Prps spectrum denoising method and device, storage medium and computer device

Abstract

The PRPS spectrum denoising method and device, the storage medium and the computer device provided by the application comprise the following steps: obtaining an original discharge signal and determining a corresponding PRPS spectrum; performing a first denoising operation on the original discharge signal to obtain a first denoising result and determine a reconstructed discharge signal; then performing a second denoising operation on the PRPS spectrum to obtain a second denoising result and determine a reconstructed PRPS spectrum; performing denoising on the PRPS spectrum by using a denoising operation capable of denoising the PRPS spectrum, thereby obtaining a reconstructed PRPS spectrum with high quality; and finally performing linear weighted fusion on the reconstructed discharge signal and the reconstructed PRPS spectrum based on the accuracy of the first denoising operation and the accuracy of the second denoising operation, thereby obtaining a target PRPS spectrum. In this way, the contribution of different denoising operations to subsequent identification results can be considered, thereby reducing noise in the target PRPS spectrum obtained finally and improving the quality of partial discharge data.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a PRPS spectral noise reduction method, apparatus, storage medium, and computer equipment. Background Technology

[0002] The condition of electrical equipment is a crucial factor affecting the stable operation of power systems. Partial discharge (PD) problems, caused by various factors, can lead to irreversible damage to equipment due to their initially insidious nature and long-term accumulation. However, the high degree of uncertainty at the equipment's location, the high coupling of internal components, and the extremely complex actual discharge process result in a wide range of atypical partial discharge signals acquired in the field. This makes current algorithms for partial discharge pattern recognition perform poorly in practical applications.

[0003] Atypical features in partial discharge signals often include interference signal characteristics caused by the influence of the on-site environment and the high coupling of internal components of the equipment. That is, the discharge signal and interference signal are superimposed. It is precisely because different signals have different amplitudes, frequencies, phases, etc. that the original signal characteristics are destroyed and it is difficult for pattern recognition algorithms to correctly identify them.

[0004] Currently, interference removal methods typically remove noise signals from the time-domain waveforms acquired by partial discharge monitoring systems through filtering and other means. However, after converting the time-domain waveforms into PRPS (Phase Resolved Pluse Sequence) spectral data and uploading it, the PRPS spectral data still retains many interference features. Since PRPS spectral data is two-dimensional, existing interference removal methods are difficult to apply directly to this type of data. Therefore, existing noise removal methods have certain limitations and cannot cover every data processing step before discharge mode identification, leading to a decrease in the quality of partial discharge data and consequently reducing the accuracy of subsequent identification, diagnosis, and other algorithms. Summary of the Invention

[0005] The purpose of this application is to at least solve one of the above-mentioned technical defects, in particular the technical defect that existing noise removal methods have certain limitations, making it difficult to cover every data processing process before discharge mode identification, resulting in a reduction in the quality of partial discharge data, and thus reducing the accuracy of subsequent identification, diagnosis and other algorithms in application.

[0006] In a first aspect, this application provides a PRPS map denoising method, the method comprising:

[0007] Acquire the original discharge signal and determine the corresponding PRPS spectrum based on the original discharge signal;

[0008] The original discharge signal is subjected to a first denoising operation to obtain a first denoising result, and the first denoising result is reconstructed to obtain a reconstructed discharge signal;

[0009] A second denoising operation is performed on the PRPS map to obtain a second denoising result, and the second denoising result is reconstructed to obtain a reconstructed PRPS map;

[0010] The accuracy rates of the first denoising operation and the second denoising operation are obtained respectively, and the first weight and the second weight are determined based on the accuracy rates of the first denoising operation and the second denoising operation respectively.

[0011] Based on the first weight and the second weight, the reconstructed discharge signal and the reconstructed PRPS spectrum are linearly weighted and fused to obtain the target PRPS spectrum.

[0012] In one embodiment, the step of performing a first denoising operation on the original discharge signal includes:

[0013] Based on the original discharge signal, determine the corresponding wavelet basis function and scale parameter set;

[0014] Each scale parameter of the scale parameter set and the original discharge signal are respectively input into the wavelet basis function to determine the wavelet coefficients of the original discharge signal under each scale parameter of the scale parameter set;

[0015] Based on the wavelet coefficients corresponding to the original discharge signal at each scale parameter in the scale parameter set, a wavelet coefficient matrix is ​​generated, and the wavelet coefficient matrix is ​​thresholded. The thresholded wavelet coefficient matrix is ​​then determined as the first denoising result.

[0016] In one embodiment, the step of performing a second denoising operation on the PRPS map includes:

[0017] The PRPS map is iteratively decomposed;

[0018] In a single decomposition iteration, the decomposition object in this decomposition iteration is determined; wherein, in the first decomposition iteration, the decomposition object is the PRPS map;

[0019] The decomposed object is filtered to obtain its high-frequency coefficients and low-frequency coefficients.

[0020] The high-frequency coefficients are thresholded to update them, and the low-frequency coefficients are determined as the decomposition objects in the next decomposition iteration. This process continues until the number of decomposition iterations reaches a preset number of decomposition layers. Based on the high-frequency and low-frequency coefficients in each decomposition iteration, a second denoising result is generated.

