A cable partial discharge detection method and system based on multi-voiceprint feature spectrum recognition

Abstract

The present invention discloses a cable partial discharge detection method and system based on multi-voiceprint feature spectrum recognition. The method comprises collecting sound signals; preprocessing the sound signals to obtain preprocessed sound signals; generating a time-frequency spectrum and a Mel-frequency spectrum based on the preprocessed sound signals; extracting the resonance peak positions and single discharge times in the spectrum as feature parameters; utilizing a fuzzy c-means clustering method to obtain the discrimination results of a sample to be identified under a certain spectrum; and comprehensively analyzing the discrimination results of different spectrums to determine the fault type of the partial discharge sample to be identified. The present invention can mine the rich features contained in ultrasonic signals and improve the accuracy of partial discharge detection.

Description

Technical Field

[0001] The present invention relates to the technical field of cable partial discharge fault detection, and in particular to a cable partial discharge detection method and system based on multi-voiceprint feature spectrum recognition. Background Art

[0002] Cable equipment provides power to primary and secondary equipment within the substation and is a crucial component. A failure in this equipment can force an emergency shutdown of key equipment, severely impacting the safe and stable operation of the power system. With the development of my country's power grid, partial discharge (PD) and overheating have become major causes of cable equipment failure. Due to the hidden nature of cable line installation and the limitations of testing equipment, locating a fault once one occurs is a complex and time-consuming process, requiring significant effort and manpower.

[0003] In recent years, with the development of power systems, the environments in which cable equipment operate have become increasingly complex, leading to a higher probability of partial discharge (PD) in these devices. Consequently, PD detection in cable equipment has attracted extensive research attention. Most current methods rely on single features, making it difficult to exploit the rich features contained in ultrasonic signals, thus limiting their accuracy. Summary of the Invention

[0004] The purpose of the present invention is to provide a cable partial discharge detection method and system based on multi-voiceprint feature spectrum recognition, which can mine the rich features contained in ultrasonic signals and improve the accuracy of local detection.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A cable partial discharge detection method based on multi-voiceprint feature spectrum recognition includes:

[0007] Collect sound signals;

[0008] Preprocessing the sound signal to obtain a preprocessed sound signal;

[0009] Generating a time-frequency spectrum and a Mel-frequency spectrum according to the preprocessed sound signal;

[0010] The resonance peak position and single discharge time in the extracted spectrum are used as characteristic parameters;

[0011] According to the characteristic parameters, a fuzzy c-means clustering method is used to obtain a discrimination result of the sample to be identified under a certain spectrum judgment;

[0012] The discrimination results of different spectra are analyzed comprehensively to determine the fault type of the partial discharge sample to be identified.

[0013] Optionally, preprocessing the sound signal to obtain a preprocessed sound signal specifically includes:

[0014] Emphasizing the high frequency portion of the sound signal to obtain a sound signal with a flattened spectrum;

[0015] Dividing a long section of voice data of the sound signal in the spectrum-flattened sound signal into a plurality of segments;

[0016] Performing windowing processing on the plurality of segments to obtain a windowed sound signal;

[0017] Endpoint detection is performed on the sound signal after the windowing process, silent segments in the sound signal are removed, and effective sound segments are retained to obtain a preprocessed sound signal.

[0018] Optionally, generating a time-frequency spectrum and a Mel-frequency spectrum according to the preprocessed sound signal specifically includes:

[0019] Performing a discrete Fourier transform on the preprocessed sound signal to obtain a time-frequency spectrum corresponding to the sound signal;

[0020] Obtaining a power spectrum of the sound signal according to the time-frequency spectrum diagram;

[0021] Passing the power spectrum through a set of Mel-scale triangular filters, calculating the energy in each filter;

[0022] Performing a logarithmic operation on the energy of each filter to obtain a Mel spectrogram corresponding to the sound signal;

[0023] Optionally, the formula used for the Fourier transform is:

[0024]

[0025] Among them, x(n) is the time domain signal, x i (n) is the data of the i-th frame, X i (k) is the k-th complex coefficient of the i-th frame.

