Transformer partial discharge defect voiceprint feature extraction method and system

Abstract

The invention belongs to the technical field of transformers, and provides a transformer partial discharge defect voiceprint feature extraction method and system, and the technical scheme comprises the steps: obtaining the original voiceprint data of a transformer; performing noise reduction on the original voiceprint data of the transformer based on an improved spectral subtraction method to obtain noise-reduced voiceprint data, and performing data expansion on the noise-reduced voiceprint data to obtain data-expanded voiceprint data; based on the voiceprint data after data expansion, in combination with the fundamental frequency and frequency distribution characteristics of the transformer, feature extraction is carried out to obtain the features of the partial discharge defect of the transformer, the voiceprint feature distribution characteristics and the deep relation with the transformer fault are fully considered, and the problems that the time domain, frequency domain and time-frequency domain features of the power transformer are multiple in variety, and the power transformer fault is poor are solved. The extraction of effective features is improved, the extraction of redundant and invalid features is avoided, and the prediction precision of the partial discharge defect of the transformer is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of transformers, and in particular relates to a method and system for extracting soundprint features of partial discharge defects in transformers. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] During operation, the transformer will emit vibration signals containing mechanical fault information, which can be used as a data source for evaluating the operating status of the transformer. The transformer's voiceprint signals include dozens of time domain, frequency domain and time-frequency domain features. The existing partial discharge defect feature extraction methods have the following technical defects:

[0004] 1. The distribution characteristics of voiceprint features and their in-depth connection with transformer faults are not fully considered, so most of the extracted features are redundant and invalid.

[0005] 2. When the types and degrees of external environmental noise in the voiceprint signal are diverse and complex, the universality and effectiveness of the existing spectral subtraction denoising may be affected;

[0006] 3. As the quality of equipment is getting better and better, there are not many transformer voiceprint defect samples. Most of the existing sample expansion technologies are based on single samples, which does not fully take into account that the sounds collected in the substation are mostly the superposition of direct and reflected sounds from different sound sources, resulting in differences in voiceprint data collected at different locations. In addition, due to differences in structure and working conditions, the voiceprint data of different transformers are even more different, resulting in the inability to accurately predict defects with the extracted features. Summary of the invention

[0007] In order to solve at least one technical problem existing in the above-mentioned background technology, the present invention provides a method and system for extracting voiceprint features of partial discharge defects of transformers, which reduces the noise of the original voiceprint data of the transformer through an improved spectral subtraction method, and at the same time increases the sample size, fully considers the distribution characteristics of the voiceprint features and the in-depth connection with the transformer fault, and extracts the partial discharge defect features, which can effectively monitor the transformer voiceprint and identify defects.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] A first aspect of the present invention provides a method for extracting transformer partial discharge defect soundprint features, comprising the following steps:

[0010] Get the original voiceprint data of the transformer;

[0011] Based on spectral subtraction, the original voiceprint data of the transformer is de-noised to obtain the de-noised voiceprint data;

[0012] Performing data expansion on the denoised voiceprint data to obtain data-expanded voiceprint data;

[0013] Based on the voiceprint data after data expansion, combined with the fundamental frequency and frequency distribution characteristics of the transformer, feature extraction is performed to obtain the resonance demodulation power frequency correlation;

[0014] Based on the resonance demodulation power frequency correlation, it is determined whether there is a partial discharge defect.

[0015] Furthermore, the original voiceprint data of the transformer is denoised based on the improved spectral subtraction method to obtain the denoised voiceprint data, specifically including: finding the signal of the non-noise interference period as the target signal, and then spectrally replacing the signal of the noise interference period, retaining the original phase information of the signal, and replacing it with the amplitude information of the frame with the minimum energy in the interference period.