[0021] In one embodiment, the thresholding of the high-frequency coefficients includes:

[0022] Determine the target threshold and threshold function for this decomposition iteration;

[0023] Based on the target threshold and threshold function in this decomposition iteration, the high-frequency coefficients are quantized, and the high-frequency coefficients in this decomposition iteration are updated based on the quantized high-frequency coefficients.

[0024] In one embodiment, the step of linearly weighting and fusing the reconstructed discharge signal with the reconstructed PRPS spectrum based on the first weight and the second weight includes:

[0025] The reconstructed PRPS map is normalized to obtain the first PRPS map;

[0026] Using the first weight as the fusion weight of the reconstructed discharge signal and the second weight as the fusion weight of the reconstructed PRPS spectrum, the reconstructed discharge signal and the reconstructed PRPS spectrum are linearly weighted and fused to obtain the target PRPS spectrum.

[0027] In one embodiment, the step of determining the corresponding PRPS spectrum based on the original discharge signal includes:

[0028] The signal data in the original discharge signal are rearranged to obtain the corresponding two-dimensional array;

[0029] The two-dimensional array is filled with zero values ​​in the phase dimension, and the phase window corresponding to the zero-filled two-dimensional array is divided to obtain each target phase window;

[0030] Average pooling is used to filter the data within each target phase window, and PRPS maps are generated based on the filtered data.

[0031] In one embodiment, the method further includes:

[0032] Obtain historical discrimination data of the first denoising operation and historical discrimination data of the second denoising operation;

[0033] The discrimination accuracy is calculated in the historical discrimination data of the first denoising operation, and the accuracy corresponding to the first denoising operation is updated according to the discrimination accuracy.

[0034] The discrimination accuracy is calculated from the historical discrimination data of the second denoising operation, and the accuracy corresponding to the second denoising operation is updated based on the discrimination accuracy.

[0035] Secondly, this application provides a PRPS map noise reduction device, the device comprising:

[0036] The signal acquisition module is used to acquire the original discharge signal and determine the corresponding PRPS spectrum based on the original discharge signal.

[0037] The first denoising module is used to perform a first denoising operation on the original discharge signal to obtain a first denoising result, and to reconstruct the first denoising result to obtain a reconstructed discharge signal.

[0038] The second denoising module is used to perform a second denoising operation on the PRPS map to obtain a second denoising result, and to reconstruct the second denoising result to obtain a reconstructed PRPS map.

[0039] The weight determination module is used to obtain the accuracy of the first denoising operation and the second denoising operation respectively, and determine the first weight and the second weight respectively based on the accuracy of the first denoising operation and the second denoising operation.

[0040] The weighted fusion module is used to linearly weight and fuse the reconstructed discharge signal with the reconstructed PRPS spectrum based on the first weight and the second weight to obtain the target PRPS spectrum.

[0041] Thirdly, this application provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the PRPS map denoising method as described in any of the above embodiments.

[0042] Fourthly, this application provides a computer device, including: one or more processors, and a memory;

[0043] The memory stores computer-readable instructions, and when the one or more processors execute the computer-readable instructions, they perform the steps of the PRPS map denoising method as described in any of the above embodiments.

[0044] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0045] This application provides a PRPS image denoising method, apparatus, storage medium, and computer equipment. The method includes: acquiring an original discharge signal and determining a corresponding PRPS image based on the original discharge signal; performing a first denoising operation on the original discharge signal to obtain a first denoising result, and reconstructing the first denoising result to obtain a reconstructed discharge signal. A suitable denoising operation is used to denoise the original discharge signal, thereby reducing interference features in the original discharge signal. Next, a second denoising operation is performed on the PRPS image to obtain a second denoising result, and the second denoising result is reconstructed to obtain a reconstructed PRPS image; a denoising operation capable of denoising the PRPS image is used to denoise it, thereby obtaining a high-quality reconstructed PRPS image. Finally, based on the accuracy of the first denoising operation and the accuracy of the second denoising operation, the reconstructed discharge signal and the reconstructed PRPS image are linearly weighted and fused to obtain a target PRPS image. By determining the weights of reconstructing the discharge signal and reconstructing the PRPS spectrum in linear weighted fusion based on the accuracy of the first and second denoising operations, the contribution of different denoising operations to the subsequent identification results can be considered, thereby further reducing the noise in the final target PRPS spectrum, improving the quality of partial discharge data, and thus improving the accuracy of subsequent identification, diagnosis and other algorithms in application. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A schematic flowchart of a PRPS map denoising method provided in an embodiment of this application;

[0048] Figure 2 A flowchart illustrating the steps of performing a first denoising operation on the original discharge signal according to an embodiment of this application;

[0049] Figure 3 A flowchart illustrating the steps of performing a second denoising operation on a PRPS map, as provided in an embodiment of this application;

[0050] Figure 4 An example diagram illustrating a PRPS map denoising method provided in this application embodiment;

[0051] Figure 5 This is a schematic diagram of the structure of a PRPS image noise reduction device provided in an embodiment of this application;

[0052] Figure 6 This is an internal structural diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] In one embodiment, this application provides a PRPS map denoising method, and the following embodiments illustrate the application of this method to a server. It is understood that the PRPS map denoising method can be performed on a single server or a server cluster consisting of multiple servers; this application does not impose any specific limitations on this.

[0055] like Figure 1 As shown, this application provides a PRPS map denoising method, the method comprising:

[0056] Step S101: Obtain the original discharge signal and determine the corresponding PRPS spectrum based on the original discharge signal.

