Power equipment partial discharge signal denoising method and system

Abstract

The invention relates to the technical field of signal processing, and discloses a power equipment partial discharge signal denoising method and system, and the method comprises the steps: S1, generating an exponential decay type pulse signal simulating partial discharge, and superposing white noise and narrow-band interference into the exponential decay type pulse signal to generate a noisy signal; s2, decomposing the noisy signal into a plurality of intrinsic mode components through an ensemble empirical mode decomposition method, calculating a correlation coefficient and a kurtosis index of each intrinsic mode component, and identifying a noise component and a signal component based on the correlation coefficient and the kurtosis index; s3, wavelet threshold denoising is carried out on the noise components, and a wavelet threshold is dynamically adjusted for each noise component through a particle swarm optimization algorithm; and S4, reconstructing the de-noised noise component and the reserved signal component into a final de-noised signal, and evaluating the de-noising performance based on the signal-to-noise ratio, the root mean square error, the peak signal-to-noise ratio and the correlation coefficient.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular to a method and system for denoising partial discharge signals of power equipment. Background Art

[0002] Partial discharge (PD) is an important signal indicative of insulation degradation in power equipment, and its accurate detection is crucial for safe equipment operation. However, actual PD signals collected are often affected by white noise and narrowband interference, resulting in a low signal-to-noise ratio (SNR), making it difficult to extract effective signal features.

[0003] Common denoising methods used in existing technologies include Fourier transforms, wavelet transforms, and empirical mode decomposition (EMD). For example, wavelet transforms denoise signals using a preset threshold, but threshold selection relies on experience and is difficult to adapt to dynamic noise environments. While EMD can decompose non-stationary signals, it is susceptible to modal aliasing and is sensitive to white noise. Furthermore, existing methods often require additional filtering steps when processing narrowband interference, increasing computational complexity and the need for manual intervention. However, existing methods still have the following shortcomings in partial discharge signal processing: 1. After EEMD decomposition, the identification rules of noise IMF are relatively simple, which may misjudge the signal components; 2. Wavelet threshold optimization lacks specificity and cannot adapt to the noise characteristics of different IMFs; 3. The effect of removing complex interference (such as narrowband interference) is limited.

[0004] To address these issues, a PD signal processing method with strong adaptability and excellent denoising performance is urgently needed to improve signal extraction accuracy in complex noisy environments. To this end, this paper proposes an improved denoising method that optimizes the IMF identification rules of EEMD and combines it with the PSO algorithm to adaptively adjust the wavelet threshold for each noise IMF, further improving the denoising performance of PD signals. Summary of the Invention

[0005] The present invention provides a method and system for denoising partial discharge signals of electric power equipment, so as to solve the problems of poor signal extraction accuracy and limited denoising effect in existing discharge signal denoising methods.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions: In a first aspect, the present invention provides a method for denoising a partial discharge signal of an electric power device, comprising: S1. Generate an exponentially decaying pulse signal simulating partial discharge, and superimpose white noise and narrowband interference on the exponentially decaying pulse signal to generate a noisy signal; S2. Decomposing the noisy signal into multiple intrinsic modal components by using an ensemble empirical mode decomposition method, calculating the correlation coefficient and kurtosis index of each intrinsic modal component, and identifying the noise component and the signal component based on the correlation coefficient and kurtosis index; S3, performing wavelet threshold denoising on the noise component, wherein the wavelet threshold is dynamically adjusted for each noise component through a particle swarm optimization algorithm; S4. Reconstruct the final denoised signal from the denoised noise component and the retained signal component, and evaluate the denoising performance based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio, and correlation coefficient.

[0007] Optionally, in S1, the generation expression of the exponential decay pulse signal satisfies the following relationship: ; Where A is the pulse amplitude, α is the attenuation coefficient, f is the pulse frequency, t1 and t2 are the pulse generation times, and u(t) is the unit step function.

[0008] Optionally, in S1, superimposing white noise and narrowband interference onto the exponentially decaying pulse signal to generate a noisy signal includes: Generate Gaussian white noise. The generation of Gaussian white noise satisfies the following relationship: ; Generate narrowband interference by superimposing two sine waves. The generation of narrowband interference satisfies the following relationship: ; The generated Gaussian white noise and narrowband interference are superimposed on the exponential decay pulse signal to obtain a noisy signal, wherein the superposition satisfies the following relationship: ; Where, is a noisy signal, is an exponential decay pulse signal. is Gaussian white noise, It is narrowband interference.

[0009] Optionally, in S2, decomposing the noisy signal into multiple intrinsic mode components by using an ensemble empirical mode decomposition method includes: Decomposition parameters are set according to the denoising accuracy requirements, where the decomposition parameters include: number of collections, noise standard deviation, and maximum number of natural mode components; Performing empirical mode decomposition on the noisy signal based on the number of collections, the noise standard deviation, and the maximum number of intrinsic modal components to obtain intrinsic modal components equal to the maximum number of intrinsic modal components.