[0016] Furthermore, the method of searching for a signal in a non-noise interference period as a target signal, and then performing spectrum replacement on the signal in a noise interference period, retaining the original phase information of the signal and replacing it with the amplitude information of the frame with the minimum energy in the interference period, includes:

[0017] The original voiceprint signal is divided into frames, and the background voiceprint is estimated to obtain the amplitude spectrum information of the background voiceprint signal;

[0018] According to the background voiceprint estimation result, the signal-to-noise ratio of each frame of data is calculated, and the sequence number of the frame to be denoised is determined according to the set noise consistency threshold;

[0019] Perform FFT calculation on the noise segment signal based on the sequence number of the frame to be denoised to obtain the phase spectrum information of the noise segment signal;

[0020] The amplitude spectrum information of the background voiceprint signal and the phase spectrum information of the noise segment signal are fused to obtain the fused full spectrum information;

[0021] Based on the fused full spectrum information, IFFT calculation is performed to obtain the denoised time domain signal.

[0022] Furthermore, the step of performing data expansion on the denoised voiceprint data to obtain the voiceprint data after data expansion specifically includes:

[0023] Select a benchmark voiceprint sample and calculate the effective value;

[0024] Calculate the amplitude spectrum of two or more samples respectively, and retain the phase spectrum of the reference sample;

[0025] Based on the sample amplitude spectrum, the peak amplitude sequence is extracted for all samples according to the set peak conditions to obtain the peak amplitude sequence;

[0026] Performing random weighting processing on each peak amplitude sequence to obtain an enhanced peak amplitude sequence;

[0027] The enhanced peak amplitude sequence is summed up according to the sample weights to obtain a mixed peak sequence;

[0028] IFFT calculation is performed according to the effective value and phase spectrum of the reference sequence and the amplitude spectrum of the peak sequence after mixing to obtain the multi-sample mixed enhanced samples.

[0029] Furthermore, before data expansion is performed on the voiceprint data after noise reduction, the signal-to-noise ratio of the transformer body voiceprint is calculated based on the time domain signal after noise reduction, and corresponding data processing is performed according to the comparison between the signal-to-noise ratio of the transformer body voiceprint and the set threshold, including obtaining the body voiceprint signal-to-noise ratio according to the transformer body voiceprint and the total value ratio. If the signal-to-noise ratio is greater than the set threshold, the sound pressure level, harmonic ratio, high-frequency energy ratio, entropy spectrum, and demodulated power frequency correlation characteristics are calculated according to the body frequency; if the signal-to-noise ratio is less than the set threshold, the data for this segment or the data within this time period is discarded.

[0030] Furthermore, the feature extraction process of transformer partial discharge defects includes:

[0031] The voiceprint data after data expansion is filtered and then the envelope is calculated to obtain an envelope signal;

[0032] After performing FFT calculation on the envelope signal, the resonance demodulation power frequency correlation is calculated in combination with the high-frequency envelope spectrum of the voiceprint signal; the resonance demodulation power frequency correlation is the proportion of 50Hz and 100Hz components in the envelope spectrum of the frequency band above 10000Hz in the 50-500Hz frequency band.

[0033] The second aspect of the present invention provides a transformer partial discharge defect soundprint feature extraction system, comprising:

[0034] A data acquisition module, which is used to obtain the original voiceprint data of the transformer;

[0035] A noise reduction module, which is used to reduce the noise of the original voiceprint data of the transformer based on spectral subtraction to obtain the noise-reduced voiceprint data;

[0036] A data expansion module, which is used to expand the voiceprint data after noise reduction to obtain the voiceprint data after data expansion;

[0037] A feature extraction module is used to extract features based on the voiceprint data after data expansion, combined with the fundamental frequency and frequency distribution characteristics of the transformer, to obtain the resonant demodulation power frequency correlation;

[0038] The defect judgment module is used to judge whether there is a partial discharge defect based on the resonance demodulation power frequency correlation.

[0039] A third aspect of the present invention provides a computer-readable storage medium.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the transformer partial discharge defect soundprint feature extraction method as described above.

[0041] A fourth aspect of the present invention provides a computer device.

[0042] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the transformer partial discharge defect voiceprint feature extraction method as described above are implemented.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention proposes a new feature extraction method for partial discharge defects of transformers, which fully considers the distribution characteristics of voiceprint features and their in-depth connection with transformer faults, solves the problem of multiple types of time domain, frequency domain and time-frequency domain features of power transformers, improves the extraction of effective features, avoids the extraction of redundant and invalid features, and improves the prediction accuracy of partial discharge defects of transformers.