[0057] The original discharge signal refers to the discharge signal acquired by using a partial discharge detection instrument to detect partial discharge. The PRPS (Partial Discharge Perimeter) spectrum is a graphical representation used to analyze partial discharge signals in power equipment. It can display the distribution and characteristics of partial discharge signals at different phases (angles).

[0058] In this step, partial discharge can be collected using a partial discharge detection instrument to obtain the original discharge signal. The original discharge signal is then further processed to determine the PRPS spectrum corresponding to the original discharge signal.

[0059] Furthermore, data can be collected manually using instruments and uploaded to a server for processing, or it can be collected and uploaded using a partial discharge detection device. This application does not impose specific limitations in this regard.

[0060] It is understandable that partial discharge detection instruments can use capacitive partial discharge detectors, UHF (ultra-high frequency) partial discharge detectors, and other detectors.

[0061] Step S102: Perform a first denoising operation on the original discharge signal to obtain a first denoising result, and reconstruct the first denoising result to obtain a reconstructed discharge signal.

[0062] The first denoising operation is a method that combines data processing and thresholding to denoise.

[0063] In this step, when the original discharge signal is obtained, the original discharge signal is denoised using a first denoising operation. The original discharge signal after the first denoising operation is taken as the first denoising result. The first denoising result is then reconstructed using a reconstruction method corresponding to the first denoising operation to obtain the reconstructed discharge signal.

[0064] Step S103: Perform a second denoising operation on the PRPS map to obtain a second denoising result, and reconstruct the second denoising result to obtain a reconstructed PRPS map.

[0065] The second denoising operation is a method that combines data processing and thresholding for denoising.

[0066] In this step, when the PRPS map is determined, a second denoising operation is performed on the PRPS map to denoise it, and the PRPS map after the second denoising operation is used as the second denoising result. Then, the second denoising result is reconstructed using the reconstruction method corresponding to the second denoising operation to obtain the reconstructed PRPS map.

[0067] Specifically, since the original discharge signal and the PRPS spectrum have different data types—the original discharge signal is one-dimensional data while the PRPS spectrum is two-dimensional data—the denoising operations used are also different, namely, the specific implementations of the first denoising operation and the second denoising operation are different.

[0068] It is understandable that the first and second denoising operations are applicable to different scenarios or data. In actual use, the specific first and second denoising operations can be determined according to the data to be processed and the requirements.

[0069] Step S104: Obtain the accuracy of the first denoising operation and the second denoising operation respectively, and determine the first weight and the second weight respectively based on the accuracy of the first denoising operation and the second denoising operation.

[0070] The accuracy rate of the first denoising operation refers to its recognition accuracy rate in historical applications. The accuracy rate of the second denoising operation refers to its recognition accuracy rate in historical applications.

[0071] In this step, since the accuracy of the first denoising operation and the second denoising operation will be updated irregularly or periodically, it is necessary to obtain the accuracy corresponding to the first denoising operation and the accuracy corresponding to the second denoising operation, and then determine the first weight and the second weight based on the accuracy of the first denoising operation and the accuracy of the second denoising operation, respectively.

[0072] Specifically, the first weight can be determined by the proportion of the accuracy of the first denoising operation to the sum of the accuracy of the first denoising operation and the accuracy of the second denoising operation, and the second weight can be determined by the proportion of the accuracy of the second denoising operation to the sum of the accuracy of the first denoising operation and the accuracy of the second denoising operation.

[0073] It is understandable that the first weight is used as the fusion weight for reconstructing the discharge signal, and the second weight is used as the fusion weight for reconstructing the PRPS map. Determining the fusion weight based on the accuracy can take into account the accuracy of different denoising operations, thereby adjusting the weight of the data after denoising through the denoising operation in the linear weighted fusion, further improving the accuracy and data quality of the PRPS map, and thus improving the accuracy of subsequent recognition, diagnosis and other algorithms in application.

[0074] Step S105: Based on the first weight and the second weight, the reconstructed discharge signal and the reconstructed PRPS spectrum are linearly weighted and fused to obtain the target PRPS spectrum.

[0075] In this step, based on the first and second weights, the reconstructed discharge signal and the reconstructed PRPS map are linearly weighted and fused. This can take into account the contribution of different denoising operations to the subsequent recognition results. The denoised reconstructed discharge signal and the reconstructed PRPS map are linearly weighted and fused, thereby further reducing the noise of the final target PRPS map and improving the accuracy of subsequent recognition, diagnosis and other algorithms.

[0076] This application provides a PRPS image denoising method, apparatus, storage medium, and computer equipment. The method includes: acquiring an original discharge signal and determining a corresponding PRPS image based on the original discharge signal; performing a first denoising operation on the original discharge signal to obtain a first denoising result, and reconstructing the first denoising result to obtain a reconstructed discharge signal. A suitable denoising operation is used to denoise the original discharge signal, thereby reducing interference features in the original discharge signal. Next, a second denoising operation is performed on the PRPS image to obtain a second denoising result, and the second denoising result is reconstructed to obtain a reconstructed PRPS image; a denoising operation capable of denoising the PRPS image is used to denoise it, thereby obtaining a high-quality reconstructed PRPS image. Finally, based on the accuracy of the first denoising operation and the accuracy of the second denoising operation, the reconstructed discharge signal and the reconstructed PRPS image are linearly weighted and fused to obtain a target PRPS image. By determining the weights of reconstructing the discharge signal and reconstructing the PRPS spectrum in linear weighted fusion based on the accuracy of the first and second denoising operations, the contribution of different denoising operations to the subsequent identification results can be considered, thereby further reducing the noise in the final target PRPS spectrum, improving the quality of partial discharge data, and thus improving the accuracy of subsequent identification, diagnosis and other algorithms in application.

