A method and system for extracting dual-time and dual-frequency parameters of pulse current waveform

Abstract

The present invention relates to a method and system for extracting dual-time and dual-frequency parameters of a pulse current waveform, the method comprising: step 1) using lifting wavelet to denoise a single pulse current waveform in a pulse group and normalize the amplitude to form a time domain waveform 1; step 2) performing Hilbert transform on the time domain waveform 1 to obtain a time domain waveform 2, and performing fast Fourier transform and short-time Fourier transform on the time domain waveform 1 to normalize the amplitude respectively to obtain frequency domain waveform 1 and frequency domain waveform 2; step 3) calculating the time standard deviation and frequency standard deviation of the time domain waveform and the frequency domain waveform to realize the distribution of dual-time and dual-frequency standard deviation characteristic parameters corresponding to a single pulse current waveform; step 4) completing all waveform processing of the pulse group, thereby obtaining a dual-time and dual-frequency standard deviation characteristic parameter distribution plane of the pulse current waveform-time series. Compared with the prior art, the present invention has the advantages of stronger robustness, simple algorithm and convenient implementation.

Description

Technical Field

[0001] The invention relates to partial discharge detection technology, in particular to a method and system for extracting dual-time and dual-frequency parameters of pulse current waveform for ultra-wideband detection. Background Art

[0002] When detecting or monitoring partial discharge (PD) of insulation of high-voltage equipment such as AC or converter transformers, the traditional detection system based on PD pulse peak-time series will have multiple PD sources (including two) or abnormal interference noise sources. If the signal source spectrum overlaps, the data obtained will be a random aliased peak-time / phase sequence, which is not conducive to the judgment of actual engineering test results. In view of the above working conditions, Bologna University in Italy began to propose pulse source detection technology based on ultra-wideband detection in 2002 and Xi'an Jiaotong University in China began to propose pulse source detection technology based on ultra-wideband detection in 2008. That is, the traditional pulse peak-time series detection is changed to pulse current waveform-time series detection, that is, a single pulse current waveform and its acquisition time point (phase) are recorded; a certain "method" is used to quickly classify the obtained mixed original pulse group, and each sub-class pulse group with "self-similar" pulse composition is converted into a peak-time series, which not only solves the aliasing problem of the peak-time series, but also can detect and identify multiple PD sources with interference pulse sources.

[0003] The key to the implementation of the ultra-wideband technology is to use a certain "method" to quickly classify the obtained mixed original pulse groups. It is divided into two parts: ① is the pulse waveform characteristic parameter extraction method; ② is the clustering technology based on the characteristic parameter distribution. Among them, for the first part, the "Fast Classification of Multiple Partial Discharge Pulse Groups Based on Waveform Nonlinear Mapping" published in the Journal of the Chinese Society of Electrotechnical Engineering in March 2009 proposed the following Figure 1 The nonlinear mapping method of the pulse waveform shown in FIG. That is, the extraction results are distributed and displayed in a two-dimensional plane or a three-dimensional space, and then the pulse group is separated into sub-pulse groups with their own characteristics by means of intelligent clustering analysis (the technology in part ②), thereby realizing the separation of multiple PD sources and noise sources.

[0004] However, in actual engineering applications, due to the influence of the propagation path, the pulse current waveforms generated by the noise source or PD source obtained by ultra-wideband detection have very similar working conditions in terms of time domain waveform representation. Figure 1 The pulse waveform nonlinear mapping method shown cannot meet the needs of later separation of multiple PD sources and noise sources using means such as intelligent clustering analysis. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a method and system for extracting dual-time and dual-frequency parameters of a pulse current waveform.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] According to one aspect of the present invention, a method for extracting dual-time and dual-frequency parameters of a pulse current waveform is provided, the method comprising:

[0008] Step 1) using lifting wavelet to denoise a single pulse current waveform in the pulse group and normalize the amplitude to form a time domain waveform 1;

[0009] Step 2) performing Hilbert transform on the time domain waveform 1 to obtain the time domain waveform 2, and performing fast Fourier transform and short-time Fourier transform on the time domain waveform 1 to normalize the amplitude respectively to obtain frequency domain waveform 1 and frequency domain waveform 2, thereby obtaining dual-time dual-frequency 4-type waveforms of a single pulse current waveform;

[0010] Step 3) The time standard deviation and frequency standard deviation of the time domain waveform and the frequency domain waveform are calculated to achieve the dual-time and dual-frequency standard deviation characteristic parameter distribution corresponding to the single pulse current waveform;

[0011] Step 4) Repeat steps 1) to 3) to complete all waveform processing of the pulse group, thereby obtaining a dual-time and dual-frequency standard deviation characteristic parameter distribution plane of the pulse current waveform-time series.