[0045] 2. The present invention proposes a new transformer voiceprint data denoising method, which reduces the noise of the original transformer voiceprint data through an improved spectral subtraction method, finds the signal of the non-noise interference period as the target signal, and then performs spectral replacement on the signal of the noise interference period, retains the original phase information of the signal, and replaces it with the amplitude information of the frame with the minimum energy in the interference period, which can effectively remove noise and solves the problem that the universality and effectiveness of the existing spectral subtraction method may be affected when the types and degrees of external environmental noise are diverse and complex.

[0046] 3. The present invention proposes a transformer voiceprint multi-sample data enhancement method, which fully considers that the sounds collected in the substation are mostly the superposition of direct and reflected sounds from different sound sources, solves the problems of unbalanced voiceprint data and small number of samples, and realizes the expansion of transformer voiceprint samples.

[0047] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0049] Figure 1 is a flow chart of a transformer partial discharge defect voiceprint feature extraction method provided by an embodiment of the present invention;

[0050] Figure 2 It is a flow chart of denoising the original voiceprint data of a transformer provided by an embodiment of the present invention;

[0051] Figure 3 is a flow chart of improved spectrum subtraction calculation provided by an embodiment of the present invention;

[0052] Figure 4 It is a flow chart of multi-sample mixed weighted enhancement and expansion based on voiceprint spectrogram provided by an embodiment of the present invention;

[0053] Figure 5 is a time-frequency diagram of the voiceprint signal 1 provided in an embodiment of the present invention;

[0054] Figure 6 is a time-frequency diagram of the voiceprint signal 2 provided in an embodiment of the present invention;

[0055] Figure 7 This is a diagram showing the effect of mixed weighted directional enhancement and expansion of voiceprint signal 1 and voiceprint signal 2 provided in an embodiment of the present invention;

[0056] Figure 8 is a time domain diagram of a partial discharge signal provided by an embodiment of the present invention;

[0057] Fig. 9 It is a frequency domain characteristic diagram of a partial discharge signal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0059] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0060] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0061] Embodiment 1

[0062] like Figure 1 As shown, this embodiment provides a transformer partial discharge defect voiceprint feature extraction method, comprising the following steps:

[0063] Step 1: Obtain the original voiceprint data of the transformer;

[0064] Step 2: De-noise the original voiceprint data of the transformer based on the improved spectral subtraction method to obtain the de-noised voiceprint data; specifically, the steps include:

[0065] Step 201, based on the improved spectral subtraction method, the original voiceprint is processed to obtain a de-noised time domain signal to improve the signal-to-noise ratio, specifically including:

[0066] If there is intermittent interference near the transformer, and the signal interference frequency band overlaps with the transformer body soundprint frequency band, spectral subtraction is used to suppress noise.

[0067] 1. Background voiceprint assessment is not exactly the same as speech signal spectral subtraction. Speech signal spectral subtraction uses the noise statistical characteristics of the speech intermittent segment to replace the noise statistical characteristics of the superimposed noise period, and then performs spectral subtraction to obtain the noise-suppressed speech signal. Since the transformer voiceprint signal is a stable and continuous signal, it is impossible to obtain the intermittent segment signal. Therefore, to solve this problem, it is necessary to determine the type and power of the noise. There are three conventional ideas:

[0068] (1) Several typical noises are recorded in advance. When the signal-to-noise ratio of the actual collected signal decreases, the actual collected signal and the noise are correlated to determine the noise type. The noise intensity is determined based on the signal power difference between the noise superposition segment and the noise-free segment, thereby determining the noise segment to be spectrally subtracted. However, this method requires certain prior knowledge.

[0069] (2) Based on the typical interference signals collected in advance, the noise type recognition model is trained to determine in real time whether there is noise interference and the noise period. Then, based on the signal power difference between the noise superposition section and the noise-free section, the noise intensity is determined, thereby determining the noise section to be spectrally subtracted. This method still requires certain prior knowledge.

[0070] (3) Taking typical fan noise as an example, since the soundprint signals generated by transformer operation and fan operation are both stable noise, it is possible to collect the soundprint of a single fan during operation, determine whether the fan is running based on the sound pressure level or the energy of the specified frequency band (fan rotation frequency band), and perform real-time spectrum subtraction. This method is only applicable to stable and measurable interference.