[0077] like Figure 2 As shown, in one embodiment, the step of performing a first denoising operation on the original discharge signal includes:

[0078] Step S201: Determine the corresponding wavelet basis function and scale parameter set based on the original discharge signal.

[0079] Wavelet basis functions are used to analyze the frequency and time characteristics of signals in wavelet transform. A wavelet basis function is an oscillating function with zero mean and rapid decay. The scaling parameter set includes multiple scaling parameters, which control the scaling or expansion of the wavelet basis function. Specifically, the scaling parameters affect the frequency characteristics and time-domain resolution of the wavelet basis function, determining its width and position in the frequency domain.

[0080] In this step, the wavelet basis function and scale parameter set that match the original discharge signal can be determined based on parameters such as the type, spectral distribution, and signal characteristics of the original discharge signal.

[0081] Understandably, different wavelet basis functions are suitable for different types of signal characteristics and analysis needs. The specific choice can be made based on the original discharge signal and actual requirements. In one example, the wavelet basis function can be the Daubechies (DB) wavelet function, with the vanishing moment order selected as 8th.

[0082] Step S202: Input each scale parameter of the scale parameter set and the original discharge signal into the wavelet basis function respectively to determine the wavelet coefficients of the original discharge signal under each scale parameter of the scale parameter set.

[0083] In this step, each scale parameter of the scale parameter set and the wavelet basis function of the original discharge signal input value are respectively used to match the original discharge signal with wavelet basis functions at various scales and locations, thereby analyzing the local features of the original discharge signal in frequency and time. Next, the original discharge signal is convolved with the wavelet basis function at each scale parameter to obtain the wavelet coefficients at each scale. Wavelet coefficients can be understood as the local energy or amplitude of the signal at the corresponding scale, reflecting the signal's variation characteristics in time and frequency. Smaller scale parameters correspond to higher frequency resolution, capable of capturing rapidly changing high-frequency components in the signal; while larger scale parameters correspond to lower frequency resolution, suitable for analyzing slowly changing low-frequency components or overall trends in the signal.

[0084] Step S203: Generate a wavelet coefficient matrix based on the wavelet coefficients corresponding to each scale parameter in the scale parameter set of the original discharge signal, and perform threshold processing on the wavelet coefficient matrix. The wavelet coefficient matrix after threshold processing is determined as the first denoising result.

[0085] In this step, when the wavelet coefficients corresponding to each scale parameter of the scale parameter set of the original discharge signal are obtained, these wavelet coefficients can be used to form a wavelet coefficient matrix. Then, the wavelet coefficient matrix can be thresholded by determining the threshold and selecting the threshold function, thereby reducing noise or insignificant signal components while retaining larger signal variations.

[0086] Understandably, signal denoising using wavelet transform and thresholding can effectively separate useful information from noise components in the original discharge signal. Wavelet transform decomposes the original discharge signal into frequency components at different scales, while thresholding removes noise components by setting appropriate thresholds to zero or adjust the low-energy (i.e., noise) parts of the wavelet coefficients. Compared to traditional denoising methods, this approach effectively preserves the local details and edge information of the original discharge signal while reducing global noise interference, resulting in a clearer and more accurate denoised and reconstructed original discharge signal.

[0087] like Figure 3 As shown, in one embodiment, the step of performing a second denoising operation on the PRPS map includes:

[0088] Step S301: Perform iterative decomposition on the PRPS map.

[0089] Step S302: In a decomposition iteration, determine the decomposition objects in this decomposition iteration.

[0090] In the first decomposition iteration, the decomposition object is the PRPS map. The decomposition object refers to the data to be decomposed.

[0091] Step S303: Filter the decomposed object to obtain the high-frequency coefficients and low-frequency coefficients of the decomposed object.

[0092] Among them, the high-frequency coefficients represent the low-frequency (approximate) components in the decomposed object, and the low-frequency coefficients represent the high-frequency (detail) components in the decomposed object.

[0093] In this step, the decomposed object is subjected to low-pass and high-pass filtering respectively to obtain the corresponding low-frequency and high-frequency coefficients. It can be understood that by combining high-pass and low-pass filtering, different frequency band information of the PRPS spectrum can be obtained. The low-frequency coefficients obtained through low-pass filtering reflect the overall trend and slowly changing parts of the PRPS spectrum, while the high-frequency coefficients obtained through high-pass filtering reflect the rapid changes and fine structure of the PRPS spectrum.

[0094] Step S304: Threshold processing is performed on the high-frequency coefficients to update them, and the low-frequency coefficients are determined as the decomposition objects in the next decomposition iteration until the number of decomposition iterations reaches the preset number of decomposition layers. Based on the high-frequency coefficients and low-frequency coefficients in each decomposition iteration, a second denoising result is generated.

[0095] Specifically, setting the preset decomposition layer number too low will result in insufficient noise removal, while setting it too high will cause some partial discharge signals to be removed along with the noise. Therefore, it is necessary to set an appropriate preset decomposition layer number based on experience or experiments. In one example, the preset decomposition layer number can be set to 4.

