Wavelet decomposition-based multi-layer adaptive ripple recognition method and related equipment

Abstract

The invention relates to the technical field of electrical detection, in particular to a multilayer adaptive ripple recognition method based on wavelet decomposition and related equipment. The method comprises the following steps: acquiring a high-voltage signal; the obtained high-voltage signal is preprocessed; performing signal amplification and digital processing on the preprocessed high-voltage signal through a signal amplification technology and an analog-to-digital conversion technology to obtain a preliminary digital signal; processing the preliminary digital signal through a wavelet decomposition algorithm to obtain direct current ripple data; performing feature extraction on the direct current ripple data through a wavelet basis to obtain an insulation state feature; and comparing the insulation state characteristics with a self-adaptive threshold value, and starting an early warning operation according to a comparison result. According to the method, the number of layers of wavelet decomposition and an analysis frequency band can be adaptively selected, energy evaluation and frequency domain positioning are carried out on decomposed detail components of each layer, effective layers containing ripple characteristics are dynamically identified, and online detection, alarm and data uploading are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical detection, and in particular to a multi-layer adaptive ripple identification method based on wavelet decomposition and related equipment. BACKGROUND

[0002] In a DC high-voltage device such as a withstand voltage tester or a power supply performance analysis system, the stability of the output voltage is one of the core indicators. The ideal DC voltage should be constant, but in actual applications, due to the influence of factors such as rectifier circuit, electromagnetic interference, load change, and switching device oscillation, the output voltage often contains periodic or non-periodic high-frequency components, i.e., so-called DC ripple. In existing ripple measurement methods: the oscilloscope detection method for displaying ripple components through AC coupling is relatively intuitive but not suitable for online continuous detection; the band-pass filtering method for extracting ripple through a fixed frequency band filter is difficult to cope with frequency changes; the Fourier transform method (FFT) can analyze frequency energy but cannot accurately locate the time-domain characteristics of non-stationary signals; the hardware integration or differentiation method is simple to implement but has poor anti-interference ability and weak adaptability. These ripple measurement methods all have necessary shortcomings in specific situations, especially in non-stationary ripple scenarios such as significant changes in ripple amplitude-frequency characteristics during voltage rise. Existing methods cannot effectively extract the instantaneous characteristics of the ripple signal, resulting in unstable detection and serious false positives and false negatives.

[0003] In summary, the technical problems in the related art need to be improved. SUMMARY

[0004] The main purpose of the embodiments of the present application is to propose a multi-layer adaptive ripple identification method based on wavelet decomposition and related equipment, which can adaptively select the number of layers of wavelet decomposition and the analysis frequency band, evaluate the energy of the detail components after each layer of decomposition, and dynamically identify the effective layer containing ripple characteristics, and realize online detection, alarm, and data uploading.

[0005] To achieve the above purpose, one aspect of the embodiments of the present application proposes a multi-layer adaptive ripple identification method based on wavelet decomposition, which comprises the following steps: obtaining a high-voltage signal; preprocessing the obtained high-voltage signal, the preprocessing including voltage reduction processing, buffer processing, and filtering processing; performing signal amplification and digitization processing on the preprocessed high-voltage signal through signal amplification technology and analog-to-digital conversion technology to obtain a preliminary digital signal; processing the preliminary digital signal through a wavelet decomposition algorithm to obtain DC ripple data; extracting features from the DC ripple data through a wavelet basis to obtain insulation state features; The insulation state feature is compared with an adaptive threshold, and a pre-warning operation is started according to a comparison result.

[0006] In some embodiments, the processing of the preliminary digital signal by the wavelet decomposition algorithm comprises the following steps: The preliminary digital signal is subjected to multi-scale decomposition by a binary-scale low-pass filter and a binary-scale high-pass filter respectively to obtain a first approximation sub-signal and a first detail sub-signal; The first approximation sub-signal is subjected to iterative multi-scale decomposition by a binary-scale low-pass filter and a binary-scale high-pass filter respectively according to a wavelet decomposition layer number; The first approximation sub-signal, the first detail sub-signal, and signals obtained after iterative multi-scale decomposition of the first approximation sub-signal are arranged to obtain an approximation sub-signal set and a detail sub-signal set; The approximation sub-signal set and the detail sub-signal set are reconstructed by inverse transformation to obtain a ripple signal component; The ripple signal component is calculated to obtain direct-current ripple data.

[0007] In some embodiments, the wavelet decomposition layer number is obtained by the following steps: The high-voltage signal is obtained; The high-voltage signal is subjected to mean removal processing to obtain a direct-current bias removal signal; The direct-current bias removal signal is subjected to pre-analysis processing by segment window fast Fourier transform to obtain a main ripple frequency; The main ripple frequency is calculated according to a current sampling rate by a wavelet transform band formula to obtain a wavelet decomposition layer number.

[0008] In some embodiments, the wavelet base is determined by selecting an orthogonal wavelet base; The orthogonal wavelet base comprises a Daubechies wavelet base function.

[0009] In some embodiments, the feature extraction of the direct-current ripple data by the wavelet base to obtain an insulation state feature comprises the following steps: The direct-current ripple data is subjected to discrete wavelet transform by the wavelet base to obtain a detail coefficient and an approximation coefficient; The energy and the maximum amplitude of each layer are calculated according to the detail coefficient and the approximation coefficient to obtain an insulation state feature.