[0026] Optionally, obtaining a power spectrum of the sound signal according to the time-frequency spectrum graph specifically includes:

[0027] According to the time-frequency spectrum diagram, each frequency spectrum is obtained;

[0028] For each of the spectra, the formula Perform modular square processing to obtain the power spectrum of the sound signal;

[0029] Among them, P i (k) is the power spectrum of the i-th frame.

[0030] Optionally, the function H of the Mel-scale triangular filter m (k) is:

[0031]

[0032] Where m is the filter bank number, f(m) is the center frequency of the triangular filter, and k is the Mel-scale frequency in Hz.

[0033] Optionally, extracting the resonance peak position and single discharge time in the time-frequency spectrum and the Mel spectrum as characteristic parameters specifically includes:

[0034] grayscale processing is performed on the time-frequency spectrum and the Mel spectrum to obtain a processed time-frequency spectrum and a processed Mel spectrum;

[0035] According to the processed time-frequency spectrum and the processed Mel spectrum, a frequency window and a time window of preset sizes are respectively used to slide sequentially from the frequency starting point and the time starting point to perform detection processing on the resonance peak position and the single discharge time to obtain characteristic parameters.

[0036] Optionally, the method of obtaining a discrimination result of a sample to be identified under a certain spectrum judgment by using a fuzzy c-means clustering method specifically includes:

[0037] The sample set x consisting of n samples is x={x1,x2,…,x n} is divided into C categories, and the cluster center of each category is found. Suppose a sample x j The probability of belonging to the i-th type of partial discharge is u ij , the classification structure can be represented by a c×n matrix U, called a fuzzy matrix. The steps to determine the optimal U matrix and cluster center are as follows:

[0038] 1) Initialize the membership matrix u to satisfy the constraints

[0039]

[0040] 2) Use the formula Calculate the cluster center c of C clusters i (i=1, 2, ..., c);

[0041] 3) According to the formula Calculate the evaluation function. If it is less than a certain threshold, or the change relative to the last evaluation function is less than a certain threshold, the algorithm stops, where p∈[1,+∞) is a weighted index, d ij =||c i -x j || is the distance from the jth sample to the i-th category;

[0042] 4) Use Calculate the new U matrix and return to step 2).

[0043] Optionally, the comprehensive analysis of the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified specifically includes:

[0044] Performing weighted summation on the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified;

[0045] The weighted summation formula is:

[0046]

[0047] Among them, w m is the weight of m judgment methods, u m is the result value of a certain discharge type obtained by m judgment methods, and PE is the final judgment result value of a certain discharge type obtained by comprehensive calculation. If the PE value exceeds the threshold, it is determined that this type of partial discharge has occurred.

[0048] Optionally, the method for determining the weights of the different spectrum judgment methods is an entropy value determination weighting method, and the formula is:

[0049] For the n samples obtained, the data matrix Y of m judgment methods is ij} n×m ,

[0050] Information entropy e j Used to measure the utility value of information of j indicators, the information entropy value of j indicators is:

[0051]

[0052] Then the weight of the j-item indicator is:

[0053]

[0054] Among them, h j It represents the information utility value of a certain indicator, which is the difference between the information entropy of the indicator and 1.

[0055] A cable partial discharge detection system based on multi-voiceprint feature spectrum recognition includes:

[0056] A sound signal acquisition module, used for collecting sound signals;

[0057] A sound signal preprocessing module, used to preprocess the sound signal to obtain a preprocessed sound signal;

[0058] A spectrogram generation module, configured to generate a time-frequency spectrum and a Mel-frequency spectrum according to the preprocessed sound signal;

[0059] a discrimination result determination module, configured to determine the type of partial discharge based on the resonance peak position in the time-frequency spectrum and the Mel spectrum and the single discharge time;

[0060] The fault type determination module is used to comprehensively analyze the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified.

[0061] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0062] The present invention conducts comprehensive analysis based on multiple feature spectra, overcoming the difficulty of insufficient prediction accuracy caused by a single feature. The method provided by the present invention can alleviate the technical problem in the prior art of increasing the probability of partial discharge failures in cable equipment with a complex working environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0064] Figure 1 This is a flow chart of the cable partial discharge detection method based on multi-voiceprint feature spectrum recognition of the present invention;

[0065] Figure 2 This is a structural diagram of the cable partial discharge detection system based on multi-voiceprint feature spectrum recognition of the present invention. DETAILED DESCRIPTION

[0066] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0067] The purpose of the present invention is to provide a cable partial discharge detection method and system based on multi-voiceprint feature spectrum recognition, which can mine the rich features contained in ultrasonic signals and improve the accuracy of local detection.

