Defect detection model training, GIS device defect detection method and related apparatus

Abstract

This invention discloses a defect monitoring model training method, a GIS equipment defect detection method, and related devices, applied to the field of GIS equipment mechanical defect detection. The method employs an improved adaptive noise complete set empirical mode decomposition to perform feature decomposition on the mechanical vibration signal of GIS equipment under variable frequency current excitation. It then uses normalized mutual information calculation to effectively screen the intrinsic mode functions and reconstruct the GIS equipment mechanical vibration signal. Finally, it extracts the feature matrix of the reconstructed GIS equipment mechanical vibration signal for model training to obtain a defect detection model. This invention achieves feature decomposition and signal reconstruction of the GIS equipment mechanical vibration signal under variable frequency current excitation through mode decomposition and normalized mutual information calculation. By extracting and training the feature of the reconstructed GIS equipment mechanical vibration signal, the defect detection model is trained. Compared with existing technologies that detect GIS mechanical defects based on a single power frequency current, this invention improves defect detection accuracy.

Description

Technical Field

[0001] This invention relates to the field of mechanical defect detection in GIS equipment, and particularly to a defect detection model training method, a GIS equipment defect detection method, a GIS equipment defect detection device, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Gas-insulated switchgear (GIS) possesses unique advantages such as supporting high-voltage, high-capacity power transmission, compact structure, and convenient installation and maintenance, and has become an important component of the power system. Mechanical defects are a significant factor leading to GIS equipment failures; in GIS equipment with voltage levels greater than 126kV, mechanical failures account for 39.3%. Mechanical failures in GIS equipment are diverse, complex, and pose serious risks. Variable frequency excitation vibration analysis can uncover the inherent properties of the mechanical structure, enhancing the structural characterization and mechanical defect detection of GIS equipment. However, current research on mechanical defect detection methods for GIS equipment is mostly based on power frequency current, which cannot reflect the structural frequency response characteristics of GIS equipment, resulting in low accuracy in mechanical defect detection under single-frequency current. Summary of the Invention

[0003] The purpose of this invention is to provide a defect detection model training, a method for detecting defects in GIS equipment, and related devices, which are applied to the field of mechanical defect detection in GIS equipment. This method achieves feature decomposition and signal reconstruction of the mechanical vibration signal of GIS equipment under variable frequency current excitation through modal decomposition and normalized mutual information calculation. By extracting and reconstructing the features of the mechanical vibration signal of GIS equipment, a defect detection model is trained. Compared with the existing technology that detects mechanical defects in GIS based on a single power frequency current, the detection accuracy is improved.

[0004] To address the aforementioned technical problems, this invention provides a defect detection model training method, comprising:

[0005] Acquire mechanical vibration signals of GIS equipment under variable frequency current excitation;

[0006] The mechanical vibration signal of the GIS equipment is subjected to modal decomposition using an improved adaptive noise complete set empirical mode decomposition algorithm to obtain multiple intrinsic mode functions.

[0007] Calculate the normalized mutual information between the mechanical vibration signal of the GIS equipment and the multiple intrinsic mode functions;

[0008] Based on the normalized mutual information, spurious intrinsic mode functions are removed from the multiple intrinsic mode functions to obtain multiple target intrinsic mode functions;

[0009] The mechanical vibration signal of the GIS equipment is obtained by linearly superimposing multiple target intrinsic mode functions;

[0010] Extract features from the mechanical vibration signal of the reconstructed GIS equipment and multiple intrinsic mode functions, and construct a feature matrix based on the features;

[0011] A defect detection model is obtained by training based on the feature matrix.

[0012] Optionally, extracting features from the reconstructed GIS equipment mechanical vibration signal and multiple intrinsic mode functions includes:

[0013] Extract the amplitude, fundamental frequency amplitude, skewness index, kurtosis index, odd-even response ratio, and vibration entropy of the reconstructed GIS equipment mechanical vibration signal;

[0014] Extract the modal energy ratio and modal centroid ratio of multiple intrinsic modal functions.