[0096] In this step, since most noise is typically high-frequency, only thresholding of the high-frequency coefficients is needed. By setting an appropriate threshold, the low-energy (i.e., noise) portion of the high-frequency coefficients is zeroed or adjusted, thereby removing noise components from the PRPS spectrum. Next, the thresholded high-frequency coefficients are used to update the high-frequency coefficients determined in step S303. Since low-frequency coefficients reflect overall trends and slowly changing components, each decomposition level only decomposes the low-frequency components. Therefore, the low-frequency coefficients are identified as the decomposition targets for the next decomposition iteration. In the next decomposition iteration, the low-frequency coefficients are further decomposed to obtain the relatively high-frequency and relatively low-frequency components within them, until the number of decomposition iterations reaches the preset number of decomposition levels. Based on the high-frequency and low-frequency coefficients determined in each decomposition iteration, a second denoising result is generated.

[0097] In one example, such as Figure 4 As shown, Figure 4 This is an example diagram illustrating a PRPS image denoising method provided in this application embodiment. The original discharge signal is denoised using CWT (Continuous Wavelet Transform), and the PRPS image is denoised using DWT (Discrete Wavelet Transform). During DWT denoising of the PRPS image, H and L represent the dimensions in the PRPS image, which can be decomposed into four components: LL1, LH1, HL1, and HH1. LL1 is then further decomposed to achieve a preset number of decomposition levels. Specifically, the LL component represents the low-frequency information of the image, depicting the approximation information of the original image; the LH component represents the high-frequency information in the horizontal direction of the image, depicting the horizontal details of the original image; the HL component represents the high-frequency information in the vertical direction of the image, depicting the vertical details of the original image; and the HH component represents the high-frequency information in the diagonal direction of the image, depicting the diagonal details of the original image.

[0098] Understandably, using wavelet transform of the PRPS spectrum combined with thresholding for signal denoising can effectively separate useful information from noise components in the PRPS spectrum. Thresholding removes noise components by setting appropriate thresholds to zero or adjust the low-energy (i.e., noise) parts of the wavelet coefficients. Compared to traditional denoising methods, this embodiment effectively preserves the local details and edge information of the PRPS spectrum while reducing global noise interference, resulting in a clearer and more accurate denoised and reconstructed PRPS spectrum.

[0099] In one embodiment, thresholding of high-frequency coefficients includes:

[0100] Determine the target threshold and threshold function for this decomposition iteration;

[0101] Based on the target threshold and threshold function in this decomposition iteration, the high-frequency coefficients are quantized, and the high-frequency coefficients in this decomposition iteration are updated based on the quantized high-frequency coefficients.

[0102] The threshold function is used to determine the specific quantization method based on the target threshold. Threshold functions include soft thresholding and hard thresholding. High-frequency coefficients include detail coefficients in the horizontal direction, vertical direction, and diagonal direction.

[0103] In this embodiment, in order to balance the characteristics of the hard threshold function and the soft threshold function, the threshold function can be determined based on the adjustment variable.

[0104] In one example, the target threshold is determined according to the following expression:

[0105]

[0106] wherein, represents the target threshold, cDi represents the high-frequency coefficient of the i-th decomposition level, and N represents the data length of the original discharge signal. In combination with the actual situation of threshold processing for partial discharge signals, N may be adjusted to one tenth of the data length in this embodiment.

[0107] Meanwhile, in order to balance the respective characteristics of the hard threshold function and the soft threshold function in the selection of the threshold function, an adjustment variable a is introduced, and the formula of the threshold function is as follows:

[0108]

[0109] wherein, represents the quantized high-frequency coefficient, represents the high-frequency coefficient, a is the adjustment variable, represents the target threshold, wherein when a=0, the function is a hard threshold function; when a=1, the function is a soft threshold function; when 0<a<1, the function can have partial characteristics of both the hard threshold function and the soft threshold function at the same time.

[0110] It can be understood that the threshold processing can help remove noise or unimportant information in the signal, so as to improve the quality and reliability of the data; moreover, the threshold processing can also highlight and strengthen the key features in the signal, so as to better support the subsequent analysis, identification and decision-making processes.

[0111] In one of the embodiments, the step of performing linear weighted fusion on the reconstructed discharge signal and the reconstructed PRPS map based on the first weight and the second weight includes:

[0112] performing normalization processing on the reconstructed PRPS map to obtain a first PRPS map;

[0113] taking the first weight as the fusion weight of the reconstructed discharge signal, and taking the second weight as the fusion weight of the reconstructed PRPS map, performing linear weighted fusion on the reconstructed discharge signal and the reconstructed PRPS map, to obtain a target PRPS map.

[0114] In this embodiment, considering the characteristic law of the partial discharge signal, 98% maximum value normalization may be adopted to obtain the first PRPS map, then the first weight is used as the fusion weight of the reconstructed discharge signal, and the second weight is used as the fusion weight of the reconstructed PRPS map, then linear weighted fusion is performed on the reconstructed discharge signal and the reconstructed PRPS map, and finally the target PRPS map is obtained.

[0115] Here, 98% means that the normalized data range can cover 98% of the value range in the original data.