[0012] As a preferred technical solution, the pulse group in step 1) is a mixed pulse current waveform-time series formed by the pulse source of the PD source and the noise source for ultra-wideband detection.

[0013] According to another aspect of the present invention, a system for the dual-time and dual-frequency parameter extraction method of a pulse current waveform is provided, the system comprising a pulse current waveform-time series module of a mixture of a PD source and a noise source obtained by ultra-wideband detection, a lifting wavelet denoising module, a pulse time domain waveform module, a Hilbert transform module, a fast Fourier transform module, a short-time Fourier transform module, a dual-time and dual-frequency four-type waveform module, and a time-frequency standard deviation distribution module;

[0014] The pulse current waveform-time series module of the mixture of PD source and noise source obtained by the ultra-wideband detection, the lifting wavelet denoising module, and the pulse time domain waveform module are connected in sequence, the pulse time domain waveform module is respectively connected to the Hilbert transform module, the fast Fourier transform module, and the short-time Fourier transform module, and the dual-time and dual-frequency four-type waveform module is respectively connected to the Hilbert transform module, the fast Fourier transform module, the short-time Fourier transform module and the time-frequency standard deviation distribution module.

[0015] As a preferred technical solution, the pulse current waveform-time series module of the mixed PD source and noise source obtained by the ultra-wideband detection is a data acquisition device with an analog bandwidth of tens of MHz and a sampling rate of 100MS / s or above, and a coupling device with a frequency response of tens of MHz or above that meets the nanosecond level PD ultra-wideband detection is used to record a single pulse current time domain waveform and the pulse current waveform-time series at the corresponding triggering moment based on the pulse waveform triggering technology. j (t).

[0016] As a preferred technical solution, the lifting wavelet denoising module uses lifting wavelet to denoise a single original pulse current waveform in the pulse group and normalizes the amplitude to form a time domain waveform 1, that is,

[0017] As a preferred technical solution, the pulse time domain waveform module is used to store the time domain waveform 1 As an object processed by subsequent modules.

[0018] As a preferred technical solution, the Hilbert transform module uses the Hilbert transform algorithm to implement the HT of the time domain waveform 1, that is, H j (t), after amplitude normalization, we get

[0019] As a preferred technical solution, the fast Fourier transform module uses a discrete Fourier transform algorithm to implement the FFT of the time domain waveform 1. Amplitude normalization processing is performed to obtain

[0020] As a preferred technical solution, the short-time Fourier transform module uses a short-time Fourier transform algorithm to implement the STFT of the time domain waveform 1. Amplitude normalization processing is obtained

[0021] As a preferred technical solution, the dual-time dual-frequency 4-type waveform module is used to store the time domain waveform 1 And time domain waveform 2 And frequency domain waveform 1 and frequency domain waveform 2

[0022] The time-frequency standard deviation distribution module is used to store the time domain waveform 1 of the dual-time dual-frequency 4-type waveform module. And time domain waveform 2 And frequency domain waveform 1 and frequency domain waveform 2 Calculation is performed to obtain the characteristic parameter distribution plane of the pulse current waveform-time series.

[0023] Compared with the prior art, the present invention has the following advantages:

[0024] 1. This method makes full use of the distribution characteristics of the non-stationary representation of the pulse current waveform signal in the time domain and frequency domain, and uses two time domain waveforms and two frequency domain waveforms to participate in the standard deviation calculation.

[0025] 2. Compared with the current characteristic parameter advance method based only on the pulse current time domain and frequency domain waveform (such as the equivalent time-frequency method), it has the advantages of stronger robustness, simple algorithm and easy implementation. It is suitable for constructing a pulse group fast classification technology for ultra-wideband PD detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is the current pulse waveform nonlinear mapping method;

[0027] Figure 2 It is the main module of the method shown in the present invention;

[0028] Figure 3 The following is an example of the acquisition results of the same pulse sampling duration of 1μs, 2μs, 3μs and 4μs at a sampling rate of 100MS / s.