[0015] Optionally, training the defect detection model based on the feature matrix includes:

[0016] Use the feature matrix as a dataset, and train a random forest model based on the dataset;

[0017] The trained random forest model is then identified as the defect detection model.

[0018] Optionally, training the random forest model based on the dataset includes:

[0019] The random forest model is trained based on the dataset, and the optimal model parameters are found during the training process using the firefly algorithm.

[0020] Optionally, training the defect detection model based on the feature matrix includes:

[0021] The feature matrix is ​​reduced in dimensionality to obtain a dimensionality-reduced feature matrix.

[0022] The defect detection model is obtained by training based on the reduced-dimensional feature matrix.

[0023] Optionally, the step of reducing the dimensionality of the feature matrix to obtain a dimensionality-reduced feature matrix includes:

[0024] The dimensionality-reduced feature matrix is ​​obtained by performing PCA dimensionality reduction on the feature matrix.

[0025] To solve the above-mentioned technical problems, the present invention provides a method for detecting defects in GIS equipment, comprising:

[0026] Acquire mechanical vibration signals of the GIS equipment to be tested;

[0027] Defect detection of GIS equipment is performed based on a defect detection model and the mechanical vibration signal of the GIS equipment to be detected.

[0028] The defect detection model is a model trained according to any defect detection model training method.

[0029] To address the aforementioned technical problems, the present invention provides a defect detection model training device, comprising:

[0030] The first module is used to acquire the mechanical vibration signal of GIS equipment under variable frequency current excitation;

[0031] The second module is used to perform mode decomposition on the mechanical vibration signal of the GIS equipment using an improved adaptive noise complete set empirical mode decomposition algorithm to obtain multiple intrinsic mode functions.

[0032] The third module is used to calculate the normalized mutual information between the mechanical vibration signal of the GIS equipment and the multiple intrinsic mode functions;

[0033] The fourth module is used to remove spurious intrinsic mode functions from the multiple intrinsic mode functions based on the normalized mutual information, so as to obtain multiple target intrinsic mode functions;

[0034] The fifth module is used to linearly superimpose multiple target intrinsic mode functions to obtain the reconstructed mechanical vibration signal of the GIS equipment;

[0035] The sixth module is used to extract features from the mechanical vibration signal of the reconstructed GIS equipment and multiple intrinsic mode functions, and to construct a feature matrix based on the features;

[0036] The seventh module is used to train a defect detection model based on the feature matrix.

[0037] To solve the above-mentioned technical problems, the present invention provides an electronic device, comprising:

[0038] Memory, used to store computer programs;

[0039] A processor, configured to implement any of the defect detection model training methods and / or the GIS equipment defect detection methods when executing the computer program.

[0040] To address the aforementioned technical problems, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement any of the aforementioned defect detection model training methods and / or the GIS equipment defect detection method.

[0041] As can be seen, the method of this invention acquires the mechanical vibration signal of GIS equipment under variable frequency current excitation; performs mode decomposition on the GIS equipment mechanical vibration signal using an improved adaptive noise complete set empirical mode decomposition algorithm to obtain multiple intrinsic mode functions (EMFs); calculates the normalized mutual information between the GIS equipment mechanical vibration signal and the multiple EMFs; removes spurious EMFs from the multiple EMFs based on the normalized mutual information to obtain multiple target EMFs; linearly superimposes the multiple target EMFs to obtain the reconstructed GIS equipment mechanical vibration signal; extracts features from the reconstructed GIS equipment mechanical vibration signal and the multiple EMFs, and constructs a feature matrix based on the features; trains a defect detection model based on the feature matrix, and performs defect detection on the GIS equipment based on the defect detection model and the mechanical vibration signal of the GIS equipment to be detected. Compared with the existing technology that detects GIS mechanical defects based on a single power frequency current, this method improves the detection accuracy. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0043] Figure 1 This is a flowchart of a defect detection model training method provided in an embodiment of the present invention;

[0044] Figure 2 This is a flowchart of a GIS equipment defect detection method provided in an embodiment of the present invention;

[0045] Figure 3 This is a structural block diagram of a defect detection model training device provided in an embodiment of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] The following combination Figure 1 , Figure 1 This is a flowchart of a defect detection model training method provided in an embodiment of the present invention. The method may include:

[0048] S101: Acquire the mechanical vibration signal of GIS equipment under variable frequency current excitation.

