Insulating oil ultrasonic feature extraction method and device

Abstract

The application discloses an insulating oil ultrasonic feature extraction method and device, and the method comprises the following steps: collecting insulating oil ultrasonic signal data sets, calculating insulating oil ultrasonic signal data in each category of the insulating oil ultrasonic signal data sets, obtaining a first divergence matrix and a second divergence matrix of the insulating oil ultrasonic signal data sets, obtaining a third matrix by using the first divergence matrix and the second divergence matrix, performing feature decomposition on the third matrix, obtaining a plurality of characteristic values and a characteristic vector corresponding to the characteristic values, then calculating AIC values of each category, selecting a characteristic vector with an AIC value less than a second preset value as an optimal characteristic vector, and obtaining optimal features of the insulating oil ultrasonic signal data sets by using the optimal characteristic vector. The application extracts features from insulating oil ultrasonic data, improves the feature extraction result accuracy, and further improves the accuracy of subsequent application of ultrasonic data for oil state diagnosis.

Description

Technical Field

[0001] This invention relates to the field of ultrasonic data feature extraction technology, and in particular to a method and apparatus for ultrasonic feature extraction of insulating oil. Background Technology

[0002] Ultrasonic testing is a technique that studies the relationship between the acoustic parameters of the tested medium and the tested object to obtain the physicochemical properties of the medium. Due to its non-destructive and rapid characteristics, this technique has been widely used in the structural testing and performance analysis of solid and liquid materials.

[0003] However, when using ultrasonic technology to detect substances, transformer oil is an extremely complex mixture, and its actual ultrasonic measurement is the result of the superposition of multiple signals. Therefore, the ultrasonic data of transformer oil contains both useful information characterizing the aging state of the transformer and invalid information such as background interference and noise signals. This invalid information can affect the results of subsequent diagnostics of the transformer oil condition using ultrasonic data. Therefore, the accuracy of directly using ultrasound for oil condition diagnosis is difficult to meet the needs of practical engineering applications. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and apparatus for ultrasonic feature extraction of insulating oil, which can improve the accuracy of feature extraction results from ultrasonic data of insulating oil, thereby improving the accuracy of subsequent application of ultrasonic data for oil condition diagnosis.

[0005] A first aspect of this invention provides a method for extracting ultrasonic features from insulating oil, the method comprising:

[0006] Collect an ultrasonic signal dataset of insulating oil, which includes ultrasonic signals of insulating oil and the corresponding categories of the ultrasonic signal data of insulating oil;

[0007] The intra-class and inter-class divergence matrices of the insulating oil ultrasonic signal data in each category are calculated to obtain the first and second divergence matrices of the insulating oil ultrasonic signal dataset. The third matrix is ​​obtained using the first and second divergence matrices.

[0008] The third matrix is ​​decomposed to obtain multiple eigenvalues ​​and corresponding eigenvectors. The eigenvectors corresponding to eigenvalues ​​greater than the first preset value are calculated to obtain the AIC value of the eigenvectors. The eigenvectors with AIC values ​​less than the second preset value are selected as the optimal eigenvectors. The feature matrix is ​​constructed based on the optimal eigenvectors. The optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix.

[0009] In one possible implementation of the first aspect, intra-class and inter-class scatter matrices are calculated using ultrasonic signal data of insulating oil from each category to obtain a first and second scatter matrix of the ultrasonic signal dataset of insulating oil, including:

[0010] The first divergence matrix is ​​obtained by using the formula for calculating the intra-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the intra-class divergence matrix is ​​as follows:

[0011]

[0012] In the formula, S b Let N represent the first divergence matrix. j U represents the number of ultrasonic signal data of the j-th type of insulating oil. j is the mean value of ultrasonic signal data of the j-th type of insulating oil, and u represents the average value of ultrasonic signal data of all insulating oils;

[0013] The second divergence matrix is ​​obtained by using the formula for calculating the inter-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the inter-class divergence matrix is ​​as follows:

[0014]

[0015] In the formula, S w Let X represent the second divergence matrix. j U represents the number of ultrasonic signal data for the j-th type of insulating oil. j is the mean value of the ultrasonic signal data of the j-th type of insulating oil, and k represents the number of types.