[0010] In some embodiments, the determination of the adaptive threshold comprises the following steps: The energy parameter in the insulation state feature is calculated to obtain a mean value and a standard deviation of the energy parameter; An adaptive energy threshold is obtained by calculating the mean and standard deviation of the energy parameter; The amplitude threshold is determined according to the amplitude; The effective ripple layer is obtained by comparing the energy and the maximum amplitude of each layer with the adaptive energy threshold and the amplitude threshold respectively, when the energy and the maximum amplitude are greater than the adaptive energy threshold and the amplitude threshold at the same time; The adaptive threshold is obtained by secondary screening of the effective ripple layer in combination with the frequency bandwidth.

[0011] In some embodiments, the method further comprises the following steps: The ripple signal is obtained by component reconstruction of the detail layer set through inverse wavelet transform; The ripple peak value is obtained by calculation according to the ripple signal; The root mean square ripple amplitude is obtained by calculation according to the ripple peak value; The ripple coefficient is obtained by calculation of the root mean square ripple amplitude and the high voltage signal; The ripple peak value and the ripple coefficient are output to the host computer for display.

[0012] To achieve the above-mentioned purpose, another aspect of the embodiment of the present application proposes a multi-layer adaptive ripple recognition device based on wavelet decomposition, which comprises: An acquisition module is configured to acquire a high voltage signal; A preprocessing module is configured to preprocess the acquired high voltage signal, and the preprocessing includes voltage reduction processing, buffer processing and filtering processing; A signal amplification module is configured to amplify the preprocessed high voltage signal; An analog-to-digital converter is configured to digitize the amplified high voltage signal to obtain a preliminary digital signal; A microcontroller is configured to preliminarily control and manage; An FPGA chip is configured to process the preliminary digital signal through a wavelet decomposition algorithm to obtain direct current ripple data, extract features of the direct current ripple data through a wavelet base to obtain insulation state features, and compare the insulation state features with an adaptive threshold; A relay is configured to switch the range of the signal amplification module according to the FPGA chip; A host computer is configured to display the results of the FPGA chip.

[0013] In some embodiments, the device further comprises: A voltage verification is configured to verify whether the error of the high voltage output value is within the allowable range; Current calibration, for measuring the output current at different voltages; AC voltage distortion calibration, for analyzing the waveform distortion rate of the output AC voltage; Voltage duration calibration, for measuring the actual duration of the output voltage.

[0014] To achieve the above-mentioned purposes, another aspect of the embodiments of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method described above.

[0015] The embodiments of the present application at least have the following beneficial effects: the present application provides a multi-layer adaptive ripple identification method and related equipment based on wavelet decomposition, which effectively reduces the interference and noise in the signal by preprocessing the high-voltage signal, providing a purer signal basis for subsequent processing. Then, by using signal amplification technology and analog-to-digital conversion technology, accurate amplification and digital conversion of the high-voltage signal are realized, and the preliminary digital signal can more accurately reflect the characteristics of the original signal. Then, by using the wavelet decomposition algorithm to process the preliminary digital signal, the DC ripple data is successfully separated, providing a key basis for subsequent feature extraction. Then, the wavelet basis is used to extract the features of the DC ripple data, and the insulation state features are accurately obtained, which can effectively represent the insulation condition of the equipment. Finally, the insulation state features are compared with the adaptive threshold, and the warning operation is started according to the comparison result, which can timely discover the insulation problem of the equipment and ensure the safe and stable operation of the equipment. The whole method process is scientific and reasonable, and each step is closely connected, from signal acquisition to feature extraction to warning, forming a complete and efficient ripple identification and insulation state monitoring system, which greatly improves the accuracy and reliability of ripple identification, and provides strong support for equipment maintenance and fault prevention. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a flowchart of the multi-layer adaptive ripple identification method based on wavelet decomposition provided by the embodiments of the present application; Figure 2 is a schematic diagram of the principle of Mallat algorithm; Figure 3 is a schematic diagram of the implementation method of wavelet transform; Figure 4 is a flowchart of step S400; Figure 5 is a schematic diagram of the wavelet transform algorithm flow; Figure 6 is a schematic diagram of the multi-layer adaptive ripple identification device based on wavelet decomposition; Figure 7 is a schematic diagram of the principle of wavelet decomposition algorithm based on FPGA platform. DETAILED DESCRIPTION

[0017] For the purpose of the present application, technical solutions and advantages, the following will be further described in detail in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application, and are not intended to limit the present application. When the following description relates to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application, but are only examples of devices and methods consistent with some aspects of the embodiments of the present application as described in the appended claims.

[0018] It can be understood that the terms "first", "second" and the like used in the present application can be used herein to describe various concepts, but unless specifically stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information. Depending on the context, the word "if" as used herein can be interpreted as "when" or "when" or "in response to determining".

[0019] The terms "at least one", "multiple", "each", "any" and the like used in the present application include one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0021] Before the embodiments of the present application are described in detail, first, some nouns and terms involved in the embodiments of the present application are described, and the nouns and terms involved in the embodiments of the present application are applicable to the following explanations.

[0022] Mallat algorithm: a fast implementation method of wavelet decomposition, which iteratively decomposes signals into approximation and detail components through filter banks, simplifying the calculation.

[0023] FFT transform: fast Fourier transform, which efficiently converts time domain signals to frequency domain, is used for analyzing signal frequency components and is widely used in spectrum analysis.

[0024] Daubechies series: a set of compactly supported orthogonal wavelet bases with different orders, suitable for signal denoising, feature extraction and other scenarios.

[0025] db4: 4th order wavelet in Daubechies series, balancing smoothness and compact support.

[0026] db6: 6th order Daubechies wavelet, longer support length, better smoothness.

[0027] Inverse wavelet transform: reconstruct the coefficients obtained by wavelet decomposition into the original signal, restore the time domain characteristics of the signal, which is the inverse process of wavelet analysis.