[0049] Variable frequency excitation vibration analysis, as an advanced structural characteristic detection method, can uncover the inherent properties of mechanical structures, enhancing the characterization of structural characteristics and the detection of mechanical defects. This embodiment can acquire mechanical vibration signals of GIS equipment under variable frequency current excitation. This embodiment does not limit the frequency range or magnitude of the current. This embodiment can generally collect mechanical vibration signals of GIS equipment under various defect states with voltages of 550kV, current frequencies of 20Hz, 40Hz, 50Hz, 60Hz, 80Hz, and 100Hz, and excitation current magnitudes of 0-5000A. This embodiment does not limit the specific types of defect states; generally, four are most representative: normal, loose busbar base, faulty disconnector spring, and loose support frame bolts. This embodiment can collect mechanical vibration signals of GIS equipment under these four defect states.

[0050] S102: The mechanical vibration signal of the GIS equipment is decomposed into multiple intrinsic mode functions by using an improved adaptive noise complete set empirical mode decomposition algorithm.

[0051] This embodiment can perform mode decomposition on the mechanical vibration signal of GIS equipment using an improved Complete Ensemble EMD with Adaptive Noise (ICEEMDAN) algorithm to obtain multiple Intrinsic Mode Functions (IMFs), thereby realizing the decomposition of the mechanical vibration signal of GIS equipment.

[0052] S103: Calculate the normalized mutual information between the mechanical vibration signal of GIS equipment and multiple intrinsic mode functions.

[0053] S104: Based on normalized mutual information, spurious intrinsic mode functions are removed from multiple intrinsic mode functions to obtain multiple target intrinsic mode functions.

[0054] S105: Linearly superimpose multiple target intrinsic mode functions to obtain the reconstructed mechanical vibration signal of GIS equipment.

[0055] This embodiment can calculate the normalized mutual information (NMI) between the mechanical vibration signal of GIS equipment and multiple intrinsic mode functions (IMFs). NMI characterizes the similarity between two clustering results. This embodiment can use NMI to evaluate the validity of the decomposed IMFs and remove spurious modes. This embodiment does not limit the specific removal method; generally, a threshold can be set to remove IMFs with NMI less than the threshold. Alternatively, the calculated NMIs can be sorted, and IMFs can be removed according to a preset percentage, with the remaining IMFs being the target IMFs.

[0056] In this embodiment, multiple target intrinsic mode functions can be linearly superimposed to obtain the reconstructed mechanical vibration signal of GIS equipment. The reconstructed mechanical vibration signal of GIS equipment eliminates false modes, realizing the modal reconstruction and feature enhancement of the mechanical vibration signal of GIS equipment.

[0057] S106: Extract and reconstruct the features of the mechanical vibration signal and multiple intrinsic mode functions of the GIS equipment, and construct a feature matrix based on the features.

[0058] In this embodiment, features of the mechanical vibration signal of the GIS equipment and multiple intrinsic mode functions can be extracted and reconstructed. The feature types of the signal in this embodiment can include vibration amplitude index, time domain statistical index, frequency domain index and modal domain index. The vibration amplitude index can include amplitude features and fundamental frequency amplitude features. The time domain statistical index can include skewness index features and kurtosis index features. The frequency domain index can include odd-even response ratio features and vibration entropy features. The modal domain index can include modal energy ratio features and modal centroid ratio features.

[0059] This embodiment does not limit the extraction method of each feature. The amplitude can be extracted as follows:

[0060] y max =max(y(n));

[0061] In the formula, y max To reconstruct the amplitude of the mechanical vibration signal of the GIS equipment, y(n) is the time series of the reconstructed mechanical vibration signal of the GIS equipment, and n is the sampling point number.

[0062] The fundamental frequency amplitude can be extracted as follows:

[0063]

[0064] In the formula, A base To reconstruct the fundamental frequency amplitude of the mechanical vibration signal of GIS equipment, p(f) is the frequency domain sequence obtained by Fourier transform of the reconstructed vibration signal, where f represents the frequency.e The frequency of the excitation current.

