Partial discharge signal denoising method and device and electronic equipment

Abstract

The application provides a partial discharge signal denoising method and device and electronic equipment. The method comprises the following steps: obtaining an initial partial discharge signal of a power device; extracting a frequency domain feature of the initial partial discharge signal; performing sparse decomposition on the initial partial discharge signal based on a pre-constructed sparse dictionary to obtain a sparse feature and a reconstruction signal of the initial partial discharge signal; inputting the frequency domain feature, the sparse feature and the reconstruction signal into a preset denoising signal generation model to obtain a denoised partial discharge signal output by the denoising signal generation model; and wherein the denoising signal generation model is obtained by training a generative adversarial network model based on frequency domain features, sparse features and reconstruction signals of a plurality of historical partial discharge signals and a plurality of real denoised partial discharge signals. The application can obtain a high-fidelity partial discharge detection signal in real time, accurately and reliably.

Description

Technical Field

[0001] This invention relates to the field of discharge detection technology, and in particular to a method, apparatus and electronic device for denoising partial discharge signals. Background Technology

[0002] Partial discharge is typically caused by localized defects inside or on the surface of the insulation material in electrical equipment. Its long-term presence leads to gradual deterioration of the insulation material, potentially causing equipment malfunctions or even shutdowns. Therefore, partial discharge detection is of great significance in monitoring the insulation condition of electrical equipment.

[0003] Ultrasonic testing technology is widely used for detecting partial discharge signals due to its advantages such as non-contact operation, high sensitivity, and accurate positioning. However, in practical applications, ultrasonic partial discharge signals are easily affected by periodic narrowband interference, white noise interference, and environmental noise, making it difficult to accurately obtain ultrasonic partial discharge signals.

[0004] In related technologies, signal denoising methods such as frequency domain filtering, time domain averaging, wavelet transform, and adaptive noise cancellation are commonly used to process the acquired partial discharge signals. These methods can suppress noise interference in the partial discharge signals to a certain extent. However, these signal denoising methods also have problems such as reduced signal fidelity, weak robustness to complex background noise, and insufficient real-time performance and computational efficiency. It is difficult to obtain high-fidelity partial discharge detection signals in real time, accurately, and reliably. Summary of the Invention

[0005] This invention provides a method, apparatus, and electronic device for denoising partial discharge signals, so as to obtain high-fidelity partial discharge detection signals in real time, accurately, and reliably.

[0006] In a first aspect, embodiments of the present invention provide a method for denoising partial discharge signals, comprising:

[0007] Acquire the initial partial discharge signal of the power equipment;

[0008] Extract the frequency domain features of the initial partial discharge signal;

[0009] Based on a pre-constructed sparse dictionary, the initial partial discharge signal is sparsely decomposed to obtain the sparse features and reconstructed signal of the initial partial discharge signal.

[0010] The frequency domain features, the sparse features, and the reconstructed signal are input into a preset denoising signal generation model to obtain a denoised partial discharge signal output by the denoising signal generation model; wherein, the denoising signal generation model is obtained by training a generative adversarial network model based on the frequency domain features, sparse features, and reconstructed signals of multiple historical partial discharge signals, as well as multiple real denoised partial discharge signals.

[0011] In one possible implementation, before performing sparse decomposition on the initial partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features and reconstructed signal of the initial partial discharge signal, the method further includes:

[0012] Acquire multiple historical partial discharge signals and an initial dictionary;

[0013] Select one historical partial discharge signal from the plurality of historical partial discharge signals and use it as the current signal, and use the initial dictionary as the current dictionary;

[0014] Based on the current dictionary, sparse coding is performed on the current signal to determine the sparse features of the current signal, as well as the dictionary matrix related to the current signal in the current dictionary.

[0015] The sparse dictionary is determined based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix.