[0010] Optionally, in S2, calculating the correlation coefficient and kurtosis index of each intrinsic modal component includes: Calculate the correlation coefficient of each natural modal component, where the calculation of the correlation coefficient satisfies the following relationship: ; Calculate the kurtosis of each natural mode component, where the calculation of the kurtosis satisfies the following relationship: ; Where, is the mean.

[0011] Optionally, in S2, identifying the noise component and the signal component based on the correlation coefficient and the kurtosis index includes: Comparing the correlation coefficient with a first preset threshold, and when the correlation coefficient is greater than the first preset threshold, identifying the natural mode component corresponding to the correlation coefficient as a signal component; When the correlation coefficient is less than or equal to the first preset threshold, the correlation coefficient is compared with the second preset threshold or the kurtosis index is compared with the third preset threshold; when the correlation coefficient is less than the second preset threshold or the kurtosis index is greater than the third preset threshold, the natural mode component corresponding to the correlation coefficient is identified as a noise component; When the above steps do not identify the noise components, the first three natural mode components are identified as noise components.

[0012] Optionally, in S3, performing wavelet threshold denoising on the noise component includes: Obtain the wavelet basis function and calculate the initial threshold of the wavelet basis function, wherein the calculation method of the initial threshold satisfies the following relationship: ; Where, б is the standard deviation of wavelet coefficient noise, N is the signal length; The fitness function is determined by combining the root mean square error and the signal-to-noise ratio, where the fitness function satisfies the following relationship: ; The threshold factor is optimized independently for each noise component, and the particle swarm optimization method is used to iteratively search for the global optimal solution of the threshold factor to obtain the final threshold; The noise component is denoised using the final threshold to obtain the denoised noise component.

[0013] Optionally, in S4, reconstructing the denoised noise component and the retained signal component into a final denoised signal includes: The denoised noise component is added to the retained signal component to obtain the final denoised signal.

[0014] Optionally, in S4, evaluating the denoising performance based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio, and correlation coefficient includes: The signal-to-noise ratio, root mean square error, peak signal-to-noise ratio and correlation coefficient of the denoised noise component are calculated, and the denoising effects of different wavelet basis functions are compared based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio and correlation coefficient.

[0015] In a second aspect, an embodiment of the present application provides a system for denoising partial discharge signals of power equipment, including a processor and a memory; Memory for storing computer programs; The processor is configured to implement any one of the method steps described in the first aspect when executing a program stored in the memory.

[0016] Beneficial effects: The power equipment partial discharge signal denoising method provided by the present invention identifies the inherent modal components of noise through ensemble empirical mode decomposition combined with cross-correlation and kurtosis, and dynamically optimizes the threshold through particle swarm optimization to adapt to different noise levels. Compared with traditional methods, the signal-to-noise ratio is improved by about 80%, the root mean square error is reduced by about 38%, the signal fidelity is significantly improved, and white noise and narrowband interference are effectively suppressed. It is suitable for complex environments such as substations. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a method for denoising partial discharge signals of power equipment according to a preferred embodiment of the present invention; Figure 2 This is an IMF waveform diagram after EEMD classification in a preferred embodiment of the present invention; Figure 3 This is the convergence curve of the FITNESS value during the PSO optimization process of the preferred embodiment of the present invention; Figure 4 A comparison diagram of signal waveforms before and after denoising according to a preferred embodiment of the present invention; Figure 5 Spectrum diagram of the signal before and after denoising according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following is a clear and complete description of the technical solutions of the present invention. It should be understood that the embodiments described are only a portion of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0019] Unless otherwise defined, the technical or scientific terms used in the present invention shall have the usual meanings understood by persons of ordinary skill in the field to which the present invention belongs. The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "a" do not indicate a quantity limitation, but rather indicate the existence of at least one. Words such as "connected" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship also changes accordingly.

[0020] See Figure 1-5 , an embodiment of the present application provides a method for denoising a partial discharge signal of an electric power device, comprising: S1. Generate an exponentially decaying pulse signal simulating partial discharge, and superimpose white noise and narrowband interference on the exponentially decaying pulse signal to generate a noisy signal; S2. Decomposing the noisy signal into multiple intrinsic modal components by using an ensemble empirical mode decomposition method, calculating the correlation coefficient and kurtosis index of each intrinsic modal component, and identifying the noise component and the signal component based on the correlation coefficient and kurtosis index; S3, performing wavelet threshold denoising on the noise component, wherein the wavelet threshold is dynamically adjusted for each noise component through a particle swarm optimization algorithm; S4. Reconstruct the final denoised signal from the denoised noise component and the retained signal component, and evaluate the denoising performance based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio, and correlation coefficient.