[0016] In one possible implementation of the first aspect, a third matrix is ​​obtained using the first and second divergence matrices, including:

[0017] The third matrix is ​​obtained by calculating using the first and second divergence matrices, where the calculation formula is:

[0018]

[0019] In the formula, S b Let S represent the first divergence matrix. w Let Z represent the second divergence matrix, and Z represent the third matrix.

[0020] In one possible implementation of the first aspect, the third matrix is ​​eigenvalued to obtain multiple eigenvalues ​​and corresponding eigenvectors. The eigenvectors corresponding to eigenvalues ​​greater than a first preset value are then calculated to obtain the AIC value of the eigenvectors, including:

[0021] Perform eigenvalue decomposition on the third matrix to obtain multiple eigenvalues ​​and their corresponding eigenvectors;

[0022] The AIC value of each eigenvector is calculated according to the AIC criterion, where the formula for calculating the AIC criterion is as follows:

[0023]

[0024] In the formula, K (K≤d) is the dimension of the dimensionality reduction matrix, N is the total number of categories, and D max D represents the maximum error between ultrasonic signal data of insulating oil across all categories. min This represents the minimum error between ultrasonic signal data for insulating oil across all categories.

[0025] In one possible implementation of the first aspect, a feature matrix is ​​constructed based on the optimal feature vector, and the optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix, including:

[0026] A feature matrix is ​​constructed based on the optimal feature vector, and the optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix. The expression for the optimal features is:

[0027]

[0028] in, Let X represent the feature matrix, where X represents the ultrasonic signal data of insulating oil for all categories, and Y represents the optimal feature.

[0029] A second aspect of the present invention provides an ultrasonic feature extraction device for insulating oil, the device comprising:

[0030] The acquisition module is used to acquire the ultrasonic signal dataset of insulating oil, which includes the ultrasonic signals of insulating oil and the corresponding categories of the ultrasonic signal data.

[0031] The decomposition module is used to calculate the intra-class and inter-class scatter matrices using the ultrasonic signal data of insulating oil in each category, to obtain the first and second scatter matrices of the ultrasonic signal dataset of insulating oil, and to obtain the third matrix using the first and second scatter matrices.

[0032] The extraction module is used to perform eigenvalue decomposition on the third matrix to obtain multiple eigenvalues ​​and corresponding eigenvectors. The eigenvectors corresponding to eigenvalues ​​greater than the first preset value are calculated to obtain the AIC value of the eigenvectors. The eigenvectors with AIC values ​​less than the second preset value are selected as the optimal eigenvectors. The feature matrix is ​​constructed based on the optimal eigenvectors. The optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix.

[0033] In one possible implementation of the second aspect, the intra-class and inter-class scatter matrices of the insulating oil ultrasonic signal data from each category are calculated to obtain the first and second scatter matrices of the insulating oil ultrasonic signal dataset, including:

[0034] The first divergence matrix is ​​obtained by using the formula for calculating the intra-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the intra-class divergence matrix is ​​as follows:

[0035]

[0036] In the formula, S b Let N represent the first divergence matrix. j U represents the number of ultrasonic signal data of the j-th type of insulating oil. j is the mean value of ultrasonic signal data of the j-th type of insulating oil, and u represents the average value of ultrasonic signal data of all insulating oils;

[0037] The second divergence matrix is ​​obtained by using the formula for calculating the inter-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the inter-class divergence matrix is ​​as follows:

[0038]

[0039] In the formula, S w Let X represent the second divergence matrix. j U represents the number of ultrasonic signal data for the j-th type of insulating oil. j is the mean value of the ultrasonic signal data of the j-th type of insulating oil, and k represents the number of types.

[0040] In one possible implementation of the second aspect, a third matrix is ​​obtained using the first and second divergence matrices, including:

[0041] The third matrix is ​​obtained by calculating using the first and second divergence matrices, where the calculation formula is:

[0042]

[0043] In the formula, S b Let S represent the first divergence matrix. w Let Z represent the second divergence matrix, and Z represent the third matrix.

[0044] A third aspect of the present invention provides a computer device, comprising:

[0045] Memory, used to store computer programs;

[0046] A processor for executing computer programs to implement methods such as the ultrasonic feature extraction method for insulating oil as described in the first aspect.

[0047] A fourth aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the ultrasonic feature extraction method for insulating oil as described in the first aspect.