[0028] STM32 microcontroller: single-chip microcomputer based on ARM Cortex-M core, used for embedded system control and data processing, high cost performance.

[0029] FPGA chip: field programmable gate array, can customize hardware logic, suitable for parallel processing, high-speed signal processing and other scenarios.

[0030] In the related art, the paper "Single Ying. Research and implementation of automatic detection system of voltage withstand tester[D]. Shandong University, 2007"; application date is November 6, 2024, application number is CN202411570663.7, a kind of DC signal ripple detection method based on GMPSO-VMD-HT is disclosed; and the application date is June 28, 2024, the application number is CN202410856852.4, a kind of DC voltage and power supply ripple detection device is disclosed. The above papers and two published patents are all voltage withstand tester calibration technology, but in the DC ripple detection of these published voltage withstand tester calibration technology, the degree of automation is low and cannot be detected online in real time, more is the method of using oscilloscope for ripple measurement, the traditional method has poor recognition ability for weak ripple component, and the stability is insufficient in strong electromagnetic interference environment. At the same time, part of the high-precision algorithm is complex and has strong parameter dependence, which is not suitable for resource-limited equipment operation, and the detection device generally has low integration and poor portability. In the existing DC ripple detection technology, methods such as spectrum analysis, filter measurement or wavelet denoising are used. Among them, although wavelet denoising can effectively separate the ripple component, most of them use fixed wavelet basis and fixed decomposition level, which is difficult to adapt to the difference of ripple main frequency caused by the change of power supply type and sampling frequency in actual measurement, and cannot balance detection accuracy and real-time performance, especially in the detection of high-frequency weak ripple.

[0031] Therefore, the present application provides a multi-layer adaptive ripple identification method based on wavelet decomposition, as shown in Figure 1 , the method in Figure 1 is an optional flowchart of the multi-layer adaptive ripple identification method based on wavelet decomposition provided by the present application, Figure 1 The method can include but is not limited to steps S100-S600.

[0032] Step S100: obtaining a high-voltage signal; Step S200: preprocessing the obtained high-voltage signal, the preprocessing including voltage reduction processing, buffer processing and filtering processing; Step S300: performing signal amplification and digitization processing on the preprocessed high-voltage signal through signal amplification technology and analog-to-digital conversion technology, to obtain a preliminary digital signal; Step S400: processing the preliminary digital signal through a wavelet decomposition algorithm, to obtain direct current ripple data; Step S500: extracting features from the direct current ripple data through a wavelet basis, to obtain insulation state features; Step S600: comparing the insulation state features with an adaptive threshold, and starting a warning operation according to a comparison result.

[0033] The steps S100 to S600 shown in the embodiments of the present application obtain a high-voltage signal through a direct current high-voltage device. Then, preprocessing work is carried out on the obtained high-voltage signal, and the preprocessing process covers voltage reduction processing, buffer processing and filtering processing. After the preprocessing is completed, signal amplification and digitization processing are performed on the processed high-voltage signal through signal amplification technology and analog-to-digital conversion technology, to obtain a preliminary digital signal. Then, the preliminary digital signal is processed through a wavelet decomposition algorithm, to obtain direct current ripple data. The direct current ripple data is extracted through a wavelet basis, to obtain insulation state features. Finally, the insulation state features are compared with an adaptive threshold, and it is decided whether to start a warning operation according to a comparison result.

[0034] In some embodiments, in steps S100 to S300, to obtain a high-voltage signal, a high-voltage probe or a voltage divider can be used to collect a high-voltage signal output by a voltage withstand tester, to obtain transmissible high-voltage signal raw data, which is used for subsequent preprocessing. In the implementation process of preprocessing the obtained high-voltage signal, the collected high-voltage signal is subjected to voltage reduction processing through a voltage reduction circuit, to obtain a low-voltage signal suitable for subsequent circuits; the signal after voltage reduction is subjected to buffer processing through a buffer circuit, to obtain a signal for stable transmission; the signal after buffering is subjected to filtering processing through a filtering circuit, to remove high-frequency noise, to obtain a preprocessed signal, which is used for subsequent signal amplification and digitization processing. In the implementation process of processing through signal amplification technology and analog-to-digital conversion technology, the preprocessed signal can be amplified through an operational amplifier, to obtain a signal amplitude meeting the requirements of analog-to-digital conversion; the amplified signal is subjected to digitization processing through an analog-to-digital converter (ADC), to convert an analog signal into a digital signal, to obtain a preliminary digital signal, which can be used for subsequent further analysis and processing such as wavelet decomposition.

[0035] In some embodiments, in step S400, the wavelet transform is a time-frequency joint analysis method, which realizes the extraction of local characteristics of signals at different times and different frequency scales by convolving the signal with a set of waveform functions (i.e. wavelet bases) with time shift and scale stretching characteristics. Unlike the traditional Fourier transform which can only provide overall frequency domain information, the wavelet transform can provide high time resolution at high frequency and high frequency resolution at low frequency, and is particularly suitable for processing non-stationary signals such as transient ripple or periodic interference in direct current power supply, and can effectively separate direct current component, ripple component and random noise, thereby improving detection accuracy and anti-interference ability. At present, the ripple detection of direct current power supply often adopts oscilloscope observation method or Fourier analysis method, but these methods generally have the following shortcomings: relying on manual reading, unable to realize online automatic detection; frequency domain method (such as FFT) is difficult to accurately extract time-varying ripple component, and has poor adaptability to non-stationary signals; noise is easy to superimpose with ripple, leading to misjudgment; the device has large volume, high power consumption and high cost, and is not suitable for embedded long-term monitoring application. The wavelet transform is divided into continuous wavelet transform (CWT) and discrete wavelet transform (DWT), which have obvious differences in mathematical principles, calculation efficiency and application scenarios. Continuous wavelet transform is usually used in scientific research scenarios such as signal analysis, image processing and feature extraction, which realizes continuous time-frequency analysis of the whole signal by convolving the signal at different scales and shifts.