[0029] Figure 4 The original waveform and time-frequency domain diagram of a typical single PD pulse current of Example 1 (tip plate discharge defect) obtained by the method of the present invention in the DC withstand voltage ultra-wideband detection, wherein (a) is the original waveform, (b) is the time domain waveform 1 obtained after lifting wavelet denoising and amplitude normalization, (c) is the time domain waveform 2 obtained after HT transformation and amplitude normalization, (d) is the frequency domain waveform 1 obtained after FFT transformation and amplitude normalization, and (e) is the frequency domain waveform 2 obtained after STFT transformation and amplitude normalization;

[0030] Figure 5 The original waveform and time-frequency domain diagram of a typical single PD pulse current of Example 2 (internal air gap discharge defect) obtained by the method of the present invention in the DC withstand voltage ultra-wideband detection, wherein (a) is the original waveform, (b) is the time domain waveform 1 obtained after lifting wavelet denoising and amplitude normalization, (c) is the time domain waveform 2 obtained after HT transformation and amplitude normalization, (d) is the frequency domain waveform 1 obtained after FFT transformation and amplitude normalization, and (e) is the frequency domain waveform 2 obtained after STFT transformation and amplitude normalization;

[0031] Figure 6The original waveform and time-frequency domain diagram of a typical single PD pulse current of Example 3 (surface discharge defect) obtained by the method of the present invention in the DC withstand voltage ultra-wideband detection, wherein (a) is the original waveform, (b) is the time domain waveform 1 obtained after lifting wavelet denoising and amplitude normalization, (c) is the time domain waveform 2 obtained after HT transformation and amplitude normalization, (d) is the frequency domain waveform 1 obtained after FFT transformation and amplitude normalization, and (e) is the frequency domain waveform 2 obtained after STFT transformation and amplitude normalization;

[0032] Figure 7 It is a lifting wavelet transform decomposition and reconstruction operation block diagram used in the method shown in the present invention;

[0033] Figure 8 This is a flow chart of pulse data processing in the method of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work should fall within the scope of protection of the present invention.

[0035] The present invention attempts to propose a method for extracting dual-time and dual-frequency parameters of pulse current waveforms for ultra-wideband detection based on practical applications. That is, for ultra-wideband detection, a mixed pulse current waveform-time series, i.e., a pulse group, formed by pulse sources such as PD source and noise source is obtained, and a single pulse current waveform in the pulse group is denoised by lifting wavelet and amplitude normalized to form a time domain waveform 1; then the time domain waveform 1 is subjected to Hilbert transform to obtain time domain waveform 2, and the time domain waveform 1 is subjected to fast Fourier transform and short-time Fourier transform to obtain frequency domain waveform 1 and frequency domain waveform 2 after amplitude normalization, thereby obtaining four types of dual-time and dual-frequency waveforms of a single pulse current waveform; then the time standard deviation and frequency standard deviation of the time domain waveform and the frequency domain waveform are calculated to realize the distribution of dual-time and dual-frequency standard deviation characteristic parameters corresponding to a single pulse current waveform; the above process processes all waveforms of the pulse group, thereby obtaining a dual-time and dual-frequency standard deviation characteristic parameter distribution plane of the pulse current waveform-time series, which is used to construct a rapid classification technology for pulse groups for ultra-wideband PD detection.

[0036] This method makes full use of the distribution characteristics of the non-stationary representation of the pulse current waveform signal in the time domain and frequency domain. It has the advantages of stronger robustness, simple algorithm and easy implementation than the current feature parameter advance method based only on the pulse current time domain and frequency domain waveform (such as the equivalent time-frequency method). It is suitable for constructing a pulse group rapid classification technology for ultra-wideband PD detection. Specific embodiments