[0068] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0069] Example 1:

[0070] In recent years, with the development of power systems, the environments in which cable equipment operate have become increasingly complex, leading to a higher probability of partial discharge (PD) in these devices. Consequently, PD detection in cable equipment has attracted extensive research attention. Most current methods rely on single features, making it difficult to exploit the rich features contained in ultrasonic signals, thus limiting their accuracy.

[0071] To facilitate understanding of this embodiment, a cable partial discharge detection method based on multi-voiceprint feature spectrum recognition disclosed in Example 1 of the present invention is first introduced in detail.

[0072] Figure 1 This is a flow chart of the cable partial discharge detection method based on multi-voiceprint feature spectrum recognition of the present invention, as shown in Figure 1 As shown, the present invention provides a cable partial discharge detection method based on multi-voiceprint feature spectrum recognition, the method comprising:

[0073] Step 101: Collecting sound signals;

[0074] Step 102: Preprocess the sound signal to obtain a preprocessed sound signal.

[0075] This step specifically includes:

[0076] Emphasizing the high frequency portion of the sound signal to obtain a sound signal with a flattened spectrum;

[0077] Dividing a long section of voice data of the sound signal in the spectrum-flattened sound signal into a plurality of segments;

[0078] Performing windowing processing on the plurality of segments to obtain a windowed sound signal;

[0079] Endpoint detection is performed on the sound signal after the windowing process, silent segments in the sound signal are removed, and effective sound segments are retained to obtain a preprocessed sound signal.

[0080] Due to differences in the physical properties of the signal source, the generated sound signal voiceprint characteristics vary. However, sound signals are susceptible to various noise interferences, which may affect subsequent work.

[0081] Therefore, this step preprocesses the sound signal, including pre-emphasis, frame division and windowing, endpoint detection, and speech enhancement. Pre-emphasis is performed first, aiming to emphasize the high-frequency portion of the signal, reduce the impact of noise, and flatten the speech signal spectrum. Then, a long segment of speech data is divided into several smaller segments, known as frames.

[0082] These short speech signals can maintain a stationary state for short periods of time and can be processed using stationary process methods. Framing introduces a truncation effect to the signal. To achieve a smooth transition at the truncation point, windowing is required. Finally, endpoint detection is performed on the speech signal to remove silent segments and retain valid speech segments.

[0083] Step 103: Generate a time-frequency spectrum and a Mel-frequency spectrum according to the pre-processed sound signal;

[0084] This step specifically includes:

[0085] Step 1031: Perform discrete Fourier transform on the pre-processed sound signal to obtain a time-frequency spectrum diagram corresponding to the sound signal.

[0086] The formula used for the Fourier transform is:

[0087]

[0088] Among them, x(n) is the time domain signal, x i (n) is the data of the i-th frame, X i (k) is the k-th complex coefficient of the i-th frame.

[0089] Traditionally, using a single feature can easily lead to information loss and misjudgment. Therefore, the present invention uses a method of generating a multi-feature spectrum to reflect the characteristic information of the sound signal as completely as possible, so as to subsequently identify the fault type of the partial discharge sample.

[0090] Step 1032: Obtain the power spectrum of the sound signal according to the time-frequency spectrum diagram.

[0091] This step specifically includes:

[0092] According to the time-frequency spectrum diagram, each frequency spectrum is obtained;

[0093] For each of the spectra, the formula Perform modular square processing to obtain the power spectrum of the sound signal;

[0094] Among them, P i (k) is the power spectrum of the i-th frame.

[0095] Step 1033: Pass the power spectrum through a set of Mel-scale triangular filters to calculate the energy in each filter.

[0096] The Mel-scale triangular filter function H m (k) is:

[0097]

[0098] Where m is the filter bank number, f(m) is the center frequency of the triangular filter, and k is the Mel-scale frequency in Hz.

[0099] Step 1034: performing a logarithmic operation on the energy of each filter to obtain a Mel spectrogram corresponding to the sound signal;

[0100] Step 104: extracting the resonance peak position and single discharge time in the time-frequency spectrum and the Mel spectrum as characteristic parameters.