[0116] In one example, the reconstructed discharge signal and the reconstructed PRPS spectrum can be weighted and fused according to the following expression:

[0117]

[0118] In the formula, f(CDWT) represents the target PRPS spectrum, a is the first weight, b is the second weight, f(CWT) represents the reconstructed discharge signal, and f(DWT) represents the reconstructed PRPS spectrum.

[0119] Understandably, by linearly weighting and fusing the reconstructed discharge signal and the reconstructed PRPS map, the target PRPS map is obtained, further optimizing the signal processing and feature extraction results. Furthermore, the accuracy of the first and second denoising operations in subsequent partial discharge signal pattern recognition is used as the fusion weight for the reconstructed discharge signal obtained after the first denoising operation and the reconstructed PRPS map obtained after the second denoising operation, respectively. This approach considers the contribution of different denoising operations to the subsequent recognition results. Finally, the results of the two denoising methods are linearly weighted and fused, thereby further reducing noise in the final target PRPS map.

[0120] In one embodiment, the step of determining the corresponding PRPS spectrum based on the original discharge signal includes:

[0121] The signal data in the original discharge signal are rearranged to obtain the corresponding two-dimensional array;

[0122] The two-dimensional array is filled with zero values ​​in the phase dimension, and the phase window corresponding to the zero-filled two-dimensional array is divided to obtain the target phase window;

[0123] Average pooling is used to filter the data within each target phase window, and PRPS maps are generated based on the filtered data.

[0124] The phase window refers to a time window used for signal analysis to extract the phase information of a signal in the frequency domain.

[0125] In this embodiment, the original discharge signal data is rearranged into a two-dimensional array to fill zeros in the phase dimension, ensuring data integrity and continuity. Next, the signal is divided into multiple target phase windows by partitioning the two-dimensional array, each containing signal data within a specific time period. Subsequently, average pooling is used to filter the data within each target phase window, thereby extracting key features and reducing noise. Finally, a PRPS map is generated based on the filtered data.

[0126] In one example, assuming the original discharge signal contains 50 cycles of partial discharge parameters within 1 second at power frequency voltage, the original discharge signal is rearranged into a 50×2000 two-dimensional array, i.e., 50 discharge cycles and 2000 phase windows. Zero-value padding is applied to the above two-dimensional array along the phase dimension, re-dividing the phase into 72 phase windows. Average pooling is used to filter data within each phase window, ultimately retaining 3600 (50×72) downsampled data points, which are then arranged and combined into a 50×72 PRPS (Partial Discharge Perimeter) map.

[0127] Understandably, firstly, rearranging and zero-filling ensure the integrity and continuity of the data, which helps to effectively handle missing or outlier signals. Secondly, using average pooling for data filtering reduces noise interference and highlights the main features within the target phase window, thereby improving the accuracy and reliability of subsequent PRPS map generation.

[0128] In one embodiment, the PRPS map denoising method further includes:

[0129] Obtain historical discrimination data for the first denoising operation and historical discrimination data for the second denoising operation;

[0130] The discrimination accuracy is calculated from the historical discrimination data of the first denoising operation, and the accuracy corresponding to the first denoising operation is updated based on the discrimination accuracy.

[0131] The discrimination accuracy is calculated from the historical discrimination data of the second denoising operation, and the accuracy corresponding to the second denoising operation is updated based on the discrimination accuracy.

[0132] Historical discrimination data includes recognition records and recognition results. The current discrimination accuracy can be determined based on the recognition results in the historical discrimination data.

[0133] In this embodiment, historical discrimination data for the first and second denoising operations can be acquired periodically or irregularly to statistically analyze the accuracy rates of these historical data, thereby updating the current accuracy rates of the first and second denoising operations. This allows for timely updates to the accuracy rates based on historical discrimination data, which in turn updates the weights in the linear weighted fusion, achieving closed-loop feedback and further improving the quality of the final target PRPS map. This enhances the accuracy of subsequent recognition, diagnosis, and other algorithm applications.

[0134] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0135] The PRPS map noise reduction device provided in the embodiments of this application is described below. The PRPS map noise reduction device described below can be referred to in correspondence with the PRPS map noise reduction method described above.

[0136] like Figure 5 As shown, this application provides a PRPS map noise reduction device 400, the device comprising:

[0137] The signal acquisition module 401 is used to acquire the original discharge signal and determine the corresponding PRPS spectrum based on the original discharge signal.

[0138] The first denoising module 402 is used to perform a first denoising operation on the original discharge signal to obtain a first denoising result, and to reconstruct the first denoising result to obtain a reconstructed discharge signal.

[0139] The second denoising module 403 is used to perform a second denoising operation on the PRPS map to obtain a second denoising result, and to reconstruct the second denoising result to obtain a reconstructed PRPS map.

[0140] The weight determination module 404 is used to obtain the accuracy of the first denoising operation and the second denoising operation respectively, and determine the first weight and the second weight respectively based on the accuracy of the first denoising operation and the second denoising operation.

[0141] The weighted fusion module 405 is used to linearly weight and fuse the reconstructed discharge signal and the reconstructed PRPS spectrum based on the first weight and the second weight to obtain the target PRPS spectrum.

[0142] In one embodiment, the first noise reduction module includes:

[0143] The information determination submodule is used to determine the corresponding wavelet basis function and scale parameter set based on the original discharge signal;

[0144] The coefficient determination submodule is used to input each ceramic parameter in the scale parameter set and the original discharge signal into the wavelet basis function to determine the wavelet coefficients of the original discharge signal under each scale parameter in the scale parameter set.

