A partial discharge signal processing method and device under high-frequency square wave voltage

Abstract

The application discloses a local discharge signal processing method and device under high-frequency square wave voltage, wherein the method comprises the following steps: filtering low-frequency noise signals of a test power supply by using a high-pass filter with a preset cutoff frequency, wherein the test power supply is a high-frequency square wave power supply for local discharge test of an insulation sample; obtaining a local discharge starting voltage of the insulation sample, collecting a local discharge signal when a test voltage is the local discharge starting voltage, obtaining a characteristic frequency band of local discharge of the insulation sample according to the local discharge signal, and the test voltage is a high-frequency square wave voltage of the test power supply; performing modal decomposition on the local discharge signal of the characteristic frequency band to obtain a plurality of modal components, selecting a plurality of modal components satisfying a preset condition to perform noise reduction processing and signal reconstruction, and obtaining a denoised local discharge signal of the insulation sample under the high-frequency square wave voltage. The application can avoid interference of the high-frequency square wave power supply on the local discharge signal, and obtain a local discharge signal with high signal-to-noise ratio under the high-frequency square wave voltage.

Description

Technical Field

[0001] This invention relates to the field of partial discharge processing technology, and in particular to a method and apparatus for processing partial discharge signals under high-frequency square wave voltage. Background Technology

[0002] Currently, the insulating medium of electrical equipment is mostly composed of solid composite insulating materials. Due to differences in manufacturing processes, aging, and deterioration during production and under complex operating conditions, impurities, decomposition products, and bubbles inevitably form within the insulation, leading to insulation defects. This, in turn, causes electric field distortion within the insulating medium under an electric field. High electric field strength causes localized breakdown within the insulation, resulting in complex discharge phenomena known as partial discharge (PD).

[0003] Partial discharge signals contain significant noise, necessitating denoising processing. Compared to the characteristics of partial discharge signals from solid insulation under power frequency AC voltage, the characteristics of partial discharge under high-frequency square wave voltage differ significantly, such as larger amplitude fluctuations, higher discharge frequency, and concentrated discharge phases. Existing partial discharge signal denoising methods suffer from low accuracy and narrow applicability under high-frequency square wave voltage, resulting in low signal-to-noise ratios in the obtained partial discharge signals. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for processing partial discharge signals under high-frequency square wave voltage, so as to solve the technical problems of low accuracy and narrow applicability of existing partial discharge signal denoising methods under high-frequency square wave voltage.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for processing partial discharge signals under high-frequency square wave voltage includes:

[0007] The low-frequency noise signal of the test power supply is filtered out by a high-pass filter with a preset cutoff frequency. The test power supply is a high-frequency square wave power supply for partial discharge testing of the insulating sample.

[0008] The partial discharge initiation voltage of the insulating sample is obtained, and the partial discharge signal is acquired when the test voltage is the partial discharge initiation voltage. The characteristic frequency band of partial discharge of the insulating sample is obtained based on the partial discharge signal. The test voltage is the high-frequency square wave voltage of the test power supply.

[0009] The partial discharge signal of the characteristic frequency band is decomposed into multiple modal components. Several modal components that meet the preset conditions are selected for noise reduction and signal reconstruction to obtain the denoised partial discharge signal of the insulating sample under high-frequency square wave voltage.

[0010] Optionally, before filtering out the low-frequency noise signal of the test power supply using a high-pass filter with a preset cutoff frequency, the method further includes:

[0011] The first spectral signal of the insulating sample when no partial discharge occurs and the second spectral signal when partial discharge occurs are obtained;

[0012] By comparing and analyzing the first and second spectral signals, the frequency bands in which low-frequency noise occurs can be obtained.

[0013] The cutoff frequency of the high-pass filter is set according to the frequency band in which the low-frequency noise occurs.

[0014] Optionally, obtaining the partial discharge initiation voltage of the insulating sample includes:

[0015] A predetermined number of insulation samples were tested using a boost method to obtain the test voltage at which partial discharge occurred in each insulation sample. The average value of the test voltage was taken as the partial discharge initiation voltage of the insulation sample.

[0016] Optionally, the partial discharge signal acquired when the test voltage is the partial discharge initiation voltage includes:

[0017] The partial discharge signal is acquired using an ultra-high frequency antenna sensor when the test voltage is the partial discharge initiation voltage.

[0018] Optionally, the step of obtaining the characteristic frequency band of partial discharge of the insulating sample based on the partial discharge signal includes:

[0019] The partial discharge signal is subjected to a fast Fourier transform to obtain the spectral signal of the insulating sample when partial discharge occurs.

[0020] The characteristic frequency band of partial discharge of the insulating sample is obtained based on the spectral signal.

[0021] Optionally, the modal decomposition of the partial discharge signal in the characteristic frequency band to obtain multiple modal components includes:

[0022] The adaptive noise complete set empirical mode decomposition algorithm CEEMDAN is used to perform mode decomposition on the partial discharge signal of the characteristic frequency band to obtain multiple mode components.