[0048] This invention collects an ultrasonic signal dataset of insulating oil, which includes ultrasonic signals of insulating oil and their corresponding categories. The dataset is used to calculate intra-class and inter-class divergence matrices of the ultrasonic signal data within each category, resulting in a first and second divergence matrix. A third matrix is ​​then obtained using the first and second divergence matrices. Feature decomposition is performed on the third matrix to obtain multiple eigenvalues ​​and their corresponding eigenvectors. The eigenvectors corresponding to eigenvalues ​​greater than a first preset value are calculated to obtain their AIC values. Eigenvectors with AIC values ​​less than a second preset value are selected as the optimal eigenvectors. A feature matrix is ​​constructed based on these optimal eigenvectors, and the optimal features of the ultrasonic signal dataset are obtained using this feature matrix. This invention improves the accuracy of feature extraction results by extracting features from the ultrasonic data of insulating oil, thereby improving the accuracy of subsequent applications of ultrasonic data for oil condition diagnosis. Attached Figure Description

[0049] Figure 1 : A flowchart illustrating the configuration method of one embodiment of the ultrasonic feature extraction method for insulating oil provided by the present invention;

[0050] Figure 2 : Schematic diagram of original ultrasonic images of different oil states in one embodiment of the ultrasonic feature extraction method for insulating oil provided by the present invention;

[0051] Figure 3 : A schematic diagram of the AIC values ​​of different dimensions of the dimensionality reduction matrix of an embodiment of the ultrasonic feature extraction method for insulating oil provided by the present invention;

[0052] Figure 4 : A block diagram of an embodiment of the ultrasonic feature extraction method for insulating oil provided by the present invention. Detailed Implementation

[0053] 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.

[0054] Ultrasonic analysis of transformer oil contains both useful information characterizing the aging state of the transformer and invalid information such as background interference and noise signals. This invalid information can affect the results of subsequent diagnostic analyses of the transformer oil condition using ultrasonic data. Therefore, the accuracy of directly using raw ultrasonic signals for oil condition diagnosis is insufficient for practical engineering applications.

[0055] Please refer to Figure 1 This is a flowchart illustrating an embodiment of the ultrasonic feature extraction method for insulating oil provided by the present invention, including steps S101 to S103, each step of which is as follows:

[0056] S101. Collect the ultrasonic signal dataset of insulating oil, wherein the ultrasonic signal dataset of insulating oil includes the ultrasonic signal of insulating oil and the corresponding category of the ultrasonic signal data of insulating oil.

[0057] In this embodiment, the ultrasonic data dataset of the insulating oil sample is collected using an ultrasonic detector. The ultrasonic signal dataset of the insulating oil includes the ultrasonic signals of the insulating oil and the corresponding categories of the ultrasonic signal data. The expression of the ultrasonic data dataset is as follows:

[0058] D = (x1, y1), (x2, y2), ..., (x m ,y m )

[0059] In the formula, x m For n-dimensional ultrasound data, y m For the corresponding category.

[0060] It is worth noting that ultrasonic data is obtained by using an ultrasonic probe to emit ultrasonic waves and simultaneously receive ultrasonic signals that are reflected, refracted, and diffracted back, and then converting these sound waves into electrical frequency signals.

[0061] S102. Calculate the intra-class and inter-class scatter matrices using the ultrasonic signal data of insulating oil in each category to obtain the first and second scatter matrices of the ultrasonic signal dataset of insulating oil. Use the first and second scatter matrices to obtain the third matrix.

[0062] It is worth noting that the first scatter matrix refers to the intra-class scatter matrix, the second scatter matrix refers to the inter-class scatter matrix, and the third matrix refers to the matrix calculated from the intra-class scatter matrix and the inter-class scatter matrix.

[0063] In this embodiment, the intra-class divergence matrix and the inter-class divergence matrix are first calculated using the ultrasonic signal data of insulating oil in each category, and then the third matrix is ​​calculated using the inverse of the intra-class divergence matrix and the inter-class divergence matrix.