[0101] This step specifically includes:

[0102] Step 1041: grayscale processing is performed on the time-frequency spectrum and the Mel spectrum to obtain a processed time-frequency spectrum and a processed Mel spectrum;

[0103] Step 1042: Based on the processed time-frequency spectrum and the processed Mel spectrum, a frequency window and a time window of preset sizes are respectively used to slide sequentially from the frequency starting point and the time starting point to detect the resonance peak position and the single discharge time, and obtain characteristic parameters. The characteristic parameters include: counting the pixel value of each point in the spectrum; recording the frequency range of the point with the largest pixel value (i.e., the resonance peak position); recording the time range of the discharge area in the spectrum;

[0104] Step 105: using the fuzzy c-means clustering method to obtain a discrimination result of the sample to be identified under a certain spectrum judgment;

[0105] This step specifically includes:

[0106] The sample set x=(x1,x2,…,x n} is divided into C categories, and the cluster center of each category is found. Suppose a sample x j The probability of belonging to the i-th type of partial discharge is u ij , the classification structure can be represented by a c×n matrix U, called a fuzzy matrix. The steps to determine the optimal U matrix and cluster center are as follows:

[0107] Step 1051: Initialize the membership matrix u to satisfy the constraints

[0108]

[0109] Step 1052: Use the formula Calculate the cluster center c of C clusters i (i=1, 2, ..., c);

[0110] Step 1053: According to the formula Calculate the evaluation function. If it is less than a certain threshold, or the change relative to the last evaluation function is less than a certain threshold, the algorithm stops, where p∈[1,+∞) is a weighted index, d ij =||c i -x j || is the distance from the jth sample to the i-th category;

[0111] Step 1054: Use Calculate the new U matrix and return to step 2). ij is an element of U and also a sample x j The probability u of belonging to the i-th type of partial discharge ij The judgment result of .

[0112] Step 106: performing weighted summation on the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified.

[0113] This step specifically includes:

[0114] The weighted summation formula is:

[0115]

[0116] Among them, w m is the weight of m judgment methods, u m is the result value of a certain discharge type obtained by m judgment methods, and PE is the final judgment result value of a certain discharge type obtained by comprehensive calculation. If the PE value exceeds the threshold, it is determined that this type of partial discharge has occurred.

[0117] The method for determining the weights of different spectrum judgment methods is the entropy value determination weight method, and the formula is:

[0118] For the n samples obtained, the data matrix Y of m judgment methods is ij} n×m ,

[0119] Information entropy e j Used to measure the utility value of information of j indicators, the information entropy value of j indicators is:

[0120]

[0121] Then the weight of the j-item indicator is:

[0122]

[0123] Among them, h j It represents the information utility value of a certain indicator, which is the difference between the information entropy of the indicator and 1.

[0124] The above method only obtains the discrimination result value under a single spectrum. It is necessary to combine the discrimination result value of another spectrum for comprehensive analysis to obtain the final discrimination result value and determine the fault type. The diagnostic method provided by the present invention performs comprehensive analysis based on multi-feature spectra, overcoming the difficulty of insufficient prediction accuracy caused by a single feature. The method provided by the present invention can alleviate the technical problem of the increased probability of partial discharge faults in the complex working environment of cable equipment in the prior art. The present invention uses a novel fault diagnosis method to perform real-time detection of cable status, which has the advantages of high efficiency, high accuracy and good timeliness.

[0125] Example 2:

[0126] Figure 2 This is a structural diagram of the cable partial discharge detection system based on multi-voiceprint feature spectrum recognition of the present invention. Figure 2 As shown, the second embodiment of the present invention provides a cable partial discharge detection system based on multi-voiceprint feature spectrum recognition. The system includes:

[0127] The sound signal collection module 201 is used to collect sound signals;

[0128] The sound signal preprocessing module 202 is used to preprocess the sound signal to obtain a preprocessed sound signal;

[0129] A spectrogram generating module 203 is configured to generate a time-frequency spectrogram and a Mel-frequency spectrogram according to the preprocessed sound signal;

[0130] A determination result determination module 204 is configured to determine the type of partial discharge based on the resonance peak position in the time-frequency spectrum and the Mel spectrum and the single discharge time;

[0131] The fault type determination module 205 is configured to comprehensively analyze the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified.

