detection method of gas inversion of a high voltage switchgear as claimed in claim 1, characterized in that, The step S1 comprises: A high-voltage switch cabinet model with three types of internal insulation defects is established, N2 sensors are installed at selected N2 collection positions in the high-voltage switch cabinet model, and the internal environment of the high-voltage switch cabinet model is set; A single insulation defect type is set for fault simulation, and the concentration of mixed gas components at N positions is collected as a single set of simulated fault data; N3, N4, and N5 fault simulations are performed for N-type metal protrusion defects, P-type metal particle defects, and G-type insulation layer air gap defects, respectively, to obtain the simulated fault data for each insulation defect type.

3. The partial discharge detection method of gas inversion of a high voltage switchgear as claimed in claim 2, characterized in that, The field fault data and the simulated fault data are both gas component data affected by high-voltage switch cabinet internal environment factors, maintenance state factors, gas collection methods, and equipment factory parameters.

4. The partial discharge detection method of gas inversion of a high voltage switchgear as claimed in claim 3, characterized in that, The step S3 comprises: Step S31, define the posterior distribution: for a given x1, x2, ..., x... N } and category label set Y = {y1, y2, ..., y N }, n∈[1,N], where y n x represents n For the corresponding internal insulation defect type, the posterior distribution p(f|X,y) of the latent function f is defined as: Wherein, p(y|X) is used to approximate the posterior distribution by MCMC method; p(f|X) is the prior distribution of Gaussian process; p(y|f) is the likelihood of class label given latent function value f, expressed as: In the formula, sigma(f(x)) is a Sigmoid function; Step S32, the MCMC sampling process of p(y|X) for a single element is: Initialization: Choose an element from the extended dataset X as the initial potential function value f (0) ; Iterative update: Let t denote the current iteration number, t e [1, T], f (t) denotes the previous sample f (t-1) the new sample obtained after the tth iteration; the iteration process is: From the current sample f (t-1) , a new candidate sample f' ~ N(f (t-1) ,∑) is generated based on the jump distribution, where ∑ is a covariance matrix. Compute acceptance probability a(f (t) f') : A uniform random number A ∈ [0, 1] is generated, and the acceptance probability a(f (t) ,f') is compared to A: p(f'|X) is the target distribution probability density of the candidate sample f' under the extended dataset X; q(f (t) |X) represents the probability distribution of f (t) given the data X. If a(f (t) ,f′) ≥ A, then the candidate sample f′ is taken as the output result f (t) of the t-th iteration. If α(f (t) , f′) < A, then reject the candidate sample f′ and take the sample f (t-1) as the output result f of the t-th iteration (t) ; when the maximum number of iterations T is reached, stop the iteration and obtain f (0) as the final iteration result f (T) ; Step S33, process each element in the extended dataset X based on the process of Step S32 to obtain F = {f1, f2,..., f N}, where f n is the processing result of x n .

5. A high-voltage switchgear characteristic gas inversion partial discharge detection device, characterized in that, A method for implementing any one of claims 1-3.

6. An electronic device, comprising: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-4.

7. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are for causing the computer to perform the method of any one of claims 1-4.

8. A computer program product comprising a computer program which, when executed by a processor, implements the method of any one of claims 1-4.

Images