[0036] In the present application, discrete wavelet transform is adopted, and the specific implementation method is to use binary scale and integer shift to decompose the signal at multiple scales to obtain two sub-signals of approximation and detail, and recursively process the low-frequency part.

[0037] Specifically, the preliminary digital signal is subjected to multi-scale decomposition by a binary scale low-pass filter and a binary scale high-pass filter respectively to obtain a first approximation sub-signal and a first detail sub-signal; the first approximation sub-signal is subjected to iterative multi-scale decomposition by a binary scale low-pass filter and a binary scale high-pass filter respectively according to the number of wavelet decomposition layers; the first approximation sub-signal, the first detail sub-signal and the signals obtained after the iterative multi-scale decomposition of the first approximation sub-signal are arranged to obtain an approximation sub-signal set and a detail sub-signal set; the approximation sub-signal set and the detail sub-signal set are reconstructed by inverse transformation to obtain a ripple signal component; and the ripple signal component is calculated to obtain direct current ripple data. It can be understood that when the number of wavelet decomposition layers is 2, the approximation sub-signal set and the detail sub-signal set include the first approximation sub-signal, the first detail sub-signal, the second approximation sub-signal and the second detail sub-signal; when the number of wavelet decomposition layers is 3, the approximation sub-signal set and the detail sub-signal set include the first approximation sub-signal, the first detail sub-signal, the second approximation sub-signal, the second detail sub-signal, the third approximation sub-signal and the third detail sub-signal; and so on.

[0038] The discrete wavelet transform is simple to implement and efficient in calculation, and is suitable for embedded and real-time applications. The fast multi-layer wavelet decomposition can be realized, and the discrete wavelet transform is widely used in engineering fields such as signal compression, feature extraction, fault detection and ripple analysis.

[0039] In some embodiments, the wavelet transform is a multi-scale signal analysis method based on a wavelet basis function. For a one-dimensional time signal , the calculation formula of the discrete wavelet decomposition is: ; wherein, represents an approximate low-frequency component of the nth layer, represents a detail high-frequency component of the jth layer.

[0040] Different layers correspond to signals of different frequency bands, and the detail components D1-Dn usually contain direct current ripple and high-frequency interference information. As shown in Figure 2 , Figure 2 is a schematic diagram of the principle of Mallat algorithm, and an effective method of discrete wavelet transform is to use a filter, also known as Mallat algorithm. In Figure 2 , the measured signal is subjected to a low-pass filter and a high-pass filter, and then the low-frequency component and the high-frequency component of the measured signal are obtained. In the decomposition process of the wavelet transform algorithm principle for a one-dimensional time signal, two coefficients appear, , which correspond to the low-frequency and high-frequency parts in the Mallat algorithm respectively, and the number of layers in the entire wavelet transform depends on the number of times the measured signal passes through the filter.

[0041] Figure 3 This is a schematic diagram of the wavelet transform implementation method, which involves multi-level decomposition of the measured signal. When the measured signal passes through the first level of decomposition (through a low-pass filter and a high-pass filter), low-frequency signals and high-frequency signals are obtained. The low-frequency signals are then decomposed a second time to obtain another low-frequency signal and high-frequency signal. This process is repeated until the required number of decomposition levels is reached.

[0042] In some embodiments, in order to obtain the wavelet decomposition level, a high-voltage signal is obtained; the high-voltage signal is subjected to mean removal processing to obtain a DC bias removed signal; the DC bias removed signal is pre-analyzed using segment window fast Fourier transform to obtain the main ripple frequency; the main ripple frequency is calculated using the wavelet transform frequency band formula based on the current sampling rate to obtain the wavelet decomposition level.

[0043] Specifically, high-voltage signals are acquired using high-resolution signal acquisition equipment. These signals contain information such as DC components, ripple, and potential noise, forming the raw data foundation for all subsequent processing. The acquired high-voltage signals undergo mean-reduction processing. Specifically, the average value of the high-voltage signal over a certain time period is calculated, and then this average value is subtracted from each data point in the signal to obtain the signal with DC bias removed. After this step, only ripple and noise components remain in the signal, eliminating DC interference for subsequent ripple analysis. Next, a short-window Fast Fourier Transform (FFT) is used for pre-analysis of the DC-bias-removed signal. The DC-bias-removed signal is divided into several short time windows, and an FFT is performed on the signal within each window to obtain the spectrum corresponding to each window. By analyzing these spectra, the frequency component with the most concentrated energy is identified; this frequency is the main ripple frequency, reflecting the main characteristic frequency of the ripple signal. Finally, based on the current sampling rate and the wavelet transform frequency band formula, the number of wavelet decomposition levels is calculated. Substituting the previously obtained main ripple frequency into the calculation for this frequency band, we find the number of layers corresponding to the frequency band that completely encompasses the main ripple frequency. This number of layers is the required number of wavelet decomposition layers. Through this process, we can accurately determine the appropriate number of layers for subsequent wavelet decomposition, laying the foundation for further signal analysis and processing.

[0044] In some embodiments, the wavelet basis is determined by selecting orthogonal wavelet bases, including Daubechies wavelet basis functions. Specifically, the wavelet basis uses the Daubechies family, such as db4 or db6, which has tight support and stationarity.