[0145] The matrix generation submodule is used to generate a wavelet coefficient matrix based on the wavelet coefficients corresponding to each scale parameter in the scale parameter set of the original discharge signal, and to perform thresholding on the wavelet coefficient matrix, and to determine the thresholded wavelet coefficient matrix as the first denoising result.

[0146] In one embodiment, the second noise reduction module includes:

[0147] The decomposition submodule is used for iterative decomposition of the PRPS map;

[0148] The object determination submodule is used to determine the decomposition objects in a single decomposition iteration; in the first decomposition iteration, the decomposition object is the PRPS map.

[0149] The filtering submodule is used to filter the decomposed object to obtain the high-frequency and low-frequency coefficients of the decomposed object.

[0150] The coefficient update submodule is used to perform threshold processing on high-frequency coefficients to update them, and to determine low-frequency coefficients as the decomposition objects in the next decomposition iteration, until the number of decomposition iterations reaches the preset number of decomposition layers. Based on the high-frequency and low-frequency coefficients in each decomposition iteration, a second denoising result is generated.

[0151] In one embodiment, the coefficient update submodule includes:

[0152] The threshold determination unit is used to determine the target threshold and threshold function in this decomposition iteration.

[0153] The quantization unit is used to quantize the high-frequency coefficients according to the target threshold and threshold function in the current decomposition iteration, and update the high-frequency coefficients in the current decomposition iteration according to the quantized high-frequency coefficients.

[0154] In one embodiment, the weighted fusion module includes:

[0155] The reconstructed PRPS map was normalized to obtain the first PRPS map.

[0156] The reconstructed discharge signal and the reconstructed PRPS spectrum are linearly weighted and fused using the first weight as the fusion weight and the second weight as the fusion weight, respectively, to obtain the target PRPS spectrum.

[0157] In one embodiment, the signal acquisition module includes:

[0158] The arrangement submodule is used to rearrange the signal data in the original discharge signal to obtain the corresponding two-dimensional array;

[0159] The partitioning submodule is used to fill the two-dimensional array with zero values ​​in the phase dimension, and to divide the phase window corresponding to the zero-filled two-dimensional array to obtain each target phase window;

[0160] The filtering submodule is used to filter the data within each target phase window using average pooling and generate PRPS maps based on the filtered data.

[0161] In one embodiment, the PRPS map noise reduction device further includes:

[0162] The data acquisition module is used to acquire historical discrimination data of the first denoising operation and historical discrimination data of the second denoising operation.

[0163] The accuracy statistics module is used to calculate the discrimination accuracy in the historical discrimination data of the first denoising operation, and update the accuracy corresponding to the first denoising operation based on the discrimination accuracy.

[0164] The accuracy update module is used to calculate the discrimination accuracy in the historical discrimination data of the second denoising operation, and update the accuracy corresponding to the second denoising operation based on the discrimination accuracy.

[0165] The division of modules in the above-described PRPS image noise reduction device is merely illustrative. In other embodiments, the PRPS image noise reduction device can be divided into different modules as needed to complete all or part of its functions. Each module in the above-described PRPS image noise reduction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0166] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the PRPS map denoising method as described in any of the above embodiments.

[0167] In one embodiment, this application also provides a computer device storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the PRPS map denoising method as described in any of the above embodiments.

[0168] Indicatively, such as Figure 6 As shown, Figure 6 This is a schematic diagram of the internal structure of a computer device 500 provided in an embodiment of this application. The computer device 500 can be provided as a server. (Refer to...) Figure 6 The computer device 500 includes a processing component 502, which further includes one or more processors, and memory resources represented by memory 501 for storing instructions, such as application programs, that can be executed by the processing component 502. The application programs stored in memory 501 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 502 is configured to execute instructions to perform the PRPS map denoising method of any of the above embodiments.