[0071] In response to the above problems, this embodiment proposes a reverse thinking method for non-stationary interference signals based on the stable characteristics of the transformer voiceprint, that is, by finding the signal of the non-noise interference period as the target signal, and then spectrally replacing the signal of the noise interference period, retaining the original phase information of the signal, and replacing it with the amplitude information of the frame with the minimum energy (the frame with the weakest interference) in the time period, so as to achieve noise suppression of any instantaneous interference.

[0072] like Figure 2 and Figure 3 As shown, the specific steps include:

[0073] Step 211, background voiceprint estimation: divide the original voiceprint signal into frames, calculate the sound pressure level, and estimate the background voiceprint to obtain the background voiceprint signal amplitude spectrum information;

[0074] Step 2012, noise period estimation: according to the background voiceprint estimation result, the signal-to-noise ratio of each frame of data is calculated, and the sequence number of the frame to be denoised is determined according to the set noise consistency threshold;

[0075] Step 2013, phase spectrum information acquisition: performing FFT calculation on the noise segment signal based on the sequence number of the frame to be denoised to obtain the phase spectrum information of the noise segment signal;

[0076] Step 2014, spectrum replacement calculation: fuse the amplitude spectrum information of the background voiceprint signal and the phase spectrum information of the noise segment signal to obtain the fused full spectrum information;

[0077] Step 2015: Based on the fused full spectrum information, perform IFFT calculation to obtain a denoised time domain signal.

[0078] Step 202: Calculate the signal-to-noise ratio of the transformer body voiceprint based on the de-noised time domain signal, and perform corresponding data processing according to the comparison between the signal-to-noise ratio of the transformer body voiceprint and a set threshold value;

[0079] Specifically include:

[0080] Step 2021: extract the fundamental frequency and harmonics based on the denoised time domain signal, and calculate the total value of the sound;

[0081] Step 2022, extracting transformer body soundprint components (50 Hz and harmonic components) in the spectrum;

[0082] Step 2023, the signal-to-noise ratio of the transformer voiceprint is obtained according to the transformer voiceprint and the total value ratio. If the signal-to-noise ratio is greater than the set threshold, the reliability of the signal is high. At this time, characteristic parameters such as sound pressure level, harmonic ratio, high-frequency energy ratio, entropy spectrum, and demodulated power frequency correlation characteristics are calculated according to the body frequency; if the signal-to-noise ratio is less than the set threshold (such as 0.6), it means that the in-band interference is too large and the extracted feature distortion is serious. Considering the stable characteristics of the transformer voiceprint, the data in this section or the data in this time period is discarded;

[0083] In this embodiment, preferably, the total value is calculated within the transformer body soundprint frequency band (50-2000 Hz), because even if the out-of-band interference is large, it actually has little effect on extracting the transformer acoustic vibration characteristics.

[0084] In this embodiment, the characteristics of the transformer body soundprint signal-to-noise ratio are: when the transformer is running without strong interference, 80% of the noise is related to the power frequency. Combined with the power frequency characteristics of the transformer, the transformer body soundprint signal-to-noise ratio is calculated, and the proportion of the 50Hz harmonic component in the total sound pressure level is:

[0085]

[0086] Among them, X i*50 is the spectrum of the voiceprint signal, and the corresponding index is the i-th 50Hz frequency.

[0087] Step 3: Perform data expansion on the denoised voiceprint data to obtain voiceprint data after data expansion;

[0088] In order to solve the problems of unbalanced voiceprint data and small number of samples, and combined with the characteristics of actual voiceprint data (different from conventional image data, the voiceprint data time-frequency diagram is relatively standardized data, and there will be no image flipping, rotation, size change, etc.), it is very important to adopt a reasonable and appropriate data enhancement and expansion method.

[0089] Since the transformer's sound source is not a point sound source and there are a lot of reflection superposition effects, theoretically, the voiceprint data collected at different locations may be different. In addition, due to differences in structure and working conditions, the voiceprint data of different transformers are more different. Based on the above reasons, the most likely features of voiceprint data at different locations have the following characteristics:

[0090] (1) Change of main frequency position;

[0091] (2) The absolute and relative amplitudes of the main peak components change to a certain extent;

[0092] (3) The background noise changes to a certain extent;

[0093] (4) Some time periods or frequency bands may be affected by short-term strong interference, resulting in partial loss of information.