[0045] In some embodiments, such as Figure 4 As shown, Figure 4 This is a flowchart of step S400. Figure 4The steps in the method include, but are not limited to, steps S410 to S520: Step S410: Discrete wavelet transform is performed on the DC ripple data by using a wavelet basis to obtain detail coefficients and approximation coefficients. Step S420: Energy and maximum amplitude of each layer are calculated based on the detail coefficients and the approximation coefficients to obtain insulation state features.

[0046] Specifically, in steps S410 to S420, a suitable wavelet basis is selected, and discrete wavelet transform is performed on the DC ripple data by using the wavelet basis. In the transform process, the original data is decomposed into a plurality of detail coefficients (D1 to DL) and approximation coefficients by convolution and down-sampling operation of the wavelet basis and the DC ripple data, and the coefficients correspond to high-frequency detail information and low-frequency approximation information at different scales, respectively. Then, based on the obtained detail coefficients and approximation coefficients, the energy and maximum amplitude corresponding to each layer of detail coefficients are calculated: the energy is calculated by summing the squares of each layer of detail coefficients, reflecting the strength of the high-frequency component of the layer; the maximum amplitude is obtained by extracting the maximum value in each layer of detail coefficients, representing the extreme degree of the high-frequency fluctuation of the layer. The calculated energy and maximum amplitude of each layer together constitute the features that can reflect the insulation state, because the change of the insulation state will directly affect the energy distribution and fluctuation amplitude of the high-frequency component in the DC ripple, for example, insulation aging or defects may cause abnormal increase of ripple energy or maximum amplitude in a specific frequency band.

[0047] In some embodiments, the determination of the adaptive threshold in step S600 includes the following steps: calculating the energy parameter in the insulation state features to obtain the mean and standard deviation of the energy parameter; obtaining the adaptive energy threshold by calculating the mean and standard deviation of the energy parameter; determining the amplitude threshold according to the amplitude; comparing the energy and the maximum amplitude of each layer with the adaptive energy threshold and the amplitude threshold, respectively, to obtain the effective ripple layer when the energy and the maximum amplitude are greater than the adaptive energy threshold and the amplitude threshold at the same time; performing secondary screening on the effective ripple layer in combination with the frequency bandwidth to obtain the adaptive threshold.

[0048] Specifically, the energy parameters of each layer in the insulation state feature are calculated to obtain the mean and standard deviation of the energy parameters, which are used as the basis for subsequent threshold setting; then the adaptive energy threshold is calculated by a preset calculation formula using the obtained mean and standard deviation of the energy parameters, to ensure that the threshold can reflect the overall distribution characteristics of the energy parameters; at the same time, according to the maximum amplitude characteristics of the detail coefficients of each layer, an amplitude threshold matching the maximum amplitude is set, so that the threshold can effectively distinguish normal amplitude from abnormal amplitude; then the energy and maximum amplitude of each layer are judged respectively, and the energy of each layer is compared with the adaptive energy threshold and the maximum amplitude is compared with the amplitude threshold, to screen out the layers whose energy is greater than the adaptive energy threshold and whose maximum amplitude is greater than the amplitude threshold, which are the preliminary effective ripple layers; finally, the preliminary effective ripple layers are subjected to secondary screening in combination with the preset frequency bandwidth range, to eliminate the layers that do not meet the frequency requirement, and the final screening result is the required adaptive threshold, which can accurately reflect the characteristic range of the insulation state. The calculation formula for calculating the mean and standard deviation of the detail layer energy is: ; ; wherein, represents the energy, represents the mean, represents the standard deviation.

[0049] , recommend ; wherein, represents the set adaptive energy threshold.

[0050] All layers that satisfy or are retained as effective ripple layers, and are subjected to secondary screening in combination with the frequency bandwidth, and the ripple is mainly concentrated in the range of 500 Hz-5 kHz. Among them: represents the maximum amplitude, represents the amplitude threshold.

[0051] In some embodiments, the step S600 further includes the following steps: performing component reconstruction on the detail layer set by inverse wavelet transform to obtain a ripple signal; calculating according to the ripple signal to obtain a ripple peak value; calculating according to the ripple peak value to obtain a root mean square ripple amplitude; calculating the root mean square ripple amplitude and the high voltage signal to obtain a ripple coefficient; and outputting the ripple peak value and the ripple coefficient to the upper computer for display.

[0052] Specifically, the detail layer set selected by the secondary screening in combination with the frequency bandwidth The ripple signal is reconstructed by inverse wavelet transform (IDWT), and the expression is: ; In the ripple amplitude and coefficient calculation, the calculation formula of the ripple peak value is: ; The calculation formula of the root mean square (RMS) ripple amplitude is: ; The calculation formula of the ripple coefficient and the total voltage ratio is: .

[0053] The upper computer outputs the ripple peak value and the voltage ripple coefficient, and if the set threshold is exceeded, an alarm or stop of the calibration operation is performed.

[0054] In some embodiments, the number of wavelet transform layers and the type of wavelet base need to be determined before measuring the ripple signal. Existing DC ripple detection techniques of voltage withstand testers mainly use oscilloscope detection, spectrum analysis, filter measurement or wavelet transform method. Among them, although the wavelet transform can effectively separate the ripple component, most of them use fixed wavelet base and fixed decomposition layer number, which is difficult to adapt to the difference of ripple main frequency caused by the change of power supply type and sampling frequency in actual measurement, and cannot balance the detection accuracy and real-time performance. After signal sampling, the initial signal is pre-analyzed by a short window FFT pre-analysis algorithm, the main ripple frequency is estimated , and the appropriate wavelet decomposition layer number L is calculated according to the current sampling rate Figure 5 and the wavelet transform band formula. The wavelet transform algorithm process is shown in .