[0038] The present invention provides a method for extracting dual-time and dual-frequency parameters of pulse current waveform for ultra-wideband detection. That is, for ultra-wideband detection, a mixed pulse current waveform formed by pulse sources such as PD source and noise source is obtained - time series, i.e., pulse group p j (t), for the pulse group p j (t) Single pulse current waveform (in p 1 (t) as an example) using lifting wavelet to denoise and normalize the amplitude to form the time domain waveform 1, that is, Then perform Hilbert transform (HT) on the time domain waveform 1 to obtain the time domain waveform 2, namely And the time domain waveform 1 is subjected to fast Fourier transform (FFT) and short-time Fourier transform (STFM) to normalize the amplitude and obtain the frequency domain waveform 1 and the frequency domain waveform 2, namely P 1 1 (f) and P 1 2 (f) Thus, the dual-time and dual-frequency 4-type waveforms of a single pulse current waveform are obtained ( P 1 1 (f) and P 1 2 (f)); then use the time standard deviation and frequency standard deviation of the time domain waveform and frequency domain waveform to calculate the dual-time and dual-frequency standard deviation characteristic parameter distribution corresponding to a single pulse current waveform (T 1 1 , T 1 2 、F 1 1 and F 1 2 ); The above process processes all the waveforms of the pulse group, thereby obtaining the distribution plane of the dual-time and dual-frequency standard deviation characteristic parameters of the pulse current waveform-time series A fast pulse group classification technique for building ultra-wideband PD detection.

[0039] like Figure 2 As shown, the system of the present invention for the method for extracting dual-time and dual-frequency parameters of pulse current waveform comprises a pulse current waveform-time series module 10 of a mixture of a PD source and a noise source obtained by ultra-wideband detection, a lifting wavelet denoising module 11, a pulse time domain waveform module 12, a Hilbert transform module 13, a fast Fourier transform module 14, a short-time Fourier transform module 15, a dual-time and dual-frequency four-type waveform module 16 and a time-frequency standard deviation distribution module 17;

[0040] The pulse current waveform-time series module 10 of the mixed PD source and noise source obtained by the ultra-wideband detection, the lifting wavelet denoising module 11, and the pulse time domain waveform module 12 are connected in sequence, the pulse time domain waveform module 12 is respectively connected to the Hilbert transform module 13, the fast Fourier transform module 14, and the short-time Fourier transform module 15, and the dual-time dual-frequency four-type waveform module 16 is respectively connected to the Hilbert transform module 13, the fast Fourier transform module 14, the short-time Fourier transform module 15 and the time-frequency standard deviation distribution module 17.

[0041] The pulse current waveform-time series module 10 of the mixed PD source and noise source obtained by the ultra-wideband detection is a data acquisition device with an analog bandwidth of tens of MHz and a sampling rate of 100MS / s or above. The coupling device with a frequency response of tens of MHz or above that meets the nanosecond level PD ultra-wideband detection records the single pulse current time domain waveform and the pulse current waveform-time series at the corresponding triggering moment, i.e., the pulse group p j (t).

[0042] The definition is as follows:

[0043]

[0044] Where:

[0045] j——jth pulse (j=1, 2, ... N, N is the total number of pulse current waveforms contained in the pulse group);

[0046] k——The pulse current waveform consists of k points, and the number of points is determined by the sampling rate f s *Determined by sampling duration; Figure 3 The following are examples of acquisition results for the same pulse sampling duration of 1μs, 2μs, 3μs, and 4μs at a sampling rate of 100MS / s, that is, 100 points, 200 points, 300 points, and 400 points);

[0047] a i ——The amplitude corresponding to the i-th point (mV or V);

[0048] Δt(i-1)——The time corresponding to the i-th point (Δt is the sampling time interval, Δt=1 / f s ).

[0049] Figure 4 (a) Figure 5 (a) and Figure 6 (a) An example of a pulse current waveform-time series obtained by ultra-wideband detection with a sampling rate of 2.5 GS / s and an analog bandwidth of 1 GHz, i.e., a single original pulse current waveform contained in a pulse group.

[0050] The lifting wavelet denoising module 11 uses lifting wavelet to denoise a single original pulse current waveform in the pulse group and normalizes the amplitude to form a time domain waveform 1, that is, Lifting wavelet transform decomposition and reconstruction operations such as Figure 7 As shown, it contains 5 steps defined as follows:

[0051] 1) Splitting. The original signal p = p j (t) is split into even-numbered sample sequences With odd sample sequence

[0052] 2) Prediction. (0) The sampling prediction operator A predicts y (0) ;

[0053]

[0054] Where: a r ——prediction coefficient of prediction operator A, N——the number of predictor coefficients.

[0055] 3) Update. (0) Apply update operator B to update x (0) ;

[0056]

[0057] Where: b r ——Update coefficient of update operator B, M——the number of new coefficients.