[0023] Optionally, the step of selecting several modal components that meet preset conditions for noise reduction and signal reconstruction to obtain the denoised partial discharge signal of the insulating sample under a high-frequency square wave voltage includes:

[0024] The modal components are divided into a first modal component, a second modal component, and a third modal component based on the correlation coefficient of the modal components; the higher the correlation coefficient, the less noise is contained in the modal component; the correlation coefficient of the first modal component is greater than a first preset threshold, the correlation coefficient of the second modal component is greater than a second preset threshold and less than the first preset threshold, and the correlation coefficient of the third modal component is less than the second preset threshold;

[0025] Wavelet analysis is used to denoise the second modal component to obtain the denoised second modal component;

[0026] Discarding the third mode component, the first mode component and the denoised second mode component are reconstructed to obtain the denoised partial discharge signal of the insulating sample under a high-frequency square wave voltage.

[0027] Optionally, the signal denoising of the second mode component using wavelet analysis includes:

[0028] The wavelet function used in the wavelet analysis is the Sym8 function.

[0029] Optionally, reconstructing the signal from the first mode component and the denoised second mode component includes:

[0030] The first modal component and the denoised second modal component are superimposed on each other.

[0031] The present invention also provides a partial discharge signal processing device under high-frequency square wave voltage, comprising:

[0032] The power supply noise filtering module is used to filter out low-frequency noise signals from the test power supply using a high-pass filter with a preset cutoff frequency. The test power supply is a high-frequency square wave power supply used for partial discharge testing of the insulating sample.

[0033] The characteristic frequency band acquisition module is used to acquire the partial discharge initiation voltage of the insulating sample, collect the partial discharge signal when the test voltage is the partial discharge initiation voltage, and obtain the characteristic frequency band of partial discharge of the insulating sample based on the partial discharge signal, wherein the test voltage is the high-frequency square wave voltage of the test power supply;

[0034] The signal denoising and reconstruction module is used to perform mode decomposition on the partial discharge signal of the characteristic frequency band to obtain multiple mode components, select several mode components that meet preset conditions for denoising processing and signal reconstruction, and obtain the denoised partial discharge signal of the insulating sample under high-frequency square wave voltage.

[0035] This invention provides a method and apparatus for processing partial discharge signals under high-frequency square wave voltage. The method includes: filtering out low-frequency noise signals from a test power supply using a high-pass filter with a preset cutoff frequency, wherein the test power supply is a high-frequency square wave power supply used for partial discharge testing of an insulating sample; acquiring the partial discharge initiation voltage of the insulating sample; acquiring a partial discharge signal when the test voltage is the partial discharge initiation voltage; obtaining a characteristic frequency band in which partial discharge occurs in the insulating sample based on the partial discharge signal, wherein the test voltage is the high-frequency square wave voltage of the test power supply; performing mode decomposition on the partial discharge signal in the characteristic frequency band to obtain multiple mode components; selecting several mode components that meet preset conditions for noise reduction processing and signal reconstruction to obtain a denoised partial discharge signal of the insulating sample under high-frequency square wave voltage.

[0036] In view of this, the beneficial effects of this invention are:

[0037] This invention fully considers the noise interference of high-frequency square wave voltage on partial discharge signals. It filters out the noise of the high-frequency square wave power supply itself by using a high-pass filter with a preset cutoff frequency, thus avoiding interference from the high-frequency square wave power supply to the partial discharge signal. It obtains the partial discharge initiation voltage of the insulating sample and acquires the partial discharge signal when the test voltage is the partial discharge initiation voltage. Based on the partial discharge signal, it obtains the characteristic frequency band of partial discharge in the insulating sample. Then, it performs mode decomposition on the partial discharge signal in the characteristic frequency band to obtain multiple mode components. Selecting some mode components for signal denoising and reconstruction improves the accuracy of partial discharge signal denoising and yields a partial discharge signal with a high signal-to-noise ratio under high-frequency square wave voltage, enabling more accurate analysis of the health status and insulation life of the insulating sample. Since only some mode components are selected for processing, it reduces computation time and improves calculation accuracy, making it suitable for situations with high sampling frequencies and large data volumes, and meeting practical application requirements. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating an embodiment of the method of the present invention;

[0039] Figure 2 This is a schematic diagram of the noise signal in the ultra-high frequency detection signal when no partial discharge occurs in the method of the present invention;

[0040] Figure 3 This is a schematic diagram illustrating the modal decomposition and wavelet denoising processes in an embodiment of the method of the present invention;

[0041] Figure 4 This is a schematic diagram of the IMF components decomposed by the CEEMDAN algorithm in an embodiment of the method of the present invention;

[0042] Figure 5This is a schematic diagram of the correlation coefficients calculated by the CEEMDAN algorithm in an embodiment of the method of the present invention;

[0043] Figure 6 This is a schematic diagram of the partial discharge signal before denoising in a certain embodiment of the method of the present invention;

[0044] Figure 7 This is a schematic diagram of the wavelet denoising result in an embodiment of the method of the present invention;

[0045] Figure 8 This is a schematic diagram of the structure of an embodiment of the device of the present invention. Detailed Implementation

[0046] This invention provides a method and apparatus for processing partial discharge signals under high-frequency square wave voltage, in order to solve the technical problems of low accuracy and narrow applicability of existing partial discharge signal denoising methods under high-frequency square wave voltage.