[0055] Specifically, the first step is to sample the original DC voltage signal by a high-resolution ADC, and after obtaining the sampling signal, the sampling signal is processed to remove the DC bias, and finally a signal containing only ripple and noise is obtained. The sampling rate includes 10kHz-100kHz. The second step is to pre-analyze the signal obtained in the first step by removing the DC bias by using a short window FFT, so as to estimate the main ripple frequency ; then, according to the current sampling rate , the main ripple frequency is calculated according to the wavelet transform band formula, and the appropriate wavelet decomposition layer number L is obtained. The wavelet decomposition layer number formula is: ; The third step is to select a Daubechies series wavelet base, such as db4 or db6, and combine the decomposition layer number L obtained in the second step to perform a discrete wavelet transform (DWT) on the signal obtained after the first step, thereby obtaining a set of detail coefficients D1, D2, …, DL and approximation coefficients. The fourth step is to calculate the energy and maximum amplitude of each layer of detail coefficients Di obtained in the third step, and obtain these energy and maximum amplitude data. The calculation formulas of the energy and maximum amplitude are as follows: ; ; wherein, represents the energy, represents the maximum amplitude, represents the detail coefficient.

[0056] The fifth step is to calculate the mean and standard deviation of the detail layer energy obtained in the fourth step, set an adaptive energy threshold value according to the mean and standard deviation, and then filter out all layers that meet the energy or amplitude conditions as effective ripple layers. The calculation formulas of the mean and standard deviation of the detail layer energy are as follows: ; ; wherein, represents the energy, represents the mean, represents the standard deviation.

[0057] , recommend ; wherein, represents the set adaptive energy threshold value.

[0058] All layers that meet the conditions of or are retained as effective ripple layers, and a secondary screening is performed in combination with the frequency bandwidth, and the ripple is mainly concentrated in the range of 500 Hz-5 kHz. Among them: represents the maximum amplitude, represents the amplitude threshold value. The sixth step is to process the effective ripple layer set filtered out in the fifth step through inverse wavelet transform (IDWT) to reconstruct the ripple signal, and the reconstruction formula is as follows: The seventh step is to calculate the ripple peak value, root mean square (RMS) ripple amplitude, and ripple coefficient of the reconstructed ripple signal. The calculation formula of the ripple peak value is as follows: ; The calculation formula of the root mean square (RMS) ripple amplitude is as follows: ; The calculation formula of the ripple coefficient and the total voltage ratio is: .

[0059] In the eighth step, the obtained ripple peak value, voltage ripple coefficient and other parameters are compared with the set threshold value, and if the threshold value is exceeded, an alarm or a stop of the calibration operation is performed.

[0060] To achieve the above object, another aspect of the embodiment of the present application proposes a multi-layer adaptive ripple recognition device based on wavelet decomposition, as shown in Figure 6 Figure 6 is a schematic diagram of a multi-layer adaptive ripple recognition device based on wavelet decomposition, which comprises: an acquisition module for acquiring a high-voltage signal; a preprocessing module for preprocessing the acquired high-voltage signal, the preprocessing including voltage reduction processing, buffer processing and filtering processing, wherein: the voltage reduction processing is performed by a high-voltage attenuation module, the buffer processing is performed by a voltage follower, and the filtering processing is performed by a high-frequency wave module; a signal amplification module for signal amplification of the preprocessed high-voltage signal; an analog-to-digital converter for digital processing of the amplified high-voltage signal to obtain a preliminary digital signal; a microcontroller for preliminary control and management; an FPGA chip for processing the preliminary digital signal by a wavelet decomposition algorithm to obtain direct current ripple data; feature extraction of the direct current ripple data by a wavelet base to obtain insulation state features; and comparison of the insulation state features with an adaptive threshold value; a relay for range switching of the signal amplification module according to the FPGA chip; and an upper computer for displaying the results of the FPGA chip.

[0061] Specifically, the multi-layer adaptive ripple recognition device based on wavelet decomposition integrates various detection and control modules, realizes automatic detection and online verification of the electrical performance parameters of the voltage withstand tester, the input end of the device is connected to the measured signal, the high-voltage signal is first subjected to voltage reduction processing by a voltage attenuation module to ensure the safety of the subsequent circuit; then the signal is buffered and stably output by a voltage follower, and a high-frequency filter module is connected to filter out irrelevant high-frequency interference in direct current ripple detection. The filtered signal is sent to a signal amplification module and an ADC analog-to-digital converter for signal amplification and digitization. After the digital signal is preliminarily controlled and managed by an STM32 microcontroller, it is subjected to high-speed parallel processing by an FPGA chip, which is responsible for data distribution, fine calculation and linkage with a relay switching module (relay control) to realize range switching and signal path adjustment, and various algorithm processing of the collected digital signal. The FPGA also controls the range switching of the relay and the connection of the network communication interface and the upper computer to perform data uploading, instruction issuing and user interactive control.

[0062] ​It can be understood that the host computer can display the measured DC ripple data, including the DC ripple data without using the wavelet decomposition algorithm, and can intuitively see the data difference. The threshold comparison analysis process will only be performed when the FPGA internal program samples the ripple, and will not be displayed on the host computer. The host computer only serves as a display data, configure parameters, receive results, such as clicking the DC ripple detection, the device will automatically detect the DC ripple, and finally output the detection data result.