[0065] The skewness index can be extracted as follows:

[0066]

[0067] In the formula, S is the skewness index of the reconstructed mechanical vibration signal of the GIS equipment, and N is the number of sampling points. For, σ y To reconstruct the mean of the time series of mechanical vibration signals from GIS equipment.

[0068] The kurtosis index can be extracted as follows:

[0069]

[0070] In the formula, K is the kurtosis index of the reconstructed mechanical vibration signal of GIS equipment.

[0071] The even-order response ratio can be extracted using the following formula:

[0072]

[0073] In the formula, R is the even-order response ratio of the reconstructed mechanical vibration signal of the GIS equipment, and f max This is the highest frequency in the mechanical vibration signal spectrum of GIS equipment.

[0074] The vibration entropy can be extracted in the following way:

[0075]

[0076] In the formula, H is the vibration entropy of the reconstructed mechanical vibration signal of the GIS equipment.

[0077] This embodiment can extract the mode energy ratio from multiple intrinsic mode functions. The extraction method can be as follows:

[0078]

[0079] In the formula, E harm p is the modal energy ratio. IMFk (f) is the frequency domain sequence of the k-th mode IMFk among multiple intrinsic mode functions, preferably k is 2.

[0080] The modal centroid ratio is extracted as follows:

[0081]

[0082] In the formula, G harm This represents the modal centroid ratio.

[0083] This embodiment can construct a feature matrix based on the characteristics of the mechanical vibration signal of GIS equipment and multiple intrinsic mode functions under different frequency conversion current excitation. This embodiment does not limit the specific form of the feature matrix. The columns of the matrix can be current frequencies, and the rows of the matrix can be feature types.

[0084] S107: A defect detection model is obtained by training based on the feature matrix.

[0085] This embodiment can train a defect detection model based on the feature matrix. The specific training method for the defect detection model is not limited. In this embodiment, the feature matrix can first undergo dimensionality reduction processing to reduce the data's dimensionality and improve the accuracy of the trained defect model. This embodiment does not limit the specific method of data dimensionality reduction. Generally, Principal Component Analysis (PCA) can be used to reduce the dimensionality of the feature matrix, resulting in a PCA dimensionality-reduced matrix. Orthogonal transformations are then used to project correlated high-dimensional variables into a low-dimensional space, thereby establishing a new set of linearly independent feature variables. This approach minimizes information loss while obtaining a dimensionality-reduced optimized feature matrix with better representativeness and less overlap between dimensions.

[0086] This embodiment does not limit the type of defect detection model. Generally, it can be a random forest model. In this embodiment, the feature matrix of each defect state can be used as a dataset to train the random forest model. The trained random forest model is then determined as the defect detection model. Furthermore, in order to improve the accuracy of the defect detection model obtained through training, this embodiment can introduce the firefly optimization algorithm during the training process to optimize the parameters and find the optimal number of decision trees and the maximum number of features in the random forest model, thereby improving the accuracy of the model.

[0087] This invention achieves feature decomposition and signal reconstruction of mechanical vibration signals of GIS equipment under variable frequency current excitation through modal decomposition and normalized mutual information calculation. By extracting and reconstructing the features of the mechanical vibration signals of GIS equipment, a defect detection model is trained. Compared with the existing technology that detects mechanical defects of GIS based on a single power frequency current, the detection accuracy is improved.

[0088] The following combination Figure 2 , Figure 2 This is a flowchart of a GIS equipment defect detection method provided in an embodiment of the present invention. The method may include:

[0089] S201: Acquire the mechanical vibration signal of the GIS equipment to be tested;

[0090] S202: Defect detection of GIS equipment based on defect detection model and mechanical vibration signal of GIS equipment to be tested;

[0091] Among them, the defect detection model is a model trained according to any defect detection model training method.

[0092] This embodiment can acquire the mechanical vibration signal of the GIS equipment to be detected, and perform defect detection of the GIS equipment based on the defect detection model and the mechanical vibration signal of the GIS equipment to be detected. This embodiment can extract the feature matrix to be detected from the mechanical vibration signal of the GIS equipment under variable frequency current excitation in the process of training the defect detection model. The feature matrix to be detected is then input into the defect detection model trained according to any defect detection model training method to perform defect detection and obtain the defect detection result.