[0132] Example 3:

[0133] This embodiment provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the cable partial discharge detection method based on multi-voiceprint feature spectrum recognition of embodiment 1.

[0134] Optionally, the above-mentioned electronic device may be a server.

[0135] In addition, an embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the cable partial discharge detection method based on multi-voiceprint feature spectrum recognition of embodiment 1 is implemented.

[0136] Embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0137] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0138] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0139] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between the various embodiments can be referenced to each other. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referenced to the method description.

[0140] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A cable partial discharge detection method based on multi-voiceprint feature spectrum recognition, characterized in that: The method comprises: Collect sound signals; Preprocessing the sound signal to obtain a preprocessed sound signal; Generating a time-frequency spectrum and a Mel-frequency spectrum according to the preprocessed sound signal; Extracting the resonance peak position and single discharge time in the time-frequency spectrum and the Mel spectrum as characteristic parameters; According to the characteristic parameters, a fuzzy c-means clustering method is used to obtain a discrimination result of the sample to be identified under a certain spectrum judgment; Comprehensively analyze the discrimination results of different spectra to determine the fault type of the partial discharge sample to be identified; The preprocessing of the sound signal to obtain a preprocessed sound signal specifically includes: Emphasizing the high frequency portion of the sound signal to obtain a sound signal with a flattened spectrum; Dividing a long section of voice data of the sound signal in the spectrum-flattened sound signal into a plurality of segments; Performing windowing processing on the plurality of segments to obtain a windowed sound signal; Performing endpoint detection on the sound signal after the windowing process, removing silent segments in the sound signal, retaining valid sound segments, and obtaining a preprocessed sound signal; Generating a time-frequency spectrum and a Mel-frequency spectrum according to the preprocessed sound signal specifically includes: Performing a discrete Fourier transform on the preprocessed sound signal to obtain a time-frequency spectrum corresponding to the sound signal; Obtaining a power spectrum of the sound signal according to the time-frequency spectrum diagram; Passing the power spectrum through a set of Mel-scale triangular filters, calculating the energy in each filter; Performing a logarithmic operation on the energy of each filter to obtain a Mel spectrogram corresponding to the sound signal; The extracting the resonance peak position and the single discharge time in the time-frequency spectrum and the Mel spectrum as characteristic parameters specifically includes: grayscale processing is performed on the time-frequency spectrum and the Mel spectrum to obtain a processed time-frequency spectrum and a processed Mel spectrum; According to the processed time-frequency spectrum and the processed Mel spectrum, a frequency window and a time window of preset sizes are respectively used to slide sequentially from the frequency starting point and the time starting point to perform detection processing on the resonance peak position and the single discharge time to obtain characteristic parameters.

2. The cable partial discharge detection method based on multi-voiceprint feature spectrum recognition according to claim 1 is characterized in that: The formula used for the Fourier transform is: Among them, x(n) is the time domain signal, x i (n) is the data of the i-th frame, X i (k) is the k-th complex coefficient of the i-th frame.

3. The cable partial discharge detection method based on multi-voiceprint feature spectrum recognition according to claim 1 is characterized in that: Obtaining the power spectrum of the sound signal according to the time-frequency spectrum diagram specifically includes: According to the time-frequency spectrum diagram, each frequency spectrum is obtained; For each of the spectra, the formula Perform modular square processing to obtain the power spectrum of the sound signal; Among them, P i (k) is the power spectrum of the i-th frame.

4. The cable partial discharge detection method based on multi-voiceprint feature spectrum recognition according to claim 1 is characterized in that: The Mel-scale triangular filter function H m (k) is: Where m is the filter bank number, f(m) is the center frequency of the triangular filter, and k is the Mel-scale frequency in Hz.