[0047] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0049] According to the IEC 60270 standard, partial discharge (PD) is a discharge occurring in a partially insulated area between two conductive electrodes with a gap. Partial discharge refers to a localized discharge phenomenon caused by insulation defects in the insulating medium of electrical equipment, and it reflects the insulation level of the equipment. The main causes of partial discharge in electrical equipment are poor manufacturing processes, external damage, and insulation aging. Partial discharge will occur at insulation defects during the operation of electrical equipment. If the discharge continues, it will exacerbate the insulation defects, shorten the equipment's service life, and ultimately endanger the safe operation of the power system. Therefore, partial discharge detection is crucial in the operation and maintenance of electrical equipment.

[0050] Currently, there are many methods for detecting partial discharge in power equipment and devices, such as impedance detection, current pulse method, ultrasonic detection, fiber optic sensing, and UHF antenna measurement. Although these methods differ in their measurement principles, sensor types, and signal types, the obtained signals all contain significant noise interference. This noise affects the accurate calculation of partial discharge information, particularly the amplitude and frequency band characteristics, making it difficult to distinguish the true discharge information. Therefore, it is essential to denoise the detected partial discharge signals to accurately analyze them and assess the insulation status of electrical equipment.

[0051] Compared to the partial discharge signal characteristics of solid insulation under power frequency AC voltage, the partial discharge characteristics under high-frequency non-sinusoidal voltage exhibit significant differences, such as a larger fluctuation range in discharge amplitude, a higher number of discharges, and a concentrated discharge phase. In recent years, with the development and application of high-capacity power electronic equipment, power equipment and components, such as high-voltage power modules, reactors, and high-frequency transformers, are subjected to complex high-frequency non-sinusoidal voltages. These voltages have high frequencies, steep rise times, and exhibit polarity effects and multiple harmonic components, easily causing complex and diverse partial discharges in the insulation structure, thus threatening the safe and stable operation of power equipment. Therefore, it is urgent to study the testing and signal processing techniques for partial discharge of insulation under high-frequency non-sinusoidal voltages.

[0052] In recent years, with the development of signal processing technology, research on partial discharge signal denoising has increased significantly, including wavelet thresholding and empirical mode decomposition. Unlike traditional power frequency AC and DC partial discharge signals under pulse conditions, high-frequency non-sinusoidal voltages have pulse voltages with steep rising edges, generating strong interference signals for sensors detecting partial discharge signals. The test power supply itself also has significant interference signals. In addition, interference noise signals still exist at the same frequency during partial discharge, which brings considerable difficulties to the detection and denoising of insulation partial discharge signals under high-frequency non-sinusoidal voltages, especially when the sampling rate is high and the number of data points is large. Existing partial discharge signal denoising methods have low accuracy under high-frequency non-sinusoidal voltages and a narrow range of applications.

[0053] Partial discharge generates signals with a wide frequency range, leading to various partial discharge detection technologies targeting different frequency ranges. Among these, ultra-high frequency (UHF) detection, covering a frequency range of 300MHz to 1500MHz, works by using UHF sensors to detect the ultra-high frequency electromagnetic signals generated during partial discharge in power equipment. This allows for the acquisition of relevant information and the monitoring of partial discharge. UHF detection technology boasts high detection sensitivity and is widely used in online partial discharge monitoring systems for gas-insulated switchgear (GIS), transformers, and ring main units (RMU). Depending on the specific equipment conditions at the site, both built-in and external UHF sensors can be employed.

[0054] The application scenario of this invention is insulation partial discharge signal under high frequency square wave voltage. The test power supply for partial discharge test is high frequency square wave voltage. First, the interference noise of the test power supply itself is filtered out by a high-pass filter. Then, the detected partial discharge signal under high frequency square wave voltage is denoised and reconstructed to finally obtain a partial discharge signal with high signal-to-noise ratio, so as to further evaluate the insulation status of electrical equipment.

[0055] Firstly, please refer to Figure 1 This invention provides an embodiment of a partial discharge signal processing method under a high-frequency square wave voltage, comprising:

[0056] S100: The low-frequency noise signal of the test power supply is filtered out by a high-pass filter with a preset cutoff frequency. The test power supply is a high-frequency square wave power supply for partial discharge testing of the insulating sample.

[0057] S200: Obtain the partial discharge initiation voltage of the insulating sample, collect the partial discharge signal when the test voltage is the partial discharge initiation voltage, and obtain the characteristic frequency band of partial discharge of the insulating sample based on the partial discharge signal, wherein the test voltage is the high-frequency square wave voltage of the test power supply;

[0058] S300: The partial discharge signal of the characteristic frequency band is decomposed into multiple modal components. Several modal components that meet the preset conditions are selected for noise reduction and signal reconstruction to obtain the denoised partial discharge signal of the insulating sample under high-frequency square wave voltage.

[0059] Unlike traditional partial discharge signals under AC and DC pulses, high-frequency non-sinusoidal voltages, i.e., high-frequency square wave voltages, have pulse voltages with steep rising edges, causing strong interference signals on the partial discharge sensor and resulting in significant interference signals from the test power supply itself. Therefore, a high-pass filter with a preset cutoff frequency is used to filter out the low-frequency noise signals from the test power supply, which is a high-frequency square wave power supply used for partial discharge testing of insulating samples.