[0016] In one possible implementation, determining the sparse dictionary based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix includes:

[0017] Based on the current signal, the current dictionary, and the sparse characteristics of the current signal, determine the residual for sparse representation of the current signal;

[0018] Based on the sparse features and residuals of the current signal, as well as the dictionary matrix, the updated dictionary is obtained;

[0019] Determine whether the change between the current dictionary and the updated dictionary is less than a preset threshold;

[0020] If the change between the current dictionary and the updated dictionary is greater than or equal to a preset threshold, the updated dictionary is used as the current dictionary. One historical partial discharge signal is selected from the remaining historical partial discharge signals as the current signal, and the steps of "sparsely encoding the current signal based on the current dictionary, determining the sparse features of the current signal, and the dictionary matrix related to the current signal in the current dictionary" and subsequent steps are repeated until the change between the current dictionary and the updated dictionary is less than the preset threshold, and the updated dictionary is used as the sparse dictionary.

[0021] In one possible implementation, based on the current dictionary, the current signal is sparsely encoded to determine the sparse features of the current signal, as well as the dictionary matrix related to the current signal in the current dictionary, including:

[0022] Set relevant parameters for sparse coding; wherein, the relevant parameters include the current sparse feature, the current residual, the preset residual threshold, the inactive set, the active set, the preset number of elements, the current iteration count, and the maximum iteration count;

[0023] Use the current signal as the current residual for initialization, use each dictionary column in the current dictionary as the element of the initial inactive set, and set the initial active set to an empty set;

[0024] Identify the element in the inactive set that has the strongest correlation with the current residual;

[0025] Add the element to the active set and remove the element from the inactive set to obtain the updated active set and the updated inactive set;

[0026] Based on the updated active set and the updated inactive set, determine the direction vector and step size for adjusting the sparse features;

[0027] Based on the current sparse features, the direction vector, and the step size, the sparse features are adjusted to obtain updated sparse features, and the updated sparse features are used as the current sparse features.

[0028] Based on the current sparse features and the current dictionary, update the residual and use the updated residual as the current residual;

[0029] Determine whether the current residual is less than the preset residual threshold, whether the current iteration count is less than the maximum iteration count, and whether the number of elements in the active set is less than the preset number of elements;

[0030] If the current residual is equal to or greater than the preset residual threshold, and the current iteration count is less than the maximum iteration count, and the number of elements in the active set is less than the preset number of elements, then update the current iteration count and re-execute the step of "determining the element in the inactive set that has the greatest correlation with the current residual" and subsequent steps until the current residual is less than the preset residual threshold, or the current iteration count is equal to or greater than the maximum iteration count, or the number of elements in the active set is equal to or greater than the preset number of elements. Use the current sparse feature as the sparse feature for determining the current signal, and determine the dictionary matrix in the current dictionary that is related to the current signal based on the elements in the active set.

[0031] In one possible implementation, before inputting the frequency domain features, the sparse features, and the reconstructed signal into a preset denoising signal generation model to obtain the denoised partial discharge signal output by the denoising signal generation model, the method further includes:

[0032] Acquire multiple historical partial discharge signals and multiple real denoised partial discharge signals;

[0033] The frequency domain features of each historical partial discharge signal are extracted, and based on the pre-built sparse dictionary, each historical partial discharge signal is sparsely decomposed to obtain the sparse features and reconstructed signal of each historical partial discharge signal.

[0034] The frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal are input into the generator of the generative adversarial network to obtain the noiseless generated signal output by the generator.

[0035] Multiple real denoised partial discharge signals and the noiseless generated signal are input into the discriminator of the generative adversarial network to obtain the probability that each noiseless generated signal and each real denoised partial discharge signal output by the discriminator is a real signal.

[0036] Based on the probability, the model parameters of the discriminator and the model parameters of the generator are adjusted to obtain a denoised signal generation model.

[0037] In one possible implementation, adjusting the model parameters of the discriminator and the generator according to the probability to obtain a denoised signal generation model includes:

[0038] Calculate the first loss function value of the discriminator based on the probability;

[0039] The second loss function value of the generator is calculated based on the probability, the noiseless generated signal, and the real denoised partial discharge signal.