[0093] The following is a specific embodiment provided by the present invention, wherein the GIS equipment is a 550kV full-scale GIS equipment, and this specific embodiment may include:

[0094] The excitation current frequency is set to 20Hz, 40Hz, 50Hz, 60Hz, 80Hz and 100Hz, and the excitation current magnitude is 0-5000A;

[0095] The mechanical vibration signals of GIS equipment with six frequencies of the same current magnitude are taken as one sample, and a source domain sample set containing 400 samples is finally obtained.

[0096] Four defect states were selected: normal, loose busbar base, faulty disconnector spring, and loose support bolts. Each defect state contained 100 samples; among them, 20 samples were selected for each of the following currents: 1000A, 2000A, 3000A, 4000A, and 5000A.

[0097] Each sample is decomposed using ICEEMDAN to obtain multiple IMFs. The spurious modes in the IMFs are removed based on the normalized mutual information method, and the remaining target IMFs are linearly superimposed to obtain the reconstructed sample.

[0098] The amplitude, fundamental frequency amplitude, skewness index, kurtosis index, odd-even response ratio and vibration entropy are extracted from each reconstructed sample. The modal energy ratio and modal centroid ratio of multiple IMFs are extracted, and a feature matrix is ​​constructed based on the extracted features.

[0099] PCA is used to reduce the dimensionality of the feature matrix to obtain the dimensionality-reduced feature matrix. The dimensionality-reduced feature matrix is ​​used as the dataset to train the random forest model. During the training process, the firefly algorithm is used to optimize the parameters.

[0100] The trained random forest model is used as the defect detection model. The mechanical vibration signal of the GIS equipment to be detected is obtained, and the defect detection of the GIS equipment is performed based on the defect detection model and the mechanical vibration signal of the GIS equipment to be detected.

[0101] The following combination Figure 3 , Figure 3 This is a structural block diagram of a defect detection model training device provided in an embodiment of the present invention. The device may include:

[0102] The first module 100 is used to acquire the mechanical vibration signal of GIS equipment under variable frequency current excitation;

[0103] The second module 200 is used to perform mode decomposition on the mechanical vibration signal of the GIS equipment using an improved adaptive noise complete set empirical mode decomposition algorithm to obtain multiple intrinsic mode functions.

[0104] The third module 300 is used to calculate the normalized mutual information between the mechanical vibration signal of the GIS equipment and the multiple intrinsic mode functions;

[0105] The fourth module 400 is used to remove spurious intrinsic mode functions from the multiple intrinsic mode functions based on the normalized mutual information, so as to obtain multiple target intrinsic mode functions;

[0106] The fifth module 500 is used to linearly superimpose multiple target intrinsic mode functions to obtain the reconstructed mechanical vibration signal of the GIS equipment;

[0107] The sixth module 600 is used to extract features of the mechanical vibration signal of the reconstructed GIS equipment and multiple intrinsic mode functions, and construct a feature matrix based on the features;

[0108] The seventh module 700 is used to train a defect detection model based on the feature matrix.

[0109] Based on the above embodiments, the method of the present invention realizes feature decomposition and signal reconstruction of the mechanical vibration signal of GIS equipment under variable frequency current excitation through modal decomposition and normalized mutual information calculation. By extracting and reconstructing the features of the mechanical vibration signal of GIS equipment, a defect detection model is trained. Compared with the existing technology that detects mechanical defects of GIS based on a single power frequency current, the detection accuracy is improved.

[0110] Based on the above embodiments, the sixth module 600 may include:

[0111] The first unit is used to extract the amplitude, fundamental frequency amplitude, skewness index, kurtosis index, odd-even response ratio, and vibration entropy of the mechanical vibration signal of the reconstructed GIS equipment.

[0112] The second unit is used to extract the modal energy ratio and modal centroid ratio of multiple intrinsic modal functions.