[0060] When conducting partial discharge tests on insulation samples, a partial discharge test platform under high-frequency non-sinusoidal voltage is first built. The partial discharge model can be either a surface discharge model or an air gap discharge model. The partial discharge detection method can be the ultra-high frequency (UHF) detection method, which has the characteristics of high detection sensitivity, high frequency band, non-contact, anti-interference, and easy identification of insulation defect types.

[0061] In one embodiment, an ultra-high frequency detection instrument, such as an ultra-high frequency antenna sensor, is used to detect the ultra-high frequency electromagnetic wave signal generated when the insulating sample undergoes partial discharge. Preferably, the ultra-high frequency antenna sensor has an operating temperature of -40℃ to +85℃ and an operating frequency of 300M-1.5GHZ.

[0062] It should be noted that the partial discharge model in this embodiment of the invention can also be an air gap discharge model, and air gap discharge experiments can also be performed.

[0063] In one embodiment, the insulating sample can be an insulating sample made of solid epoxy resin; the size of the insulating sample can be a 30mm square and the thickness can be 0.1mm. Preferably, the insulating sample is made of epoxy material, curing agent and accelerator, and the mass ratio between epoxy material, curing agent and accelerator is 100:a:0.6, where a=Ev×166×0.9, and Ev represents the epoxy value of epoxy material.

[0064] It should be noted that epoxy resin (EP) is commonly used as the main insulation material for high-frequency transformers. In this embodiment, the insulation sample can be made of solid epoxy resin or other insulating materials (including gaseous, liquid and solid insulating materials), such as insulating varnish, insulating glue, insulating paper, plastic, glass, insulating oil, etc.

[0065] It should be noted that the size of the insulation sample is determined by the size of the test electrode, which is specified according to GB / T1048.1—2016 "Electrical strength test method for insulating materials - Part 1: Test at power frequency" and CIGRE Method II standard.

[0066] In one embodiment of the present invention, before filtering out the low-frequency noise signal of the test power supply using a high-pass filter with a preset cutoff frequency, the method further includes setting the cutoff frequency of the high-pass filter, specifically:

[0067] Acquire the first spectral signal when the insulation sample does not experience partial discharge and the second spectral signal when partial discharge occurs; compare and analyze the first and second spectral signals to obtain the frequency band where low-frequency noise occurs; set the cutoff frequency of the high-pass filter according to the frequency band where low-frequency noise occurs.

[0068] Specifically, a test voltage less than the partial discharge initiation voltage (PDIV) is applied to the insulating sample. At this point, no partial discharge occurs in the insulating sample, and a first spectral signal is acquired. Then, a test voltage greater than or equal to the PDIV is applied to the insulating sample. At this point, partial discharge occurs in the insulating sample, and a second spectral signal is acquired. By comparing the first and second spectral signals, the frequency band where noise occurs can be analyzed, thereby setting an appropriate cutoff frequency. For example, please refer to [link to relevant documentation]. Figure 2 It was found that the partial discharge signal of a certain insulation sample occurred in the frequency band above 400MHz, and there were some interference signals in the low frequency band below 400MHz. Therefore, the cutoff frequency can be set to 400MHz, and the low frequency noise signal in the test power supply can be filtered out by a high-pass filter with a cutoff frequency of 400MHz.

[0069] Because high-frequency square wave voltage has a pulse voltage with a steep rising edge, it will generate a strong interference signal to the ultra-high frequency antenna sensor that detects partial discharge signals, that is, the interference signal of the test power supply itself is large. Therefore, this embodiment fully considers the interference noise characteristics of the partial discharge signal under high-frequency square wave voltage. Before denoising the partial discharge signal of the insulating sample, a high-pass filter with a preset cutoff frequency is selected to filter out the noise of the high-frequency square wave power supply itself.

[0070] In step S200, the partial discharge initiation voltage of the insulating sample is obtained, the partial discharge signal when the test voltage is the partial discharge initiation voltage is collected, the characteristic frequency band of partial discharge of the insulating sample is obtained according to the partial discharge signal, and the test voltage is the high-frequency square wave voltage of the test power supply.

[0071] According to standard GB / T 22720, the partial discharge initiation voltage (PDIV) is: the lowest voltage at which partial discharge is first detected in the test circuit when the voltage applied to the test sample gradually increases from a lower value where partial discharge is not observed.

[0072] A partial discharge test platform under high-frequency square wave voltage was used, with column-plate electrodes as the test electrodes and a test temperature of 30°C. When testing the partial discharge initiation voltage (PDIV) of the insulating sample, the square wave voltage frequency could be selected as 1kHz, 5kHz, 10kHz, 15kHz, or 20kHz, with a boost rate of 10kV / s. By selecting different square wave voltage frequencies within the range of 1kHz to 20kHz, the influence of frequency on partial discharge can be observed, thus illustrating the necessity of investigating high-frequency insulation issues to facilitate further research on insulation lifetime models under high-frequency square wave voltage.

[0073] In partial discharge experiments, the boost method refers to the method of measuring the corresponding current and power factor under conditions of gradually increasing voltage. In the boost method experiment, as the voltage gradually increases, the partial discharge activity also gradually intensifies with the increase of electric field strength. At this time, the apparent discharge quantity (partial discharge quantity) of the partial discharge can be measured. The boost method is suitable for detecting the initiation and progression of partial discharge and provides more detailed information, such as the assessment of discharge level.