[0063] In some embodiments, as shown in Figure 6 the device further comprises: a voltage verification for verifying whether the error of the high-voltage output value is within the allowable range; a current verification for measuring the output current at different voltages; an AC voltage distortion verification for analyzing the waveform distortion rate of the output AC voltage; and a voltage duration verification for measuring the actual duration of the output voltage. The multi-dimensional performance index verification of the tested voltage-withstanding tester is completed by calculating the processed digital signal. Overall, the modules of the device work cooperatively to realize reliable acquisition of high-voltage signals, online detection of ripples, data processing, and remote control, and have the advantages of compact structure, high precision, and strong automation.

[0064] Specifically, the voltage verification is to verify whether the error of the high-voltage output value is within the allowable range by comparing the output voltage of the tested voltage-withstanding tester with the standard voltage source. The current verification is to measure the output current of the tester at different voltages by using a standard resistor or a current sensor. The function is to verify whether the judgment reference of the tester for insulation breakdown is accurate. The AC voltage distortion verification is to analyze the waveform distortion rate of the output AC voltage of the AC voltage-withstanding tester by using a distortion meter. The distorted voltage waveform may contain additional harmonic components, which interfere with the accuracy of the leakage current detection and even affect the judgment of the insulation performance of the measured device, so it is necessary to ensure that the waveform meets the standard. The voltage duration verification is to measure the actual duration of the output voltage of the tester by using a timer.

[0065] In some embodiments, as shown in Figure 7 Figure 7 is a schematic diagram of the wavelet decomposition algorithm based on the FPGA platform. Figure 6 ​The analog quantity u(t) of the input signal is output as digital quantity x(t) after passing through the A / D conversion unit. The digital quantity x(t) enters the calculation unit through the first data buffer unit. When the wavelet decomposition algorithm is not used, the ripple value is directly calculated and then output for display. At the same time, the digital quantity x(t) is buffered in the first data buffer unit and then passes through a subtractor to eliminate the DC bias, leaving only the noise signal and the ripple signal. At this time, the main frequency of the DC ripple is calculated by the FFT algorithm main frequency estimation calculation unit to obtain the corresponding wavelet decomposition layer number. Then, the decomposition coefficients are obtained by the multi-layer wavelet decomposition unit. Finally, the data is transmitted to the calculation unit to calculate the energy, mean value, amplitude, ripple coefficient, etc. of the DC ripple. The energy and amplitude are compared with the threshold value. The frequency components that meet the determination condition are reconstructed by the inverse wavelet transform, and the related parameters are output.

[0066] The application integrates a signal spectrum statistical module in an FPGA or a microcontroller, automatically determines an optimal wavelet decomposition layer number based on a characteristic frequency interval of a measured direct current signal, and most of ripple frequencies are distributed between tens of hertz and thousands of hertz. For example, a signal energy change rate or a frequency band inclusion rate is used as a control criterion of the decomposition layer number, the effective ripple component information is maintained, and the calculation complexity is reduced as much as possible. The layer number adaptive algorithm dynamically adjusts the wavelet transform depth in combination with actual sampling rate, signal length, expected ripple frequency band and other information, and achieves wide compatibility of ripple signals of different types of power supplies (such as switching power supplies, linear power supplies, charging power supplies and the like). The method not only improves the robustness and precision of detection, but also has good real-time performance and resource control capability, and is suitable for deployment on an embedded platform with limited computing resources such as an FPGA. The application introduces a wavelet decomposition and multi-layer adaptive ripple recognition strategy, especially proposes a method of adaptive determination of the decomposition layer number, and significantly improves the detection capability and adaptability of the direct current ripple detection in various practical application scenarios. A traditional wavelet ripple detection system usually relies on manual setting of a fixed decomposition layer number, which causes under-decomposition or over-decomposition of the algorithm when facing direct current ripples of different frequency ranges and different amplitude levels, and affects the accuracy and real-time performance of ripple recognition. The layer number adaptive algorithm proposed in the application can intelligently determine the optimal wavelet decomposition depth by dynamically analyzing the frequency spectrum distribution and energy change characteristics of the measured signal, effectively suppresses redundant noise components and calculation burden while ensuring the accuracy of ripple extraction. This mechanism not only improves the resolution capability of the detection algorithm for complex ripple signals (such as low-frequency fundamental superimposed high-frequency interference), but also optimizes system resource allocation, realizes higher data processing efficiency without increasing hardware load, and is suitable for deployment on resource-limited platforms such as FPGAs. At the same time, the adaptive mechanism improves the wide compatibility of ripple of different types of power supplies (such as industrial power supplies, vehicle-mounted power supplies, communication power supplies and the like), and enhances the universality and engineering practicability of the algorithm. Overall, the method in the application has significant gains in accuracy, real-time performance, system resource efficiency, signal compatibility and the like, can meet the ripple detection requirements of high precision and high stability, and provides an efficient, intelligent and low-cost online detection solution for power quality detection, power supply performance evaluation and voltage withstand tester calibration and the like.

[0067] The embodiment of the application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the method.

[0068] It can be understood that the contents in the method embodiments are applicable to the storage medium embodiments, the storage medium embodiments specifically realize the functions of the method embodiments, and the beneficial effects achieved by the storage medium embodiments are the same as the beneficial effects achieved by the method embodiments.

[0069] The embodiments described in the present application are used to more clearly illustrate the technical solutions of the present application, and do not constitute a limitation on the technical solutions provided by the present application. Those skilled in the art can know that, as technology evolves and new application scenarios appear, the technical solutions provided by the present application are also applicable to similar technical problems.

[0070] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation on the present application, and can include more or fewer steps than shown in the figures, or combine certain steps or different steps.

[0071] The device embodiments described above are merely illustrative, and the units described as separate components can or can not be physically separated, that is, they can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the present embodiment according to actual needs.

[0072] Those skilled in the art can understand that all or some of the steps in the above method, the functions of the modules / units in the system and the device can be implemented as software, firmware, hardware and their appropriate combinations.