[0113] Based on the above embodiments, the seventh module 700 may include:

[0114] The third unit is used to use the feature matrix as a dataset and train a random forest model based on the dataset.

[0115] The fourth unit is used to determine the trained random forest model as the defect detection model.

[0116] Based on the above embodiments, the third unit may include:

[0117] The first subunit trains the random forest model using the dataset and finds the optimal model parameters during the training process using the firefly algorithm.

[0118] Based on the above embodiments, the seventh module 700 may include:

[0119] The fifth unit is used to perform data dimensionality reduction on the feature matrix to obtain a dimensionality-reduced feature matrix;

[0120] The sixth unit is used to train the defect detection model based on the reduced-dimensional feature matrix.

[0121] Based on the above embodiments, the fifth unit includes:

[0122] The second sub-unit is used to perform PCA dimensionality reduction on the feature matrix to obtain the dimensionality-reduced feature matrix.

[0123] The following is an embodiment of a GIS equipment defect detection device provided by the present invention, which may include:

[0124] The eighth module is used to acquire the mechanical vibration signal of the GIS equipment to be tested;

[0125] The ninth module is used to perform defect detection of GIS equipment based on the defect detection model and the mechanical vibration signal of the GIS equipment to be detected;

[0126] The defect detection model is a model trained using any defect detection model training device.

[0127] Based on the above embodiments, the present invention also provides an electronic device, which may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments. Of course, the device may also include various necessary network interfaces, a power supply, and other components.

[0128] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by an execution terminal or processor, can implement the method provided in the embodiments of the present invention; the storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0129] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any such actual relationship or order between these entities or operations. Furthermore, 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 a process, method, article, or apparatus. Without further limitation, 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 said element.

[0130] The above provides a detailed description of the defect detection model training, GIS equipment defect detection method, and related apparatus provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A defect detection model training method, characterized in that, include: Acquire mechanical vibration signals of GIS equipment under variable frequency current excitation; The mechanical vibration signal of the GIS equipment is subjected to modal decomposition using an improved adaptive noise complete set empirical mode decomposition algorithm to obtain multiple intrinsic mode functions. Calculate the normalized mutual information between the mechanical vibration signal of the GIS equipment and the multiple intrinsic mode functions; Based on the normalized mutual information, spurious intrinsic mode functions are removed from the multiple intrinsic mode functions to obtain multiple target intrinsic mode functions; The mechanical vibration signal of the GIS equipment is obtained by linearly superimposing multiple target intrinsic mode functions; Extract features from the mechanical vibration signal of the reconstructed GIS equipment and multiple intrinsic mode functions, and construct a feature matrix based on the features; A defect detection model is trained based on the aforementioned feature matrix; The extraction of features from the reconstructed GIS equipment mechanical vibration signal and multiple intrinsic mode functions includes: Extract the amplitude, fundamental frequency amplitude, skewness index, kurtosis index, odd-even response ratio, and vibration entropy of the reconstructed GIS equipment mechanical vibration signal; Extract the modal energy ratio and modal centroid ratio of multiple intrinsic modal functions; The method for extracting the amplitude is as follows: ； In the formula, The amplitude of the reconstructed mechanical vibration signal of the GIS equipment. The time series of the reconstructed mechanical vibration signal of the GIS equipment. The sampling point number; The method for extracting the fundamental frequency amplitude is as follows: ； In the formula, The fundamental frequency amplitude of the reconstructed mechanical vibration signal of the GIS equipment. The frequency domain sequence obtained by Fourier transforming the reconstructed vibration signal is... Indicates frequency, The frequency of the excitation current; The even-order response ratio is extracted as follows: ； In the formula, The even-order response ratio of the reconstructed GIS equipment mechanical vibration signal. The highest frequency in the mechanical vibration signal spectrum of the GIS equipment; The method for extracting the vibration entropy is as follows: ； In the formula, The vibration entropy is the mechanical vibration signal of the reconstructed GIS equipment. The modal energy ratio is extracted as follows: ； In the formula, The modal energy ratio is... For the first of the multiple intrinsic mode functions First mode The frequency domain sequence; The modal centroid ratio is extracted as follows: ； In the formula, The modal centroid ratio is given.