[0058] 4) Reconstruction. Sampling equations (2) and (3) are inversely calculated to obtain and

[0059] 5) Merge. Merge the even-numbered sample sequence With odd sample sequence Merge Generation

[0060] The pulse time domain waveform module 12 is used to store the time domain waveform 1 As an object processed by subsequent modules. Figure 4 (a) Figure 5 (a) and Figure 6 The time domain waveform of the single original pulse current shown in (a) is obtained after being processed by the 11-lift wavelet denoising module. Figure 4 (b) Figure 5 (b) and Figure 6 As shown in (b), they are all stored in the pulse time domain waveform module 12.

[0061] The Hilbert transform module 13 uses the Hilbert transform algorithm to implement the HT of the time domain waveform 1, that is, H j (t), after amplitude normalization, we get The algorithm is as follows:

[0062] 1) HT.

[0063]

[0064] 2) Amplitude normalization.

[0065]

[0066] Figure 4 (b) Figure 5 (b) and Figure 6 (b) The time domain waveform 1 is obtained by the Hilbert transform (HT) module to obtain the time domain waveform 2, that is, like Figure 4 (c) Figure 5 (c) and Figure 6 (c) as shown.

[0067] The fast Fourier transform module 14 uses a discrete Fourier transform (DFT) algorithm to implement the FFT of the time domain waveform 1. Amplitude normalization processing is obtained The algorithm is as follows:

[0068] 1) DFT.

[0069]

[0070] 2) Amplitude normalization.

[0071]

[0072] Figure 4 (b) Figure 5 (b) and Figure 6 (b) The time domain waveform 1 is obtained by the fast Fourier transform (FFT) module to obtain the frequency domain waveform 1, that is, like Figure 4 (d) Figure 5 (d) and Figure 6 (d) as shown.

[0073] The short-time Fourier transform module 15 uses a short-time Fourier transform algorithm to implement the STFT of the time domain waveform 1. Amplitude normalization processing is obtained The algorithm is as follows:

[0074] 1) STFT.

[0075]

[0076] Where: h(t) is the short-time analysis window centered at t=0 and f=0.

[0077] 2) Amplitude normalization.

[0078]

[0079] Figure 4 (b) Figure 5 (b) and Figure 6 The frequency domain waveform 2 obtained by the short-time Fourier transform (STFM) module of the time domain waveform 1 shown in (b) is like Figure 4 (e) Figure 5 (e) and Figure 6 (e) as shown.

[0080] The dual-time dual-frequency 4-type waveform module 16 is used to store the time domain waveform 1 And time domain waveform 2 And frequency domain waveform 1 and frequency domain waveform 2

[0081] The time-frequency standard deviation distribution module 17 is used to distribute the time domain waveform 1 stored in the dual-time dual-frequency 4-type waveform module 16. And time domain waveform 2 And frequency domain waveform 1 and frequency domain waveform 2 Calculation is performed to obtain the characteristic parameter distribution plane of the pulse current waveform-time series.

[0082] The time-frequency standard deviation distribution module 17, wherein the algorithm for calculating the time-frequency standard deviation for the pulse signal p(t) and its frequency distribution P(f) is defined as follows:

[0083]

[0084] Where: |p(t)| 2 and |P(f)| 2 ——probability density;

[0085] ——The energy of the signal p(t).

[0086] Formula (10) is for the collected signal p j (t) and its spectrum P jDiscretization of (f):

[0087]

[0088] Using (11), the time domain waveform 1 and the time domain waveform 2 stored in the 16 dual-time dual-frequency 4-type waveform modules are and And frequency domain waveform 1 and frequency domain waveform 2, namely and You can get:

[0089]

[0090] Finally, the characteristic parameter distribution plane of the pulse current waveform-time series is obtained N is the pulse group p j (t) contains the total number of pulse current waveforms.

[0091] The processing flow chart of pulse current waveform data in the method of the present invention is as follows: Figure 8 shown.

[0092] Figures 4 to 5 The following is a pulse current waveform obtained by ultra-wideband detection with a sampling rate of 2.5GS / s and an analog bandwidth of 1GHz. A typical single pulse waveform in the time series is as follows: Figure 8 Examples of time domain waveforms 1, 2 and frequency domain waveforms 1, 2 corresponding to the processing flow shown.