[0094] Considering that the voiceprint of the transformer site has the characteristics of non-single sound source and non-point sound source, and the sound signal collected by the sensor is actually the superposition of different sound sources after direct and reflected, that is, under the same working conditions, the voiceprint data collected after moving the sensor position may appear as the weighted superposition of voiceprint data at other positions. In order to realize the generation of transformer site voiceprint data in more reasonable scenarios, reference is made to Figure 4 The specific steps of the flow chart based on multi-sample mixed weighted enhancement and expansion of voiceprint spectrogram are as follows:

[0095] Step 301, select a reference voiceprint sample and calculate the effective value;

[0096] Step 302, respectively calculating two or more sample amplitude spectra, and retaining the phase spectrum of the reference sample;

[0097] Step 303: Based on the sample amplitude spectrum, extract the peak amplitude sequence for all samples according to the set peak conditions. The specific peak conditions can be adjusted according to the actual situation. For example, only the maximum value within a certain range is considered as the peak value, so as to avoid the random peaks in the background noise signal being enhanced and causing signal pollution.

[0098] Step 304: performing random weighting processing on each peak amplitude sequence to obtain an enhanced peak amplitude sequence;

[0099] Step 305, summing up the enhanced peak amplitude sequence according to the sample weights to obtain a mixed peak sequence;

[0100] Step 306, perform IFFT calculation according to the effective value and phase spectrum of the reference sequence and the amplitude spectrum of the peak sequence after mixing to obtain multi-sample mixed enhanced samples;

[0101] Figure 7 For Figure 5 and Figure 6 The effect diagram after multi-sample mixed weighted directional enhancement expansion. Since sample 2 is used as the reference signal and is given a higher weight, the overall expanded sample is Figure 6 The similarity is high, but it can also be seen that the expanded sample has a lot of detailed information of sample 1 within the low frequency band of 200Hz.

[0102] Step 4: Based on the voiceprint data after data expansion, combined with the fundamental frequency and frequency distribution characteristics of the transformer, feature extraction is performed to obtain the characteristics of the transformer partial discharge defect;

[0103] There are dozens of time domain, frequency domain and time-frequency domain features of sound signals. Based on the specific characteristics of the transformer, the present invention proposes a new feature extraction method for the partial discharge defect category, combined with the transformer fundamental frequency and frequency distribution characteristics, which specifically includes:

[0104] Step 401: Filter the voiceprint data after data expansion;

[0105] In this embodiment, a 10000 Hz high pass filter is used for filtering, and the filtering formula is:

[0106]

[0107] Wherein, x(n) is the original audio signal, a and b are filter coefficients, y(n) is the filtered signal, N and P represent the number of terms in the filter, also known as the order of the filter; x(nk) is the input time series, including x(0), x(1), x(2), ..., x(nk), y(nJ) is the output time series, including y(0), y(1), y(2), ..., y(nk);

[0108] Step 402: perform Hilbert envelope calculation on the filtered signal to obtain an envelope signal:

[0109]

[0110] Where H(jw) is a 90° phase shift filter. The phase shifted signal is obtained by performing phase shift filtering on the filtered time domain data y(n), and then the phase shifted signal is combined with the original time domain signal to form the analytical signal (y e = abs(y(n)+y(n)*h(n))), The modulus value of the analytical signal is the envelope of the corresponding original waveform.