2. The defect detection model training method according to claim 1, characterized in that, The defect detection model trained based on the feature matrix includes: Use the feature matrix as a dataset, and train a random forest model based on the dataset; The trained random forest model is then identified as the defect detection model.

3. The defect detection model training method according to claim 2, characterized in that, Training a random forest model based on the dataset includes: The random forest model is trained based on the dataset, and the optimal model parameters are found during the training process using the firefly algorithm.

4. The defect detection model training method according to claim 1, characterized in that, The defect detection model trained based on the feature matrix includes: The feature matrix is ​​reduced in dimensionality to obtain a dimensionality-reduced feature matrix. The defect detection model is obtained by training based on the reduced-dimensional feature matrix.

5. The defect detection model training method according to claim 4, characterized in that, The step of reducing the dimensionality of the feature matrix to obtain a dimensionality-reduced feature matrix includes: The dimensionality-reduced feature matrix is ​​obtained by performing PCA dimensionality reduction on the feature matrix.

6. A method for detecting defects in GIS equipment, characterized in that, include: Acquire mechanical vibration signals of the GIS equipment to be tested; Defect detection of GIS equipment is performed based on a defect detection model and the mechanical vibration signal of the GIS equipment to be detected. The defect detection model is a model trained according to any one of the defect detection model training methods in claims 1 to 5.

7. A defect detection model training device, characterized in that, include: The first module is used to acquire the mechanical vibration signal of GIS equipment under variable frequency current excitation; The second module is used to perform mode decomposition on the mechanical vibration signal of the GIS equipment using an improved adaptive noise complete set empirical mode decomposition algorithm to obtain multiple intrinsic mode functions. The third module is used to calculate the normalized mutual information between the mechanical vibration signal of the GIS equipment and the multiple intrinsic mode functions; The fourth module is used to remove spurious intrinsic mode functions from the multiple intrinsic mode functions based on the normalized mutual information, so as to obtain multiple target intrinsic mode functions; The fifth module is used to linearly superimpose multiple target intrinsic mode functions to obtain the reconstructed mechanical vibration signal of the GIS equipment; The sixth module is used to extract features from the mechanical vibration signal of the reconstructed GIS equipment and multiple intrinsic mode functions, and to construct a feature matrix based on the features; The seventh module is used to train a defect detection model based on the feature matrix; The extraction of features from the reconstructed GIS equipment mechanical vibration signal and multiple intrinsic mode functions includes: Extract the amplitude, fundamental frequency amplitude, skewness index, kurtosis index, odd-even response ratio, and vibration entropy of the reconstructed GIS equipment mechanical vibration signal; Extract the modal energy ratio and modal centroid ratio of multiple intrinsic modal functions; The method for extracting the amplitude is as follows: ； In the formula, The amplitude of the reconstructed mechanical vibration signal of the GIS equipment. The time series of the reconstructed mechanical vibration signal of the GIS equipment. The sampling point number; The method for extracting the fundamental frequency amplitude is as follows: ； In the formula, The fundamental frequency amplitude of the reconstructed mechanical vibration signal of the GIS equipment. The frequency domain sequence obtained by Fourier transforming the reconstructed vibration signal is... Indicates frequency, The frequency of the excitation current; The even-order response ratio is extracted as follows: ； In the formula, The even-order response ratio of the reconstructed GIS equipment mechanical vibration signal. The highest frequency in the mechanical vibration signal spectrum of the GIS equipment; The method for extracting the vibration entropy is as follows: ； In the formula, The vibration entropy is the mechanical vibration signal of the reconstructed GIS equipment. The modal energy ratio is extracted as follows: ； In the formula, The modal energy ratio is... For the first of the multiple intrinsic mode functions First mode The frequency domain sequence; The modal centroid ratio is extracted as follows: ； In the formula, The modal centroid ratio is given.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the defect detection model training method as described in any one of claims 1 to 5, and / or the GIS equipment defect detection method as described in claim 6, when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, implement the defect detection model training method as described in any one of claims 1 to 5, and / or the GIS equipment defect detection method as described in claim 6.

Images