[0093] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A system for extracting dual-time and dual-frequency parameters of pulse current waveform, It is characterized in that The method includes: Step 1) using lifting wavelet to denoise a single pulse current waveform in the pulse group and normalize the amplitude to form a time domain waveform 1; Step 2) performing Hilbert transform on the time domain waveform 1 to obtain the time domain waveform 2, and performing fast Fourier transform and short-time Fourier transform on the time domain waveform 1 to normalize the amplitude respectively to obtain frequency domain waveform 1 and frequency domain waveform 2, thereby obtaining dual-time dual-frequency 4-type waveforms of a single pulse current waveform; Step 3) The time standard deviation and frequency standard deviation of the time domain waveform and the frequency domain waveform are calculated to achieve the dual-time and dual-frequency standard deviation characteristic parameter distribution corresponding to the single pulse current waveform; Step 4) repeating steps 1) to 3) to complete all waveform processing of the pulse group, thereby obtaining a dual-time and dual-frequency standard deviation characteristic parameter distribution plane of the pulse current waveform-time series; The system comprises a pulse current waveform-time series module (10) of a mixture of a PD source and a noise source obtained by ultra-wideband detection, a lifting wavelet denoising module (11), a pulse time domain waveform module (12), a Hilbert transform module (13), a fast Fourier transform module (14), a short-time Fourier transform module (15), a dual-time and dual-frequency four-type waveform module (16) and a time-frequency standard deviation distribution module (17); The pulse current waveform-time series module (10) of the mixed PD source and noise source obtained by the ultra-wideband detection, the lifting wavelet denoising module (11), and the pulse time domain waveform module (12) are connected in sequence; the pulse time domain waveform module (12) is respectively connected to the Hilbert transform module (13), the fast Fourier transform module (14), and the short-time Fourier transform module (15); the dual-time dual-frequency four-type waveform module (16) is respectively connected to the Hilbert transform module (13), the fast Fourier transform module (14), the short-time Fourier transform module (15), and the time-frequency standard deviation distribution module (17); The dual-time dual-frequency 4-type waveform module (16) is used to store time domain waveforms. and time domain waveform And the frequency domain waveform and frequency domain waveform The time-frequency standard deviation distribution module (17) distributes the time domain waveform stored in the dual-time dual-frequency four-type waveform module (16). and time domain waveform And the frequency domain waveform and frequency domain waveform Calculations are performed to obtain the characteristic parameter distribution plane of the pulse current waveform-time series, which is used to construct a rapid classification of pulse groups for ultra-wideband PD detection.

2. The system according to claim 1, It is characterized in that The pulse group in step 1) is a mixed pulse current waveform-time sequence formed by the pulse sources of the PD source and the noise source for ultra-wideband detection.

3. The system according to claim 1, It is characterized in that The pulse current waveform-time series module (10) of the mixed PD source and noise source obtained by the ultra-wideband detection is a data acquisition device with an analog bandwidth of tens of MHz and a sampling rate of 100MS / s or more. The coupling device with a frequency response of tens of MHz or more that meets the nanosecond level PD ultra-wideband detection records a single pulse current time domain waveform and a pulse current waveform-time series corresponding to the triggering moment, i.e., a pulse group p, based on a pulse waveform triggering technology. j (t).

4. The system according to claim 1, It is characterized in that The lifting wavelet denoising module (11) uses lifting wavelet to denoise a single original pulse current waveform in the pulse group and normalizes the amplitude to form a time domain waveform 1, that is, 5. The system according to claim 1, It is characterized in that The pulse time domain waveform module (12) is used to store the time domain waveform As an object processed by subsequent modules.

6. The system according to claim 1, It is characterized in that The Hilbert transform module (13) uses the Hilbert transform algorithm to implement the HT of the time domain waveform 1, that is, H j (t), after amplitude normalization, we get 7. The system according to claim 1, It is characterized in that The fast Fourier transform module (14) uses a discrete Fourier transform algorithm to implement the FFT of the time domain waveform 1. Amplitude normalization processing is performed to obtain 8. The system according to claim 1, It is characterized in that The short-time Fourier transform module (15) uses a short-time Fourier transform algorithm to implement the STFT of the time domain waveform 1. Amplitude normalization processing is performed to obtain

Images