[0021] In the above embodiment, the denoising method is refined into the following steps: Step 1: Signal generation and preprocessing Partial discharge signal simulation: Generates exponentially decaying pulse signals to simulate partial discharge characteristics.

[0022] The signal expression is: ; Where A is the pulse amplitude, α is the attenuation coefficient, f is the pulse frequency, t1 and t2 are the pulse generation times, and u(t) is the unit step function.

[0023] Superposition of noise and interference: Add white noise: Generate Gaussian white noise, and adjust the amplitude according to the initial signal-to-noise ratio (SNR=10). The formula is: ; Add narrowband interference: superimpose two sine waves, ; The frequencies are f1 and f2, and the amplitudes are .

[0024] The final noisy signal is: ; Step 2: EEMD decomposition 1. EEMD algorithm implementation Parameter settings: the number of collections is 50, the noise standard deviation is 0.2, and the maximum number of IMFs is 10.

[0025] Decomposition process: for noisy signals After adding random white noise multiple times, perform EMD decomposition to obtain a set of IMF components, and take the average value to eliminate the influence of modal aliasing. Each decomposition generates a set of IMFs, and the final average IMF is:

[0026] 2. Noise IMF Identification Identification indicators: Calculate the correlation coefficient and kurtosis of each IMF with the original signal: Correlation coefficient: ; Kurtosis: ; Identification rules: 1. Protect IMFs with a cross-correlation coefficient greater than 0.7 (considered as signal components); 2. For the remaining IMFs, identify the components with a cross-correlation coefficient less than 0.5 or a kurtosis greater than 4 as noise IMFs; 3. If no noise IMF meets the requirements, the first three IMFs (high-frequency components) are assumed to be noise components.

[0027] Step 3: PSO optimized wavelet threshold denoising 1. Wavelet basis selection Select a variety of wavelet basis functions (such as haar, db3, db6, db9, bior3.7, bior5.5), and the decomposition level is 3. Perform wavelet decomposition on each noise IMF and calculate the initial threshold: ; Among them, б is the standard deviation of wavelet coefficient noise, and N is the signal length.

[0028] 2. PSO optimization threshold factor Parameter settings: number of particles 50, maximum number of iterations 100, inertia weight linearly decays from 1.2 to 0.4, learning factor c1=c2=2.5.

[0029] Fitness function: Comprehensive RMSE and SNR, defined as: ; Optimization process: The threshold factor (ranging from 0.1 to 1) is optimized independently for each noise IMF, and the global optimal solution is searched through PSO iteration.

[0030] 3. Denoising Execution A soft threshold function is applied to the noise IMF for denoising, while the signal-related IMF remains unchanged.

[0031] Step 4: Signal reconstruction and performance evaluation Reconstruction: Add the denoised IMF to the retained signal IMF to obtain the denoised signal .

[0032] Evaluation indicators: Calculate the SNR, RMSE, PSNR and correlation coefficient before and after denoising, compare the effects of different wavelet bases, and compare the signal quality before and after denoising The embodiment of the present application also provides a system for denoising partial discharge signals of power equipment, including a processor and a memory; Memory for storing computer programs; The processor is configured to implement any one of the method steps described in the method for denoising partial discharge signals of electric power equipment when executing a program stored in the memory.

[0033] The above-mentioned power equipment partial discharge signal denoising system can implement various embodiments of the above-mentioned power equipment partial discharge signal denoising method and achieve the same beneficial effects, which will not be described in detail here.

[0034] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A method for denoising partial discharge signals of power equipment, characterized in that: include: S1. Generate an exponentially decaying pulse signal simulating partial discharge, and superimpose white noise and narrowband interference on the exponentially decaying pulse signal to generate a noisy signal; S2. Decomposing the noisy signal into multiple intrinsic modal components by using an ensemble empirical mode decomposition method, calculating the correlation coefficient and kurtosis index of each intrinsic modal component, and identifying the noise component and the signal component based on the correlation coefficient and kurtosis index; S3, performing wavelet threshold denoising on the noise component, wherein the wavelet threshold is dynamically adjusted for each noise component through a particle swarm optimization algorithm; S4. Reconstruct the final denoised signal from the denoised noise component and the retained signal component, and evaluate the denoising performance based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio, and correlation coefficient.

2. The method for denoising partial discharge signals of power equipment according to claim 1, characterized in that: In S1, the generation expression of the exponential decay pulse signal satisfies the following relationship: ； Where s(t) is an exponential decay pulse signal, A is the pulse amplitude, α is the attenuation coefficient, f is the pulse frequency, t is the time, t1 and t2 are the pulse generation times, and u(t) is the unit step function.