[0073] It should be understood that in the present application, "at least one" means one or more, and "multiple" means two or more. "And / or" is used to describe the association between the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean: only A, only B, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0074] The units described above as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the present embodiment according to actual needs.

[0075] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing unit, or each of the units can exist physically, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0076] The preferred embodiments of the present application are described above with reference to the accompanying drawings, and are not intended to limit the scope of the present application. Any modification, equivalent replacement and improvement made by those skilled in the art without departing from the scope and spirit of the present application shall fall within the scope of the present application.

Claims

1. A multi-layer adaptive tonal noise recognition method based on wavelet decomposition, characterized in that, The method comprises the following steps: Obtaining a high-voltage signal; Pretreating the obtained high-voltage signal, the pretreatment comprising voltage reduction processing, buffer processing and filtering processing; Signal amplification and digitization processing of the pretreated high-voltage signal through signal amplification technology and analog-to-digital conversion technology to obtain a preliminary digital signal; Processing the preliminary digital signal through a wavelet decomposition algorithm to obtain direct current ripple data; Feature extraction of the direct current ripple data through a wavelet base to obtain insulation state features; Comparing the insulation state features with an adaptive threshold value, and starting a pre-warning operation according to the comparison result.

2. The method of claim 1, wherein, The processing of the preliminary digital signal through the wavelet decomposition algorithm comprises the following steps: Multi-scale decomposition of the preliminary digital signal through a binary scale low-pass filter and a binary scale high-pass filter respectively to obtain a first approximation sub-signal and a first detail sub-signal; Iterative multi-scale decomposition of the first approximation sub-signal through a binary scale low-pass filter and a binary scale high-pass filter according to the wavelet decomposition layer number; Arranging the first approximation sub-signal, the first detail sub-signal and the signal obtained after the iterative multi-scale decomposition of the first approximation sub-signal to obtain an approximation sub-signal set and a detail sub-signal set; Reconstruction of the approximation sub-signal set and the detail sub-signal set through inverse transformation to obtain a ripple signal component; Calculation of the ripple signal component to obtain direct current ripple data.

3. The method of claim 2, wherein, The obtaining of the wavelet decomposition layer number comprises the following steps: Obtaining the high-voltage signal; De-meaning processing of the high-voltage signal to obtain a direct current bias removal signal; Pre-analysis processing of the direct current bias removal signal through segment window fast Fourier transform to obtain a main ripple frequency; Calculation of the main ripple frequency through a wavelet transform frequency band formula according to a current sampling rate to obtain a wavelet decomposition layer number.

4. The method of claim 3, wherein, The wavelet base is determined by selecting an orthogonal wavelet base; The orthogonal wavelet base comprises a Daubechies wavelet base function.

5. The method of claim 4, wherein, The feature extraction of the direct current ripple data through the wavelet base to obtain insulation state features comprises the following steps: Discrete wavelet transform of the direct current ripple data through the wavelet base to obtain a detail coefficient and an approximation coefficient; Calculation of the energy and the maximum amplitude of each layer according to the detail coefficient and the approximation coefficient to obtain insulation state features.

6. The method of claim 5, wherein, The determination of the adaptive threshold value comprises the following steps: Calculation of an energy parameter in the insulation state features to obtain a mean value and a standard deviation of the energy parameter; Calculation of an adaptive energy threshold value through the mean value and the standard deviation of the energy parameter; Determination of an amplitude threshold value according to the amplitude; Comparison of the energy and the maximum amplitude of each layer with the adaptive energy threshold value and the amplitude threshold value respectively, and obtaining an effective ripple layer when the energy and the maximum amplitude are greater than the adaptive energy threshold value and the amplitude threshold value at the same time; Secondary screening of the effective ripple layer in combination with a frequency bandwidth to obtain an adaptive threshold value.

7. The method of claim 6, wherein, The method further comprises the following steps: The ripple signal is obtained by component reconstruction of the detail layer set through inverse wavelet transform; The ripple peak value is obtained by calculation according to the ripple signal; The root mean square ripple amplitude is obtained by calculation according to the ripple peak value; The ripple coefficient is obtained by calculation of the root mean square ripple amplitude and the high voltage signal; The ripple peak value and the ripple coefficient are output to the host computer for display.

8. A multi-layer adaptive tone recognition apparatus based on wavelet decomposition, characterized by, The device comprises: An acquisition module for acquiring a high voltage signal; A preprocessing module for preprocessing the acquired high voltage signal, the preprocessing including voltage reduction processing, buffer processing and filtering processing; A signal amplification module for signal amplification of the preprocessed high voltage signal; An analog-to-digital converter for digital processing of the amplified high voltage signal to obtain a preliminary digital signal; A microcontroller for preliminary control and management; An FPGA chip for processing the preliminary digital signal through a wavelet decomposition algorithm to obtain direct current ripple data, extracting features of the direct current ripple data through a wavelet base to obtain insulation state features, and comparing the insulation state features with an adaptive threshold value; A relay for range switching of the signal amplification module according to the FPGA chip; A host computer for displaying results of the FPGA chip.

9. The wavelet-decomposition-based multilevel adaptive tone recognition device according to claim 8, wherein, The device further comprises: Voltage verification for verifying whether the error of the high voltage output value is within the allowable range; Current verification for measuring the output current under different voltages; AC voltage distortion verification for analyzing the waveform distortion rate of the output AC voltage; Voltage duration verification for measuring the actual duration of the output voltage.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by a processor to implement the method of any one of claims 1 to 7. The computer program is executed by a processor to implement the method of any one of claims 1 to 7.

Images