5. The cable partial discharge detection method based on multi-voiceprint feature spectrum recognition according to claim 1 is characterized in that: The method of using the fuzzy c-means clustering method to obtain the discrimination result of the sample to be identified under a certain spectrum judgment specifically includes: The sample set x consisting of n samples is x={x1,x2,…,x n } is divided into C categories, and the cluster center of each category is found. Suppose a sample x j The probability of belonging to the i-th type of partial discharge is u ij The classification structure of the sample set is represented by a c×n matrix U, called a fuzzy matrix. The steps to determine the optimal U matrix and cluster center are as follows: 1) Initialize the membership matrix u to satisfy the constraints 2) Use the formula Calculate the cluster center c of C clusters i (i=1, 2, ..., c); 3) According to the formula Calculate the evaluation function. If it is less than a certain threshold, or the change relative to the last evaluation function is less than a certain threshold, the algorithm for calculating the evaluation function stops, where p∈[1,+∞) is a weighted index, d ij =||c i -x j || is the distance from the jth sample to the i-th category; 4) Use Calculate the new U matrix and return to step 2).

6. The cable partial discharge detection method based on multi-voiceprint feature spectrum recognition according to claim 1 is characterized in that: The comprehensive analysis of the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified specifically includes: Performing weighted summation on the discrimination results of the different spectra to determine the fault type of the partial discharge sample to be identified; The weighted summation formula is: Among them, w m is the weight of m judgment methods, u ij is the result value of a certain discharge type obtained by m judgment methods, PE is the final judgment result value of a certain discharge type obtained by comprehensive calculation, and if the PE value exceeds the threshold, it is determined that the fault type of the partial discharge sample to be identified has occurred.

7. The cable partial discharge detection method based on multi-voiceprint feature spectrum recognition according to claim 6 is characterized in that: The method for determining the weights of the different spectra is an entropy value determination weighting method, which specifically includes: For the n samples obtained, the data matrix Y of m judgment methods is ij } n×m , Information entropy e j Used to measure the utility value of information of j indicators, the information entropy value of j indicators is: Then the weight of the j-item indicator is: Among them, h j It represents the information utility value of a certain indicator, which is the difference between the information entropy of the indicator and 1.

8. A cable partial discharge detection system based on multi-voiceprint feature spectrum recognition, characterized in that: The system comprises: A sound signal acquisition module, used for collecting sound signals; A sound signal preprocessing module, used to preprocess the sound signal to obtain a preprocessed sound signal; A spectrogram generating module, configured to generate a time-frequency spectrum and a Mel-frequency spectrum according to the preprocessed sound signal; A characteristic parameter determination module is used to extract the resonance peak position and single discharge time in the time-frequency spectrum and the Mel spectrum as characteristic parameters; A discrimination result determination module is used to obtain the discrimination result of the sample to be identified under a certain spectrum judgment using the fuzzy c-means clustering method according to the characteristic parameters; The fault type determination module is used to comprehensively analyze the discrimination results of different spectra to determine the fault type of the partial discharge sample to be identified; The sound signal preprocessing module specifically includes: a sound signal determination unit, configured to emphasize a high frequency portion of the sound signal to obtain a sound signal with a flattened spectrum; A sound signal dividing unit, configured to divide a long section of sound signal voice data in the spectrum-flattened sound signal into a plurality of segments; a windowing processing unit, configured to perform windowing processing on the plurality of segments to obtain a sound signal after the windowing processing; a preprocessing unit, configured to perform endpoint detection on the sound signal after the windowing process, remove silent segments from the sound signal, retain valid sound segments, and obtain a preprocessed sound signal; The spectrum generation module specifically includes: A discrete Fourier transform unit, configured to perform a discrete Fourier transform on the preprocessed sound signal to obtain a time-frequency spectrum corresponding to the sound signal; a power spectrum determination unit, configured to obtain a power spectrum of the sound signal according to the time-frequency spectrum diagram; a filter energy calculation unit, configured to calculate the energy of each filter according to the power spectrum passing through a set of Mel-scale triangular filters; A Mel spectrogram determination unit, configured to perform a logarithmic operation on the energy of each filter to obtain a Mel spectrogram corresponding to the sound signal; The characteristic parameter determination module specifically includes: a grayscale processing unit, configured to perform grayscale processing on the time-frequency spectrum and the Mel spectrum to obtain a processed time-frequency spectrum and a processed Mel spectrum; The characteristic parameter determination unit is used to detect the resonance peak position and the single discharge time according to the processed time-frequency spectrum and the processed Mel spectrum, respectively, by sliding the frequency window and time window of preset sizes from the frequency starting point and the time starting point in sequence to obtain characteristic parameters.

Images