[0111] Step 403: Perform FFT calculation on the envelope signal to obtain the signal amplitude spectrum:

[0112]

[0113] Where, Y(k) is the complex spectrum at frequency k, N is the total number of sampling points in the time domain, and the corresponding amplitude spectrum can be obtained by taking the modulus value of Y(k)*2 / N;

[0114] Step 404, combining the high frequency envelope spectrum of the voiceprint signal to calculate the power frequency correlation;

[0115] Resonance demodulation and power frequency correlation: Because there are more low-frequency interferences in the transformer sound, including the main noise of transformer operation is also in the low-frequency band, the discharge signal is a signal with a very wide frequency and short duration, and the signal-to-noise ratio in the high-frequency band is relatively high. Figure 8 and Fig. 9 The time domain diagram and frequency domain characteristic diagram of the partial discharge signal show an obvious peak effect in the high frequency band. When a partial discharge fault occurs, there will be modulation components related to the transformer power frequency in the high frequency band of the soundprint. Therefore, the proportion of 50Hz and 100Hz components in the envelope spectrum of the frequency band above 10000Hz in the 50-500Hz band is calculated:

[0116]

[0117] Where Y() is the envelope spectrum of the high frequency band (usually above 10000Hz) of the voiceprint signal, and the index is the corresponding frequency. This feature is mainly achieved by envelope demodulating the high frequency band of the transformer voiceprint signal to avoid the interference of high-order harmonics of the low-frequency power frequency. Then, FFT calculation is performed on the envelope signal to analyze whether there is an impact signal (i.e., discharge pulse) with the power frequency or power frequency harmonic as the period in the high frequency band. When there is a discharge occurring in the power frequency period, a large number of 50Hz and harmonic components will appear in the working demodulation spectrum, and the power frequency correlation value will increase sharply.

[0118] Step 5: Determine whether there is a partial discharge defect based on the characteristics of the transformer partial discharge defect;

[0119] In this embodiment, whether there is a partial discharge defect is determined based on the value of the resonance demodulation power frequency correlation PD_CORR. For example, if the calculated value is between 0.5 and 0.8, it can be determined that this type of defect exists.

[0120] Embodiment 2

[0121] This embodiment provides a transformer partial discharge defect soundprint feature extraction system, including:

[0122] A data acquisition module, which is used to obtain the original voiceprint data of the transformer;

[0123] A noise reduction module, which is used to reduce the noise of the original voiceprint data of the transformer based on spectral subtraction to obtain the noise-reduced voiceprint data;

[0124] A data expansion module, which is used to expand the voiceprint data after noise reduction to obtain the voiceprint data after data expansion;

[0125] A feature extraction module is used to extract features based on the voiceprint data after data expansion, combined with the fundamental frequency and frequency distribution characteristics of the transformer, to obtain the resonant demodulation power frequency correlation;

[0126] The defect judgment module is used to judge whether there is a partial discharge defect based on the resonance demodulation power frequency correlation.

[0127] In this embodiment, the implementation process of each module is consistent with the content of the first embodiment.

[0128] Embodiment 3

[0129] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps in the transformer partial discharge defect soundprint feature extraction method as described above are implemented.

[0130] Embodiment 4

[0131] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the transformer partial discharge defect voiceprint feature extraction method as described above are implemented.

[0132] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. 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 disk storage and optical storage, etc.) containing computer-usable program code.

[0133] 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 flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, 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 flowchart and / or block diagram. 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.

[0134] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate 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 A function specified in one or more boxes.

[0135] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0136] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0137] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for extracting the soundprint features of transformer partial discharge defects, characterized in that: The steps include: Get the original voiceprint data of the transformer; Based on spectral subtraction, the original voiceprint data of the transformer is de-noised to obtain the de-noised voiceprint data; Performing data expansion on the denoised voiceprint data to obtain voiceprint data after data expansion; Based on the voiceprint data after data expansion, combined with the fundamental frequency and frequency distribution characteristics of the transformer, feature extraction is performed to obtain the resonance demodulation power frequency correlation; Based on the resonance demodulation power frequency correlation, it is determined whether there is a partial discharge defect.

2. The transformer partial discharge defect soundprint feature extraction method according to claim 1, characterized in that: Based on the improved spectral subtraction method, the original voiceprint data of the transformer is denoised to obtain the denoised voiceprint data, which specifically includes: finding the signal in the non-noise interference period as the target signal, and then spectrally replacing the signal in the noise interference period, retaining the original phase information of the signal, and replacing it with the amplitude information of the frame with the minimum energy in the interference period.