[0064] In some embodiments, step S102 includes, but is not limited to, the following steps:

[0065] The first divergence matrix is ​​obtained by using the formula for calculating the intra-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the intra-class divergence matrix is ​​as follows:

[0066]

[0067] In the formula, S b Let N represent the first divergence matrix. j U represents the number of ultrasonic signal data of the j-th type of insulating oil. j is the mean value of ultrasonic signal data of the j-th type of insulating oil, and u represents the average value of ultrasonic signal data of all insulating oils;

[0068] The second divergence matrix is ​​obtained by using the formula for calculating the inter-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the inter-class divergence matrix is ​​as follows:

[0069]

[0070] In the formula, S w Let X represent the second divergence matrix. j U represents the number of ultrasonic signal data for the j-th type of insulating oil. j is the mean value of the ultrasonic signal data of the j-th type of insulating oil, and k represents the number of types.

[0071] Specifically, the intra-class scatter matrix of the insulating oil ultrasonic signal dataset is calculated using the intra-class scatter matrix calculation formula and the insulating oil ultrasonic signal data in each category. The intra-class scatter matrix calculation formula is as follows:

[0072]

[0073] In the formula, S b N represents the within-class scatter matrix. j U represents the number of ultrasonic signal data of the j-th type of insulating oil. j is the mean value of ultrasonic signal data of the j-th type of insulating oil, and u represents the average value of ultrasonic signal data of all insulating oils;

[0074] The inter-class divergence matrix of the insulating oil ultrasonic signal dataset is obtained by using the formula for calculating the inter-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the inter-class divergence matrix is ​​as follows:

[0075]

[0076] In the formula, S w Let X represent the between-class scatter matrix. jU represents the number of ultrasonic signal data for the j-th type of insulating oil. j is the mean value of the ultrasonic signal data of the j-th type of insulating oil, and k represents the number of types.

[0077] In some embodiments, step S102 includes, but is not limited to, the following steps:

[0078] The third matrix is ​​obtained by calculating using the first and second divergence matrices, where the calculation formula is:

[0079]

[0080] In the formula, S b Let S represent the first divergence matrix. w Let Z represent the second divergence matrix, and Z represent the third matrix.

[0081] Specifically, the third matrix is ​​calculated using the inverses of the intra-class scatter matrix and the inter-class scatter matrix. The calculation formula is as follows:

[0082]

[0083] In the formula, S b Let S represent the first divergence matrix. w Let Z represent the second divergence matrix, and Z represent the third matrix.

[0084] S103: Perform eigenvalue decomposition on the third matrix to obtain multiple eigenvalues ​​and corresponding eigenvectors. Calculate the eigenvectors corresponding to eigenvalues ​​greater than the first preset value to obtain the AIC value of the eigenvectors. Select the eigenvectors with AIC values ​​less than the second preset value as the optimal eigenvectors. Construct a feature matrix based on the optimal eigenvectors. Use the feature matrix to obtain the optimal features of the insulating oil ultrasonic signal dataset.

[0085] Specifically, the third matrix is ​​decomposed to obtain multiple eigenvalues ​​and corresponding eigenvectors. A dimensionality reduction matrix is ​​formed by combining the eigenvectors corresponding to a preset number of eigenvalues ​​greater than the first preset value. Then, the AIC value of each dimensionality reduction matrix is ​​calculated according to the AIC criterion. The eigenvalues ​​corresponding to the AIC values ​​less than the second preset value are selected as the optimal eigenvalues. The eigenvectors corresponding to the optimal eigenvalues ​​are used to obtain the feature matrix. The optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix.

[0086] It should be noted that the preset quantity can be set according to actual needs, and the first and second preset values ​​can be set according to actual needs.

[0087] In some embodiments, step S103 includes, but is not limited to, the following steps:

[0088] Perform eigenvalue decomposition on the third matrix to obtain multiple eigenvalues ​​and their corresponding eigenvectors;

[0089] The AIC value of each eigenvector is calculated according to the AIC criterion, where the formula for calculating the AIC criterion is as follows:

[0090]

[0091] In the formula, K (K≤d) is the dimension of the dimensionality reduction matrix, N is the total number of categories, and D max D represents the maximum error between ultrasonic signal data of insulating oil across all categories. min This represents the minimum error between ultrasonic signal data for insulating oil across all categories.