[0074] In one embodiment of the present invention, obtaining the partial discharge initiation voltage of the insulating sample includes: testing a preset number of insulating samples using a boost method to obtain the test voltage when partial discharge occurs in each insulating sample, and taking the average value of the test voltage as the partial discharge initiation voltage of the insulating sample.

[0075] Specifically, the test temperature is room temperature, and the partial discharge initiation voltage (PDIV) of the insulation sample is tested using a continuous voltage ramp method. The square wave voltage frequency can be selected as 1kHz, 5kHz, 10kHz, 15kHz, or 20kHz, the voltage ramp rate is selected as 10kV / s, and the partial discharge signal threshold is 20mV. A preset number of insulation samples can be tested at each square wave voltage frequency. For example, 5 or 10 insulation samples can be tested at each frequency. The test voltage when partial discharge occurs at each insulation sample is recorded. The average value of the test voltages when partial discharge occurs at all insulation samples is taken as the partial discharge initiation voltage (PDIV), and the partial discharge signal (PD signal) when the test voltage is the partial discharge initiation voltage (PDIV) is collected using an ultra-high frequency antenna sensor.

[0076] It is understood that the test power supply in this embodiment is a high-frequency square wave power supply, and the test voltage is the high-frequency square wave voltage of the test power supply.

[0077] In one embodiment, after acquiring the partial discharge signal when the insulating sample experiences partial discharge, a Fast Fourier Transform (FFT) can be used to observe the frequency domain partial discharge signal. A Fast Fourier Transform is then performed on the partial discharge signal to calculate the spectral signal of the insulating sample experiencing partial discharge. The characteristic frequency band of the partial discharge in the insulating sample is then obtained based on the spectral signal. For example, the characteristic frequency band of the partial discharge signal of one insulating sample is 400MHz to 1.5GHz, while the characteristic frequency band of the partial discharge signal of another insulating sample is 500MHz to 1GHz. For the partial discharge signal in this characteristic frequency band, a noise reduction algorithm can be used to reduce noise interference.

[0078] It is worth noting that the ultra-high frequency (UHF) detection method uses an UHF antenna sensor to detect the partial discharge signal of the insulating sample. Since the operating frequency of the UHF antenna sensor has a specific range, the characteristic frequency of the detected partial discharge signal should be within the operating frequency range of the UHF antenna sensor. If the operating frequency of the UHF antenna sensor is 300MHz-1.5GHz, then the minimum characteristic frequency of the partial discharge signal should not be lower than 300MHz, and the maximum characteristic frequency should not exceed 1.5GHz. For example, the characteristic frequency band of the partial discharge signal is 500MHz to 1GHz, or 400MHz to 1.5GHz.

[0079] It should be noted that when the cutoff frequency of the high-pass filter is 400MHz, the high-pass filter filters out low-frequency noise signals below 400MHz. However, the frequency band in which partial discharge occurs is 400MHz to 1.5GHz. The noise interference in this frequency band is not processed by the high-pass filter. Therefore, it is necessary to perform noise reduction processing on the partial discharge signals in the characteristic frequency band.

[0080] Understandably, in signal processing, a characteristic frequency band contains multiple characteristic frequencies, which refer to the inherent frequencies exhibited by a signal. These inherent frequencies are related to factors such as the signal source, the signal transmission medium, and its properties. Different insulating samples will have different corresponding characteristic frequencies.

[0081] Partial discharge tests are performed on insulating samples using a test voltage. The test voltage can be 1.5 to 2 times the partial discharge inception voltage (PDIV). Partial discharge tests are performed on insulating samples (such as solid epoxy resin insulating samples) at square wave voltage frequencies of 1kHz, 5kHz, 10kHz, 15kHz, and 20kHz, respectively. The partial discharge signals and test voltage waveform data are acquired and saved.

[0082] Understandably, based on the partial discharge signal and test voltage waveform data, the corresponding test voltage when the insulation sample undergoes partial discharge can be determined. Since the sampling frequency is set to be high, the amount of data obtained is large, for example, 300,000 samples can be selected.

[0083] In step S300, the partial discharge signal in the characteristic frequency band is decomposed into multiple modal components. Several modal components that meet the preset conditions are selected for noise reduction and signal reconstruction to obtain the denoised partial discharge signal of the insulating sample under high-frequency square wave voltage.

[0084] In this embodiment, the adaptive noise complete set empirical mode decomposition algorithm CEEMDAN can be used to decompose the signal to be processed (the partial discharge signal after filtering out low-frequency noise). That is, the CEEMDAN algorithm is used to perform mode decomposition on the partial discharge signal in the characteristic frequency band to obtain multiple mode components.

[0085] Specifically, firstly, Gaussian white noise with a mean of 0 is added to the signal to be processed K times to construct a sequence to be decomposed from K experiments. EMD empirical decomposition is then performed on this sequence to obtain the first intrinsic modal component (IMF), and its mean is taken as the first IMF component obtained from CEEMDAN decomposition. The remaining signals are processed by repeating the above steps until a set number of iterations is reached or the residual signal from the nth decomposition is a monotonic signal. The decomposition then stops, yielding the signal components, i.e., the intrinsic modal components (IMFs).