[0169] The computer device 500 may also include a power supply component 503 configured to perform power management of the computer device 500, a wired or wireless network interface 504 configured to connect the computer device 500 to a network, and an input / output (I / O) interface 505. The computer device 500 may operate on an operating system stored in memory 501, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0170] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0171] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a…" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this document, the singular forms "a," "an," and "the" may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms “comprising / including” or “having” specify the presence of the stated features, wholes, steps, operations, components, parts or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0172] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0173] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A PRPS image denoising method, characterized in that, The method includes: Acquire the original discharge signal and determine the corresponding PRPS spectrum based on the original discharge signal; The original discharge signal is subjected to a first denoising operation to obtain a first denoising result, and the first denoising result is reconstructed to obtain a reconstructed discharge signal; A second denoising operation is performed on the PRPS map to obtain a second denoising result, and the second denoising result is reconstructed to obtain a reconstructed PRPS map; The accuracy rates of the first denoising operation and the second denoising operation are obtained respectively, and the first weight and the second weight are determined based on the accuracy rates of the first denoising operation and the second denoising operation respectively. Based on the first weight and the second weight, the reconstructed discharge signal and the reconstructed PRPS spectrum are linearly weighted and fused to obtain the target PRPS spectrum; The step of performing a second denoising operation on the PRPS map includes: The PRPS map is iteratively decomposed; In a single decomposition iteration, the decomposition object in this decomposition iteration is determined; wherein, in the first decomposition iteration, the decomposition object is the PRPS map; The decomposed object is filtered to obtain its high-frequency coefficients and low-frequency coefficients. The high-frequency coefficients are thresholded to update them, and the low-frequency coefficients are determined as the decomposition objects in the next decomposition iteration. This process continues until the number of decomposition iterations reaches the preset number of decomposition layers. Based on the high-frequency and low-frequency coefficients in each decomposition iteration, a second denoising result is generated. The thresholding process for the high-frequency coefficients includes: Determine the target threshold and threshold function for this decomposition iteration; Based on the target threshold and threshold function in this decomposition iteration, the high-frequency coefficients are quantized, and the high-frequency coefficients in this decomposition iteration are updated based on the quantized high-frequency coefficients. The step of determining the corresponding PRPS spectrum based on the original discharge signal includes: The signal data in the original discharge signal are rearranged to obtain the corresponding two-dimensional array; The two-dimensional array is filled with zero values ​​in the phase dimension, and the phase window corresponding to the zero-filled two-dimensional array is divided to obtain each target phase window; Average pooling is used to filter the data within each target phase window, and PRPS maps are generated based on the filtered data. The method further includes: Obtain historical discrimination data of the first denoising operation and historical discrimination data of the second denoising operation; The discrimination accuracy is calculated in the historical discrimination data of the first denoising operation, and the accuracy corresponding to the first denoising operation is updated according to the discrimination accuracy. The discrimination accuracy is calculated from the historical discrimination data of the second denoising operation, and the accuracy corresponding to the second denoising operation is updated based on the discrimination accuracy.

2. The PRPS image denoising method according to claim 1, characterized in that, The step of performing a first denoising operation on the original discharge signal includes: Based on the original discharge signal, determine the corresponding wavelet basis function and scale parameter set; Each scale parameter of the scale parameter set and the original discharge signal are respectively input into the wavelet basis function to determine the wavelet coefficients corresponding to the original discharge signal under each scale parameter of the scale parameter set; Based on the wavelet coefficients corresponding to the original discharge signal at each scale parameter in the scale parameter set, a wavelet coefficient matrix is ​​generated, and the wavelet coefficient matrix is ​​thresholded. The thresholded wavelet coefficient matrix is ​​then determined as the first denoising result.

3. A PRPS image noise reduction device, characterized in that, The device includes: The signal acquisition module is used to acquire the original discharge signal and determine the corresponding PRPS spectrum based on the original discharge signal. The first denoising module is used to perform a first denoising operation on the original discharge signal to obtain a first denoising result, and to reconstruct the first denoising result to obtain a reconstructed discharge signal. The second denoising module is used to perform a second denoising operation on the PRPS map to obtain a second denoising result, and to reconstruct the second denoising result to obtain a reconstructed PRPS map. The weight determination module is used to obtain the accuracy of the first denoising operation and the second denoising operation respectively, and determine the first weight and the second weight respectively based on the accuracy of the first denoising operation and the second denoising operation. The weighted fusion module is used to linearly weight and fuse the reconstructed discharge signal with the reconstructed PRPS spectrum based on the first weight and the second weight to obtain the target PRPS spectrum. The second noise reduction module includes: The decomposition submodule is used for iterative decomposition of the PRPS map; The object determination submodule is used to determine the decomposition object in a decomposition iteration; wherein, in the first decomposition iteration, the decomposition object is the PRPS map; The filtering submodule is used to perform filtering processing on the decomposed object to obtain the high-frequency coefficients and low-frequency coefficients of the decomposed object; The coefficient update submodule is used to perform threshold processing on the high-frequency coefficients to update the high-frequency coefficients, and to determine the low-frequency coefficients as the decomposition objects in the next decomposition iteration, until the number of decomposition iterations reaches the preset number of decomposition layers, and to generate a second denoising result based on the high-frequency coefficients and low-frequency coefficients in each decomposition iteration. The coefficient update submodule includes: The threshold determination unit is used to determine the target threshold and threshold function in this decomposition iteration. The quantization unit is used to quantize the high-frequency coefficients according to the target threshold and threshold function in the current decomposition iteration, and update the high-frequency coefficients in the current decomposition iteration according to the quantized high-frequency coefficients. The signal acquisition module includes: The arrangement submodule is used to rearrange the signal data in the original discharge signal to obtain the corresponding two-dimensional array; The partitioning submodule is used to fill the two-dimensional array with zero values ​​in the phase dimension, and to partition the phase window corresponding to the two-dimensional array after zero-value filling to obtain each target phase window; The filtering submodule is used to filter the data within each target phase window using average pooling and generate PRPS maps based on the filtered data. The device further includes: The data acquisition module is used to acquire historical discrimination data of the first denoising operation and historical discrimination data of the second denoising operation; The accuracy statistics module is used to calculate the discrimination accuracy in the historical discrimination data of the first denoising operation, and update the accuracy corresponding to the first denoising operation based on the discrimination accuracy. The accuracy update module is used to calculate the discrimination accuracy in the historical discrimination data of the second denoising operation, and update the accuracy corresponding to the second denoising operation according to the discrimination accuracy.

4. A storage medium, characterized in that: The storage medium stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the PRPS map denoising method as described in any one of claims 1 or 2.

5. A computer device, characterized in that, include: One or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the PRPS map denoising method as described in any one of claims 1 or 2.

Images