[0092] Specifically, the third matrix is ​​subjected to eigenvalue decomposition to obtain multiple eigenvalues ​​and their corresponding eigenvectors. A predetermined number of eigenvectors corresponding to eigenvalues ​​greater than a first predetermined value are used to form a dimensionality reduction matrix. The AIC value of each dimensionality reduction matrix is ​​calculated according to the AIC criterion. Then, the eigenvector with an AIC value less than a second predetermined value is selected as the optimal eigenvector. The formula for calculating the AIC criterion is as follows:

[0093]

[0094] In the formula, K (K≤d) is the dimension of the dimensionality reduction matrix, N is the total number of categories, and D max D represents the maximum error between ultrasonic signal data of insulating oil across all categories. min This represents the minimum error between ultrasonic signal data for insulating oil across all categories.

[0095] It should be noted that the preset values ​​can be set according to actual needs.

[0096] In some embodiments, step S103 includes, but is not limited to, the following steps:

[0097] A feature matrix is ​​constructed based on the optimal eigenvectors. The optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix. The expression for the optimal features is:

[0098]

[0099] in, Let X represent the feature matrix, where X represents the ultrasonic signal data of insulating oil for all categories, and Y represents the optimal feature.

[0100] Specifically, the eigenvalue corresponding to the lowest AIC value is selected as the optimal eigenvalue. The eigenvector corresponding to this optimal eigenvalue is used to obtain the feature matrix. The feature matrix is ​​then used to obtain the optimal features of the insulating oil ultrasonic signal dataset. The expression for the optimal feature is:

[0101]

[0102] in, Let X represent the feature matrix, where X represents the ultrasonic signal data of insulating oil for all categories, and Y represents the optimal feature.

[0103] As an example of this embodiment, the ultrasonic feature extraction method for insulating oil is further illustrated below. Oil samples with different aging times were prepared through accelerated thermal aging tests. The raw ultrasonic data of the transformer oil were obtained by detecting the oil samples using an ultrasonic testing platform. To verify the effectiveness of the feature extraction, SVM was used for diagnostic verification. The specific steps are as follows:

[0104] First, Raman spectra of insulating oil at different aging times were obtained. According to DL / T 984, the prepared oil samples were experimentally divided into four stages (I: Good, II: Operable, III: Caution, IV: Exit from operation). Then, raw ultrasonic data for different oil states were obtained, such as... Figure 2 As shown. The ultrasonic data feature extraction method for insulating oil extracts the ultrasonic features of the oil, such as... Figure 3 As shown, Figure 3 To calculate the AIC values ​​for different feature dimensions, SVM was used to validate the AIC-LDA ultrasonic features. The results are shown in Tables 1 and 2 below. The results show that using this method for ultrasonic data extraction of insulating oil significantly improves the accuracy of feature extraction.

[0105] Table 1. Diagnostic results using raw ultrasound data as SVM input.

[0106]

[0107] Table 2. Diagnostic results using oil ultrasonic AIC-LDA features as SVM input.

[0108]

[0109] It should be noted that LDA refers to the linear discriminant analysis method, which is a method that can classify the features of two or more objects. This method extracts features from the ultrasonic signal data of insulating oil based on the AIC-LDA method. It does not require user intervention and can effectively extract the ultrasonic feature information of the oil, highlight the characteristics of the ultrasonic signal data of the insulating oil under test, ensure the validity of the ultrasonic data, and effectively improve the accuracy of subsequent application of ultrasonic data for oil condition diagnosis.

[0110] It should be understood that, although Figure 1The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.

[0111] In some embodiments, such as Figure 4 The diagram shows a block diagram of an ultrasonic feature extraction device 400 for insulating oil provided in an embodiment of this application, including: an acquisition module 401, a decomposition module 402, and an extraction module 403, wherein:

[0112] The acquisition module 401 is used to acquire an insulating oil ultrasonic signal dataset, wherein the insulating oil ultrasonic signal dataset includes insulating oil ultrasonic signals and the categories corresponding to the insulating oil ultrasonic signal data;

[0113] The decomposition module 402 is used to calculate the intra-class divergence matrix and inter-class divergence matrix using the ultrasonic signal data of insulating oil in each category, to obtain the first divergence matrix and the second divergence matrix of the ultrasonic signal dataset of insulating oil, and to obtain the third matrix using the first divergence matrix and the second divergence matrix.

[0114] The extraction module 403 is used to perform feature decomposition on the third matrix to obtain multiple feature values ​​and corresponding feature vectors. The feature vectors corresponding to feature values ​​greater than the first preset value are calculated to obtain the AIC value of the feature vectors. The feature vectors with AIC values ​​less than the second preset value are selected as the optimal feature vectors. The feature matrix is ​​constructed based on the optimal feature vectors. The optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix.