[0086] In one embodiment of the present invention, several suitable IMF components are selected based on the correlation coefficients calculated by CEEMDAN for wavelet denoising. These IMF components contain both a significant amount of noise and a high proportion of layout discharge signals. A higher correlation coefficient indicates less noise in the IMF component. IMF components with high correlation coefficients are retained, IMF components with medium correlation coefficients are selected for wavelet denoising, and IMF components with low correlation coefficients are discarded.

[0087] Please see Figure 4 and Figure 5 , Figure 4 The image shows the IMF components obtained from CEEMDAN decomposition. After signal decomposition, 14 modal components (IMFs) are obtained. Based on the correlation coefficients calculated by CEEMDAN, the correlation coefficients of this experiment are analyzed. Since IMF1 and IMF2 components contain less noise, their correlation coefficients are higher. Therefore, IMF1 and IMF2 components are retained. Since IMF3 to IMF12 components contain more noise, but partial discharge signals still account for a large proportion, their correlation coefficients are moderate (0.2 to 0.5). Therefore, a noise reduction algorithm is used to denoise IMF3 to IMF12 components. Since IMF13 and IMF14 components are almost all irregular noise signals, their correlation coefficients are close to zero (less than 0.2). Therefore, IMF13 and IMF14 components are discarded.

[0088] It should be noted that the correlation coefficient is used to measure the correlation of the processed IMF component signals. The larger the correlation coefficient, the higher the correlation, indicating that there is less noise in the IMF mode components; conversely, the lower the correlation, the higher the noise in the IMF mode components.

[0089] In this embodiment, the IMF components after wavelet denoising (such as the denoised IMF3 to IMF12 components) and the retained IMF components (such as IMF1 and IMF2 components) are reconstructed (such as signal superposition) to obtain the denoised partial discharge signal of the solid epoxy resin insulating sample under high frequency square wave voltage.

[0090] In one embodiment, the step of selecting several modal components that meet preset conditions for noise reduction and signal reconstruction to obtain the denoised partial discharge signal of the insulating sample under a high-frequency square wave voltage includes:

[0091] Based on the correlation coefficient of the modal components, multiple modal components are divided into: first modal component, second modal component and third modal component; the higher the correlation coefficient, the less noise is contained in the modal component; the correlation coefficient of the first modal component is greater than the first preset threshold, the correlation coefficient of the second modal component is greater than the second preset threshold and less than the first preset threshold, and the correlation coefficient of the third modal component is less than the second preset threshold;

[0092] Wavelet analysis was used to denoise the second mode component, resulting in the denoised second mode component.

[0093] Discarding the third mode component, the first mode component and the denoised second mode component are reconstructed to obtain the denoised partial discharge signal of the insulating sample under high-frequency square wave voltage.

[0094] like Figure 5 The correlation coefficients calculated by CEEMDAN, as shown, comprise 19 IMF components. Two IMF components (with correlation coefficients greater than 0.5, the first preset threshold) are retained. IMF components with correlation coefficients between 0.1 and 0.5 undergo noise reduction processing, while IMF components with correlation coefficients lower than 0.1, the second preset threshold, are discarded. Of course, the first and second preset thresholds can be adjusted based on the actual signal-to-noise ratio after noise reduction.

[0095] Specifically, when using wavelet analysis to denoise the second modal component, three factors are considered when selecting wavelet threshold denoising: the wavelet function for wavelet denoising can be sym8, the wavelet threshold can be selected using the ant colony algorithm, and the number of wavelet decomposition levels can be reasonably selected based on the calculation results. The Sym8 wavelet has a support range of 15, a vanishing moment of 8, and also possesses good regularity. Compared to the db wavelet, it has better symmetry, meaning it can reduce phase distortion during signal analysis and reconstruction to some extent. By assigning wavelet levels from 1 to N, a wavelet decomposition level with a high signal-to-noise ratio and short processing time can be found.

[0096] It should be noted that the Symlet wavelet function is an approximately symmetric wavelet function proposed by IngridDaubechies, and it is an improvement on the db function. The Symlet wavelet system is usually represented as symN (N=2,3,…,8).

[0097] In this embodiment of the invention, the interference noise characteristics of the partial discharge signal of the insulating sample under high-frequency square wave voltage are fully considered. First, the spectrum signal detected when PDIV is not reached is compared with the spectrum signal of the partial discharge signal to analyze and obtain the frequency band with noise. The cutoff frequency of the high-pass filter is set according to the noise frequency band. Before denoising the partial discharge signal, the noise of the high-frequency square wave power supply itself is filtered out by a high-performance high-order high-pass filter. Then, the high-frequency part of the partial discharge signal is further denoised. Specifically, the CEEMDAN decomposition method is used to decompose the noisy partial discharge signal to obtain multiple IMF components. Different processing methods are adopted for these IMF components to reduce the calculation time and improve the calculation accuracy. The IMF components with low noise content are retained and the IMF components with high noise content are discarded. The wavelet denoising method is used to denoise the IMF components with a lot of noise and partial discharge signal to obtain the denoised IMF components. The retained IMF components and the denoised IMF components are reconstructed to obtain the denoised partial discharge signal, which is a partial discharge characteristic signal with high signal-to-noise ratio.