[0040] Based on the first loss function value, the model parameters of the discriminator are adjusted, and based on the second loss function value, the model parameters of the generator are adjusted to obtain a trained generative adversarial network model. Based on the trained generative adversarial network model, a denoised signal generation model is obtained.

[0041] In one possible implementation, the second loss function corresponding to the second loss function value includes an adversarial loss function, a feature matching loss function, and a spectrum reconstruction loss function;

[0042] Based on the probability, the noiseless generated signal, and the actual denoised partial discharge signal, the second loss function value of the generator is calculated, including:

[0043] Calculate the adversarial loss function value of the generator based on the probability;

[0044] The feature matching loss function value of the generator is calculated based on the intermediate features of the noiseless generated signal in the discriminator and the intermediate features of the real denoised partial discharge signal in the discriminator.

[0045] Based on the spectrum diagram of the noiseless generated signal and the spectrum diagram of the real denoised partial discharge signal, calculate the spectrum reconstruction loss function value of the generator;

[0046] The second loss function value of the generator is calculated based on the adversarial loss function value, the feature matching loss function value, and the spectrum reconstruction loss function value.

[0047] In one possible implementation, after inputting the frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal into the generator of the generative adversarial network to obtain the noiseless generated signal output by the generator, the method further includes:

[0048] Based on a preset feature extraction network model, the first feature of the noiseless generated signal and the second feature of the real denoised partial discharge signal are extracted.

[0049] The perceptual loss function value of the generator is determined based on the first feature and the second feature;

[0050] After calculating the second loss function value of the generator based on the probability, the noiseless generated signal, and the real denoised partial discharge signal, the method further includes:

[0051] Update the second loss function value based on the perceived loss function value.

[0052] Secondly, embodiments of the present invention provide a noise reduction device for partial discharge signals, comprising:

[0053] The acquisition module is used to acquire the initial partial discharge signal of the power equipment;

[0054] An extraction module is used to extract the frequency domain features of the initial partial discharge signal;

[0055] The reconstruction module is used to perform sparse decomposition on the initial partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features of the initial partial discharge signal and the reconstructed signal.

[0056] The denoising module is used to input the frequency domain features, the sparse features, and the reconstructed signal into a preset denoising signal generation model to obtain a denoised partial discharge signal output by the denoising signal generation model; wherein, the denoising signal generation model is obtained by training a generative adversarial network model based on the frequency domain features, sparse features, and reconstructed signals of multiple historical partial discharge signals, as well as multiple real denoised partial discharge signals.

[0057] Thirdly, embodiments of the present invention provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described in the first aspect or any possible implementation thereof.

[0058] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described in the first aspect or any possible implementation thereof.

[0059] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0060] This invention employs a pre-constructed sparse dictionary to sparsely decompose the initial partial discharge signal, obtaining its sparse features and reconstructed signal. This allows for the extraction of key features from the initial partial discharge signal, dimensionality reduction, and improved denoising performance. Furthermore, by inputting the extracted frequency domain features, sparse features, and reconstructed signal of the initial partial discharge signal into a pre-defined denoising signal generation model, the model outputs a denoised partial discharge signal. This enables the model to focus on key components of the initial partial discharge signal during processing, effectively restoring a high-fidelity signal and improving the generation effect of the denoised signal, thus achieving efficient and accurate denoising. Moreover, by determining the frequency domain features, sparse features, and reconstructed signal of the initial partial discharge signal and inputting them into the denoising signal generation model, this invention removes redundant information from the signal, reduces unnecessary computational complexity, and facilitates convenient and efficient signal generation, thereby improving the speed and efficiency of generating denoised partial discharge signals. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 This is a flowchart illustrating the implementation of the partial discharge signal denoising method provided in this embodiment of the invention.