[0115] In some embodiments, intra-class and inter-class divergence matrices are calculated using ultrasonic signal data of insulating oil from each category to obtain a first divergence matrix and a second divergence matrix of the ultrasonic signal dataset of insulating oil, including:

[0116] The first divergence matrix is ​​obtained by using the formula for calculating the intra-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the intra-class divergence matrix is ​​as follows:

[0117]

[0118] In the formula, S b Let N represent the first divergence matrix.j U represents the number of ultrasonic signal data of the j-th type of insulating oil. j is the mean value of ultrasonic signal data of the j-th type of insulating oil, and u represents the average value of ultrasonic signal data of all insulating oils;

[0119] The second divergence matrix is ​​obtained by using the formula for calculating the inter-class divergence matrix and the ultrasonic signal data of insulating oil in each category. The formula for calculating the inter-class divergence matrix is ​​as follows:

[0120]

[0121] In the formula, S w Let X represent the second divergence matrix. j U represents the number of ultrasonic signal data for the j-th type of insulating oil. j is the mean value of the ultrasonic signal data of the j-th type of insulating oil, and k represents the number of types.

[0122] In some embodiments, obtaining a third matrix using a first divergence matrix and a second divergence matrix includes:

[0123] The third matrix is ​​obtained by calculating using the first and second divergence matrices, where the calculation formula is:

[0124]

[0125] In the formula, S b Let S represent the first divergence matrix. w Let Z represent the second divergence matrix, and Z represent the third matrix.

[0126] In some embodiments, eigenvalue decomposition is performed on the third matrix to obtain multiple eigenvalues ​​and corresponding eigenvectors. The AIC value of the eigenvectors corresponding to the eigenvalues ​​greater than a first preset value is calculated, including:

[0127] Perform eigenvalue decomposition on the third matrix to obtain multiple eigenvalues ​​and their corresponding eigenvectors;

[0128] The AIC value of each eigenvector is calculated according to the AIC criterion, where the formula for calculating the AIC criterion is as follows:

[0129]

[0130] In the formula, K (K≤d) is the dimension of the dimensionality reduction matrix, N is the total number of categories, and D max D represents the maximum error between ultrasonic signal data of insulating oil across all categories. min This represents the minimum error between ultrasonic signal data for insulating oil across all categories.

[0131] In some embodiments, a feature matrix is ​​constructed based on the optimal feature vector, and the optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix, including:

[0132] A feature matrix is ​​constructed based on the optimal feature vector, and the optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix. The expression for the optimal features is:

[0133]

[0134] in, Let X represent the feature matrix, where X represents the ultrasonic signal data of insulating oil for all categories, and Y represents the optimal feature.

[0135] In one embodiment of this application, a computer device is provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above steps. The implementation principle and technical effects of the computer device provided in this embodiment are similar to those of the above method embodiments, and will not be repeated here.

[0136] In one embodiment of this application, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it performs the above steps; the implementation principle and technical effects of the computer-readable storage medium provided in this embodiment are similar to those of the above method embodiments, and will not be repeated here.

[0137] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for ultrasonic feature extraction of insulating oil, characterized in that, include: Collect an ultrasonic signal dataset of insulating oil, wherein the ultrasonic signal dataset of insulating oil includes ultrasonic signals of insulating oil and the categories corresponding to the ultrasonic signal data of insulating oil; The first divergence matrix is ​​obtained by calculating the intra-class divergence matrix using the formula and the ultrasonic signal data of the insulating oil in each category. The formula for calculating the intra-class divergence matrix is ​​as follows: In the formula, Let represent the first divergence matrix. N j Indicates the first j The number of ultrasonic signal data of the insulating oil described above. u j It is the first j The mean value of the ultrasonic signal data of the insulating oil described above. This represents the average value of all the ultrasonic signal data of the insulating oil. The second divergence matrix is ​​obtained by using the inter-class divergence matrix calculation formula and the ultrasonic signal data of the insulating oil in each category. The inter-class divergence matrix calculation formula is as follows: In the formula, Denotes the second divergence matrix. Indicates the first j The number of ultrasonic signal data of the insulating oil described above. u j It is the first j The mean value of the ultrasonic signal data of the insulating oil described above. Indicates the number of categories; The third matrix is ​​obtained using the first scatter matrix and the second scatter matrix, wherein the calculation formula is: Z= In the formula, Let represent the first divergence matrix. Let Z represent the second divergence matrix, and Z represent the third matrix. The third matrix is ​​subjected to eigenvalue decomposition to obtain multiple eigenvalues ​​and corresponding eigenvectors. The dimension reduction matrix composed of eigenvectors corresponding to eigenvalues ​​greater than a first preset value is calculated to obtain the AIC value of the dimension reduction matrix. The eigenvectors in the dimension reduction matrix whose AIC value is less than a second preset value are selected as the optimal eigenvectors. A feature matrix is ​​constructed based on the optimal eigenvectors. The optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix.