[0098] This invention performs detection when no partial discharge signal occurs, filtering out low-frequency noise interference. For interference in the high-frequency portion of the partial discharge signal, a combination of CEEMDAN and wavelet denoising is used. Compared to traditional wavelet denoising methods, the larger the data volume and the higher the partial discharge frequency, the higher the signal-to-noise ratio and the smaller the root mean square error (RMSE), indicating better processing performance in this invention. While the error rate is not significantly different when the data volume is 100,000 rows, the error rate of this invention is significantly lower than that of traditional methods when the data volume is 200,000 rows or more.

[0099] This invention fully considers the interference and noise characteristics of partial discharge signals. Before denoising, a high-performance, high-order high-pass filter is selected to filter out the intrinsic power supply noise. This invention also fully considers the mode aliasing problem in EMD decomposition and the slow decomposition speed of EEMD and CEEMD, and therefore adopts the CEEMDAN decomposition method to decompose and reconstruct noisy partial discharge signals. This invention also fully considers different processing methods for each IMF component. The first mode component containing only partial discharge signals is retained, the second mode component containing both noise and partial discharge signals is denoised to obtain the denoised second mode component, and the third mode component containing only noise is discarded. The denoised second and third mode components are then reconstructed, which reduces computation time and improves computational accuracy.

[0100] This invention provides a method for processing partial discharge signals under high-frequency square wave voltage. It fully considers the noise interference of high-frequency square wave voltage on partial discharge signals, filtering out the noise of the high-frequency square wave power supply itself through a high-pass filter with a preset cutoff frequency, thus avoiding interference from the high-frequency square wave power supply to the partial discharge signal. The method acquires the partial discharge initiation voltage of the insulating sample, collects the partial discharge signal when the test voltage is the partial discharge initiation voltage, and obtains the characteristic frequency band of partial discharge in the insulating sample based on the partial discharge signal. Then, it performs mode decomposition on the partial discharge signal in the characteristic frequency band to obtain multiple mode components, selecting some mode components for signal denoising and signal reconstruction, which improves the accuracy of partial discharge signal denoising and obtains a partial discharge signal with a high signal-to-noise ratio under high-frequency square wave voltage, enabling more accurate analysis of the health status and insulation life of the insulating sample. Since only some mode components are selected for processing, the computation time is reduced while improving calculation accuracy. This method is applicable to situations with high sampling frequencies and large data volumes, meeting practical application requirements.

[0101] Please see Figure 7 The present invention provides an embodiment of a partial discharge signal processing device under high-frequency square wave voltage, comprising:

[0102] The power supply noise filtering module 11 is used to filter out the low-frequency noise signal of the test power supply using a high-pass filter with a preset cutoff frequency. The test power supply is a high-frequency square wave power supply for partial discharge testing of the insulating sample.

[0103] The characteristic frequency band acquisition module 22 is used to acquire the partial discharge initiation voltage of the insulating sample, collect the partial discharge signal when the test voltage is the partial discharge initiation voltage, and obtain the characteristic frequency band of the partial discharge of the insulating sample based on the partial discharge signal, wherein the test voltage is the high-frequency square wave voltage of the test power supply.

[0104] The signal denoising and reconstruction module 33 is used to perform mode decomposition on the partial discharge signal of the characteristic frequency band to obtain multiple mode components, select several mode components that meet preset conditions for denoising processing and signal reconstruction, and obtain the denoised partial discharge signal of the insulating sample under high frequency square wave voltage.

[0105] This invention provides a partial discharge signal processing device under high-frequency square wave voltage. It fully considers the noise interference of high-frequency square wave voltage on partial discharge signals, filtering out the noise of the high-frequency square wave power supply itself through a high-pass filter with a preset cutoff frequency, thus avoiding interference from the high-frequency square wave power supply to the partial discharge signal. The device acquires the partial discharge initiation voltage of the insulating sample, collects the partial discharge signal when the test voltage is the partial discharge initiation voltage, and obtains the characteristic frequency band of partial discharge in the insulating sample based on the partial discharge signal. Then, it performs mode decomposition on the partial discharge signal in the characteristic frequency band to obtain multiple mode components, selecting some mode components for signal denoising and signal reconstruction, which improves the accuracy of partial discharge signal denoising and obtains a partial discharge signal with a high signal-to-noise ratio under high-frequency square wave voltage, enabling more accurate analysis of the health status and insulation life of the insulating sample. Since only some mode components are selected for processing, the computation time is reduced while improving calculation accuracy. This device is suitable for situations with high sampling frequency and large data volume, meeting practical application requirements.