3. The transformer partial discharge defect soundprint feature extraction method according to claim 2, characterized in that: The method of searching for a signal in a non-noise interference period as a target signal, and then performing spectrum replacement on the signal in a noise interference period, retaining the original phase information of the signal and replacing it with the amplitude information of the frame with the minimum energy in the interference period, includes: The original voiceprint signal is divided into frames, and the background voiceprint is estimated to obtain the amplitude spectrum information of the background voiceprint signal; According to the background voiceprint estimation result, the signal-to-noise ratio of each frame of data is calculated, and the sequence number of the frame to be denoised is determined according to the set noise consistency threshold; Perform FFT calculation on the noise segment signal based on the sequence number of the frame to be denoised to obtain the phase spectrum information of the noise segment signal; The amplitude spectrum information of the background voiceprint signal and the phase spectrum information of the noise segment signal are fused to obtain the fused full spectrum information; Based on the fused full spectrum information, IFFT calculation is performed to obtain the denoised time domain signal.

4. The transformer partial discharge defect soundprint feature extraction method according to claim 1, characterized in that: The step of performing data expansion on the noise-reduced voiceprint data to obtain the voiceprint data after data expansion specifically includes: Select a benchmark voiceprint sample and calculate the effective value; Calculate the amplitude spectrum of two or more samples respectively, and retain the phase spectrum of the reference sample; Based on the sample amplitude spectrum, the peak amplitude sequence is extracted for all samples according to the set peak conditions to obtain the peak amplitude sequence; Performing random weighting processing on each peak amplitude sequence to obtain an enhanced peak amplitude sequence; The enhanced peak amplitude sequence is summed up according to the sample weights to obtain a mixed peak sequence; IFFT calculation is performed according to the effective value and phase spectrum of the reference sequence and the amplitude spectrum of the peak sequence after mixing to obtain the multi-sample mixed enhanced samples.

5. The transformer partial discharge defect soundprint feature extraction method according to claim 1, characterized in that: Before data expansion of the voiceprint data after noise reduction, the signal-to-noise ratio of the transformer body voiceprint is calculated based on the time domain signal after noise reduction, and the corresponding data processing is performed according to the comparison between the signal-to-noise ratio of the transformer body voiceprint and the set threshold, including obtaining the body voiceprint signal-to-noise ratio according to the transformer body voiceprint and the total value ratio. If the signal-to-noise ratio is greater than the set threshold, the sound pressure level, harmonic ratio, high-frequency energy proportion, entropy spectrum, and demodulated power frequency correlation characteristics are calculated according to the body frequency; If the signal-to-noise ratio is less than the set threshold, the data in that segment or the data in that time period will be discarded.

6. The transformer partial discharge defect soundprint feature extraction method according to claim 1, characterized in that: The process of feature extraction of transformer partial discharge defects includes: The voiceprint data after data expansion is filtered and then the envelope is calculated to obtain an envelope signal; After performing FFT calculation on the envelope signal, the resonance demodulation power frequency correlation is calculated in combination with the high-frequency envelope spectrum of the voiceprint signal; the resonance demodulation power frequency correlation is the proportion of 50Hz and 100Hz components in the envelope spectrum of the frequency band above 10000Hz in the 50-500Hz frequency band.

7. Transformer partial discharge defect soundprint feature extraction system, characterized by: include: A data acquisition module, which is used to obtain the original voiceprint data of the transformer; A noise reduction module, which is used to reduce the noise of the original voiceprint data of the transformer based on spectral subtraction to obtain the noise-reduced voiceprint data; A data expansion module, which is used to expand the voiceprint data after noise reduction to obtain the voiceprint data after data expansion; A feature extraction module is used to extract features based on the voiceprint data after data expansion, combined with the fundamental frequency and frequency distribution characteristics of the transformer, to obtain the resonant demodulation power frequency correlation; The defect judgment module is used to judge whether there is a partial discharge defect based on the resonance demodulation power frequency correlation.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the transformer partial discharge defect voiceprint feature extraction method as described in any one of claims 1 to 6 are implemented.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the transformer partial discharge defect voiceprint feature extraction method according to any one of claims 1 to 6 are implemented.

10. A program product, the program product being a computer program product, comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps in the transformer partial discharge defect soundprint feature extraction method according to any one of claims 1 to 6 are implemented.

Images