[0063] Figure 2 This is a flowchart illustrating the implementation of constructing a sparse dictionary according to an embodiment of the present invention;

[0064] Figure 3 This is a schematic diagram of the structure of a generative adversarial network provided in an embodiment of the present invention;

[0065] Figure 4 This is a schematic diagram of the structure of the noise reduction device for partial discharge signals provided in an embodiment of the present invention;

[0066] Figure 5This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0067] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0068] The inventors have discovered that common denoising methods for acquiring partial discharge signals, such as frequency domain filtering, time domain averaging, wavelet transform, and adaptive noise cancellation, reduce signal fidelity, lack robustness to complex background noise, and suffer from insufficient real-time performance and computational efficiency. Therefore, it is necessary to consider a method that can obtain high-fidelity partial discharge detection signals accurately and reliably in real time.

[0069] Based on the idea of ​​obtaining high-fidelity partial discharge detection signals in real time, accurately and reliably, in the embodiments of the present invention, the sparse features and reconstructed signals of the partial discharge signal are determined by a sparse dictionary, which can extract important features of the partial discharge signal and perform preliminary noise reduction on the signal; then the frequency domain features, sparse features and reconstructed signals of the initial partial discharge signal are input into the denoising signal generation model, which can effectively restore the high-fidelity partial discharge signal and perform denoising on the signal in real time, efficiently and accurately.

[0070] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0071] Figure 1 The implementation flowchart of the partial discharge signal denoising method provided in the embodiments of the present invention is described in detail below:

[0072] Step S101: Obtain the initial partial discharge signal of the power equipment.

[0073] In this embodiment, the initial partial discharge signal can be a partial discharge signal acquired from the power equipment that has experienced partial discharge.

[0074] Step S102: Extract the frequency domain features of the initial partial discharge signal.

[0075] In this embodiment, the acquired initial partial discharge signal is a time-domain signal. The initial partial discharge signal can be transformed into a frequency domain signal by frequency domain transformation, thereby extracting the frequency domain features of the initial partial discharge signal.

[0076] Here, the time-domain signal can be converted to the frequency domain using short-time Fourier transform, or frequency domain transformation can be performed using wavelet transform or other methods.

[0077] For example, through the expression: A frequency domain transformation is performed, where X(f,t) represents the frequency domain representation of the initial partial discharge signal, ω(nt) represents the window function, x(n) represents the original signal, which is a discrete signal that varies with time n, f represents the frequency variable, the frequency component of the Fourier transform, n represents the discrete time index, traversing all points of the signal x(n), t represents the time variable, and represents the position of the center of the window function.

[0078] Step S103: Based on the pre-constructed sparse dictionary, the initial partial discharge signal is sparsely decomposed to obtain the sparse features and reconstructed signal of the initial partial discharge signal.

[0079] In this embodiment, the initial partial discharge signal is sparsely decomposed using a sparse dictionary to obtain sparse features. The signal can then be reconstructed using the sparse features and the sparse dictionary to obtain the reconstructed signal of the initial partial discharge signal.

[0080] Among them, the sparse dictionary is a frequency domain dictionary constructed using the frequency domain characteristics of historical partial discharge signals.

[0081] Optionally, this embodiment can be based on the expression: Determine the sparsity features. Based on the expression: Determine the reconstructed signal. Where α i x represents the sparse features corresponding to the initial partial discharge signal. i Let represent the initial partial discharge signal, D represent the sparse dictionary, and α represent the sparse representation. Indicates the reconstructed signal. This indicates the reconstruction error.

[0082] Step S104: Input the frequency domain features, sparse features, and reconstructed signal into the preset denoising signal generation model to obtain the denoised partial discharge signal output by the denoising signal generation model; wherein, the denoising signal generation model is obtained by training the generative adversarial network model based on the frequency domain features, sparse features, and reconstructed signal of multiple historical partial discharge signals, as well as multiple real denoised partial discharge signals.

[0083] In this embodiment, the preset denoising signal generation model can be a Generative Adversarial Network (GAN) or a Generative Adversarial Network for Efficient and High Fidelity Speech Synthesis (HiFi-GAN).