[0106] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0107] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0108] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0110] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0111] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A partial discharge signal processing method under a high frequency square wave voltage, characterized by, The application relates to a method for testing partial discharge of an insulation sample. The low-frequency noise signal of a test power supply is filtered by a high-pass filter with a preset cutoff frequency, the test power supply being a high-frequency square wave power supply for testing partial discharge of the insulation sample; The partial discharge starting voltage of the insulation sample is obtained, the partial discharge signal when the test voltage is the partial discharge starting voltage is collected, and the characteristic frequency band of the partial discharge of the insulation sample is obtained according to the partial discharge signal, the test voltage being the high-frequency square wave voltage of the test power supply; The partial discharge signal of the characteristic frequency band is subjected to modal decomposition to obtain a plurality of modal components, a plurality of modal components satisfying preset conditions are selected for noise reduction processing and signal reconstruction to obtain the noise-reduced partial discharge signal of the insulation sample under the high-frequency square wave voltage. The step of selecting a plurality of modal components satisfying preset conditions for noise reduction processing and signal reconstruction to obtain the noise-reduced partial discharge signal of the insulation sample under the high-frequency square wave voltage comprises: The plurality of modal components are divided into a first modal component, a second modal component and a third modal component according to the correlation coefficients of the modal components; the higher the correlation coefficient, the less noise contained in the modal component; the correlation coefficient of the first modal component is greater than a first preset threshold, the correlation coefficient of the second modal component is greater than a second preset threshold and less than the first preset threshold, and the correlation coefficient of the third modal component is less than the second preset threshold; The second modal component is subjected to signal noise reduction by wavelet analysis to obtain a noise-reduced second modal component; The third modal component is discarded, and the first modal component and the noise-reduced second modal component are subjected to signal reconstruction to obtain the noise-reduced partial discharge signal of the insulation sample under the high-frequency square wave voltage.

2. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The step of filtering the low-frequency noise signal of the test power supply by the high-pass filter with the preset cutoff frequency further comprises: A first frequency spectrum signal when the insulation sample does not have partial discharge and a second frequency spectrum signal when the insulation sample has partial discharge are obtained; The first frequency spectrum signal and the second frequency spectrum signal are compared and analyzed to obtain a frequency band where the low-frequency noise appears; The cutoff frequency of the high-pass filter is set according to the frequency band where the low-frequency noise appears.

3. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The step of obtaining the partial discharge starting voltage of the insulation sample comprises: A plurality of preset insulation samples are tested by using a voltage boosting method to obtain test voltages when the insulation samples have partial discharge, and the average value of the test voltages is taken as the partial discharge starting voltage of the insulation sample.

4. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The step of collecting the partial discharge signal when the test voltage is the partial discharge starting voltage comprises: The partial discharge signal when the test voltage is the partial discharge starting voltage is collected by using a UHF antenna sensor.

5. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The step of obtaining the characteristic frequency band of the partial discharge of the insulation sample according to the partial discharge signal comprises: The partial discharge signal is subjected to fast Fourier transform to obtain a frequency spectrum signal when the insulation sample has partial discharge; The characteristic frequency band of the partial discharge of the insulation sample is obtained according to the frequency spectrum signal.

6. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The step of obtaining a plurality of modal components by subjecting the partial discharge signal of the characteristic frequency band to modal decomposition comprises: The adaptive noise complete ensemble empirical mode decomposition algorithm CEEMDAN is used for mode decomposition of the partial discharge signal in the characteristic frequency band, to obtain a plurality of mode components.

7. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The wavelet analysis is used for signal denoising of the second mode component. The wavelet function used in the wavelet analysis is a Sym8 function.

8. The partial discharge signal processing method under high frequency square wave voltage according to claim 1, characterized by, The first mode component and the denoised second mode component are subjected to signal reconstruction. The first mode component and the denoised second mode component are subjected to signal superposition.

9. A partial discharge signal processing apparatus under a high frequency square wave voltage, characterized by, Comprise: The power supply noise filtering module is configured to filter low-frequency noise signals of a test power supply by using a high-pass filter with a preset cutoff frequency, wherein the test power supply is a high-frequency square wave power supply for partial discharge testing of an insulation sample; The characteristic frequency band acquisition module is configured to acquire a partial discharge inception voltage of the insulation sample, collect a partial discharge signal when a test voltage is the partial discharge inception voltage, and obtain a characteristic frequency band of partial discharge of the insulation sample according to the partial discharge signal, wherein the test voltage is a high-frequency square wave voltage of the test power supply; The signal denoising and reconstruction module is configured to perform mode decomposition on the partial discharge signal in the characteristic frequency band to obtain a plurality of mode components, select a plurality of mode components satisfying a preset condition for denoising and signal reconstruction, and obtain a denoised partial discharge signal of the insulation sample under the high-frequency square wave voltage. The selection of a plurality of mode components satisfying a preset condition for denoising and signal reconstruction to obtain a denoised partial discharge signal of the insulation sample under the high-frequency square wave voltage specifically comprises: According to the correlation coefficient of the mode component, the plurality of mode components are divided into a first mode component, a second mode component, and a third mode component; the higher the correlation coefficient, the less noise contained in the mode component; the correlation coefficient of the first mode component is greater than a first preset threshold, the correlation coefficient of the second mode component is greater than a second preset threshold and less than the first preset threshold, and the correlation coefficient of the third mode component is less than the second preset threshold; The wavelet analysis is used for signal denoising of the second mode component, to obtain a denoised second mode component; The third mode component is discarded, and the first mode component and the denoised second mode component are subjected to signal reconstruction, to obtain a denoised partial discharge signal of the insulation sample under the high-frequency square wave voltage.

Images