3. The method for denoising partial discharge signals of power equipment according to claim 1, characterized in that: In S1, superimposing white noise and narrowband interference on the exponentially decaying pulse signal to generate a noisy signal includes: Generate Gaussian white noise. The generation of Gaussian white noise satisfies the following relationship: ； Generate narrowband interference by superimposing two sine waves. The generation of narrowband interference satisfies the following relationship: ； The generated Gaussian white noise and narrowband interference are superimposed on the exponential decay pulse signal to obtain a noisy signal, wherein the superposition satisfies the following relationship: ； Where, is a noisy signal, is an exponential decay pulse signal. is Gaussian white noise, is narrowband interference, is the noise variance, is the standard deviation of the original signal, To convert the signal-to-noise ratio from decibels to amplitude ratio, is the standard deviation of the standard normal distribution random numbers, To preset the signal-to-noise ratio, is the interference amplitude, 、 is the interference frequency.

4. The method for denoising partial discharge signals of power equipment according to claim 1, characterized in that: In S2, the noisy signal is decomposed into multiple intrinsic modal components by using an ensemble empirical mode decomposition method, including: Decomposition parameters are set according to the denoising accuracy requirements, where the decomposition parameters include: number of collections, noise standard deviation, and maximum number of natural mode components; Performing empirical mode decomposition on the noisy signal based on the number of collections, the noise standard deviation, and the maximum number of intrinsic modal components to obtain intrinsic modal components equal to the maximum number of intrinsic modal components.

5. The method for denoising partial discharge signals of electric power equipment according to claim 1, characterized in that: In S2, the correlation coefficient and kurtosis index of each intrinsic modal component are calculated, including: Calculate the correlation coefficient of each natural modal component, where the calculation of the correlation coefficient satisfies the following relationship: ； Calculate the kurtosis of each natural mode component, where the calculation of the kurtosis satisfies the following relationship: ； Where, is the mean, is the jth natural mode component, is the correlation coefficient between the j-th IMF component and the original signal, is the covariance, is the variance, is the kurtosis of the j-th IMF component, N is the signal length, and n is the sampling point index.

6. The method for denoising partial discharge signals of electric power equipment according to claim 1, characterized in that: In S2, identifying the noise component and the signal component based on the correlation coefficient and the kurtosis index includes: Comparing the correlation coefficient with a first preset threshold, and when the correlation coefficient is greater than the first preset threshold, identifying the natural mode component corresponding to the correlation coefficient as a signal component; When the correlation coefficient is less than or equal to the first preset threshold, the correlation coefficient is compared with the second preset threshold or the kurtosis index is compared with the third preset threshold; when the correlation coefficient is less than the second preset threshold or the kurtosis index is greater than the third preset threshold, the natural mode component corresponding to the correlation coefficient is identified as a noise component; When the above steps do not identify the noise components, the first three natural mode components are identified as noise components.

7. The method for denoising partial discharge signals of power equipment according to claim 1, characterized in that: In S3, performing wavelet threshold denoising on the noise component includes: Obtain the wavelet basis function and calculate the initial threshold of the wavelet basis function, wherein the calculation method of the initial threshold satisfies the following relationship: ； Where N is the signal length, is the basic wavelet threshold of the j-th noise component, is the noise standard deviation of the j-th layer wavelet coefficient, is the detail coefficient of the Lth layer of wavelet decomposition, is the median of the absolute values of detail coefficients; The fitness function is determined by combining the root mean square error and the signal-to-noise ratio, where the fitness function satisfies the following relationship: ； Where, is the fitness function value, is the root mean square error, is the signal-to-noise ratio; The threshold factor is optimized independently for each noise component, and the particle swarm optimization method is used to iteratively search for the global optimal solution of the threshold factor to obtain the final threshold; The noise component is denoised using the final threshold to obtain the denoised noise component.

8. The method for denoising partial discharge signals of electric power equipment according to claim 1, characterized in that: In S4, reconstructing the denoised noise component and the retained signal component into a final denoised signal includes: The denoised noise component is added to the retained signal component to obtain the final denoised signal.

9. The method for denoising partial discharge signals of electric power equipment according to claim 1, characterized in that: In S4, the denoising performance is evaluated based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio, and correlation coefficient, including: The signal-to-noise ratio, root mean square error, peak signal-to-noise ratio and correlation coefficient of the denoised noise component are calculated, and the denoising effects of different wavelet basis functions are compared based on the signal-to-noise ratio, root mean square error, peak signal-to-noise ratio and correlation coefficient.

10. A partial discharge signal denoising system for power equipment, characterized in that: Including processor and memory; Memory for storing computer programs; A processor, configured to implement the method steps described in any one of claims 1 to 9 when executing a program stored in a memory.

Images