2. The method for ultrasonic feature extraction of insulating oil as described in claim 1, characterized in that, The step of performing eigenvalue decomposition on the third matrix to obtain multiple eigenvalues ​​and corresponding eigenvectors, and calculating the AIC value of the dimensionality-reduced matrix composed of the eigenvectors corresponding to the eigenvalues ​​greater than a first preset value, includes: The third matrix is ​​subjected to eigenvalue decomposition to obtain multiple eigenvalues ​​and eigenvectors corresponding to the eigenvalues; The AIC value of each of the eigenvectors is calculated according to the AIC criterion, wherein the formula for calculating the AIC criterion is as follows: In the formula, K Let be the dimension of the reduced-dimensional matrix. The total number of categories, The maximum error between ultrasonic signal data of insulating oil across all categories. This represents the minimum error between ultrasonic signal data for insulating oil across all categories.

3. The method for ultrasonic feature extraction of insulating oil as described in claim 1, characterized in that, The step of constructing a feature matrix based on the optimal feature vector and using the feature matrix to obtain the optimal features of the insulating oil ultrasonic signal dataset includes: A feature matrix is ​​constructed based on the optimal feature vector, and the optimal features of the insulating oil ultrasonic signal dataset are obtained using the feature matrix. The expression for the optimal features is: in, Represents the characteristic matrix, X This represents ultrasonic signal data for all categories of insulating oil. This represents the optimal feature.

4. An ultrasonic feature extraction device for insulating oil, characterized in that, include: The acquisition module is used to acquire an ultrasonic signal dataset of insulating oil, wherein the ultrasonic signal dataset of insulating oil includes ultrasonic signals of insulating oil and the categories corresponding to the ultrasonic signal data of insulating oil; The decomposition module is used to calculate the first divergence matrix using the intra-class divergence matrix calculation formula and the ultrasonic signal data of the insulating oil in each category. The intra-class divergence matrix calculation formula is as follows: In the formula, Let represent the first divergence matrix. N j Indicates the first j The number of ultrasonic signal data of the insulating oil described above. u j It is the first j The mean value of the ultrasonic signal data of the insulating oil described above. This represents the average value of all the ultrasonic signal data of the insulating oil. The second divergence matrix is ​​obtained by using the inter-class divergence matrix calculation formula and the ultrasonic signal data of the insulating oil in each category. The inter-class divergence matrix calculation formula is as follows: In the formula, Denotes the second divergence matrix. Indicates the first j The number of ultrasonic signal data of the insulating oil described above. u j It is the first j The mean value of the ultrasonic signal data of the insulating oil described above. Indicates the number of categories; The third matrix is ​​obtained using the first scatter matrix and the second scatter matrix; wherein the calculation formula is: Z= In the formula, Let represent the first divergence matrix. Let Z represent the second divergence matrix, and Z represent the third matrix. The extraction module is used to perform feature decomposition on the third matrix to obtain multiple feature values ​​and feature vectors corresponding to the feature values. It calculates the dimension reduction matrix composed of feature vectors corresponding to feature values ​​greater than a first preset value to obtain the AIC value of the dimension reduction matrix. It selects the feature vectors in the dimension reduction matrix whose AIC value is less than a second preset value as the optimal feature vectors. It constructs a feature matrix based on the optimal feature vectors and uses the feature matrix to obtain the optimal features of the insulating oil ultrasonic signal dataset.

5. A computer device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the ultrasonic feature extraction method for insulating oil as described in any one of claims 1 to 3.

6. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the ultrasonic feature extraction method for insulating oil as described in any one of claims 1 to 3.

Images