[0084] This invention employs a pre-constructed sparse dictionary to sparsely decompose the initial partial discharge signal, obtaining its sparse features and reconstructed signal. This allows for the extraction of key features from the initial partial discharge signal, dimensionality reduction, and improved denoising performance. By inputting the extracted frequency domain features, sparse features, and reconstructed signal of the initial partial discharge signal into a pre-defined denoising signal generation model, the model outputs a denoised partial discharge signal. This allows the model to focus on key components of the initial partial discharge signal during processing, effectively restoring a high-fidelity signal and improving the generation effect of the denoised signal, thus achieving efficient and accurate denoising. Furthermore, by determining the frequency domain features, sparse features, and reconstructed signal of the initial partial discharge signal and inputting them into the denoising signal generation model, redundant information in the signal can be removed, reducing unnecessary computational complexity and enabling convenient and efficient signal generation, thereby improving the speed and efficiency of generating denoised partial discharge signals.

[0085] In some embodiments, before sparsely decomposing the initial partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features and reconstructed signal of the initial partial discharge signal, multiple historical partial discharge signals and an initial dictionary may be acquired first; one historical partial discharge signal may be selected from the multiple historical partial discharge signals and used as the current signal, and the initial dictionary may be used as the current dictionary; then, based on the current dictionary, sparse coding may be performed on the current signal to determine the sparse features of the current signal, as well as the dictionary matrix related to the current signal in the current dictionary; finally, the sparse dictionary may be determined based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix.

[0086] In this embodiment, the dictionary is updated using historical partial discharge signals. When updating the dictionary, the frequency domain characteristics and instantaneous characteristics of the historical partial discharge signals are considered, and elements that can capture these characteristics are selected preferentially.

[0087] By using the given current dictionary and solving for the sparse features corresponding to the selected historical partial discharge ultrasonic signal, the historical partial discharge signal can be represented by the current dictionary. Then, the element most relevant to the historical partial discharge signal, i.e., the dictionary matrix, can be found from the current dictionary, thus determining the sparse dictionary.

[0088] Here, the initial dictionary can be constructed by introducing prior knowledge, such as the frequency domain characteristics and transient characteristics of the partial discharge signal, to ensure that the initial dictionary can effectively represent the main components of the signal.

[0089] The initial dictionary can be represented as: D0 = [d1, d2, ..., d j ,...,d K ], where D0 represents the initial dictionary, d j Let K represent the j-th element in the initial dictionary, and K represent the total number of elements in the initial dictionary.

[0090] Optionally, in this embodiment, the sparse dictionary is determined based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix. This can be achieved by: first, determining the residual for sparse representation of the current signal based on the current signal, the current dictionary, and the sparse features of the current signal; then, obtaining an updated dictionary based on the sparse features of the current signal, the residual, and the dictionary matrix; next, determining whether the change between the current dictionary and the updated dictionary is less than a preset threshold; if the change between the current dictionary and the updated dictionary is greater than or equal to the preset threshold, then the updated dictionary is used as the current dictionary, and one historical partial discharge signal is selected from the remaining historical partial discharge signals as the current signal. The steps of "sparsely encoding the current signal based on the current dictionary, determining the sparse features of the current signal, and the dictionary matrix related to the current signal in the current dictionary" and subsequent steps are repeated until the change between the current dictionary and the updated dictionary is less than the preset threshold, and the updated dictionary is used as the sparse dictionary.

[0091] In this embodiment, the residual of the sparse representation of the current signal, i.e., the difference between the current signal and the sparse representation of the current signal, can indicate the error in reconstructing the current signal using the current dictionary. By updating the dictionary matrix using the sparse features and residual of the current signal, the transient features and frequency domain characteristics of the signal can be taken into account, and elements that can capture these features can be preferentially selected for dictionary updates.

[0092] If the change between the current dictionary and the updated dictionary is greater than or equal to a preset threshold, it means that there is still room for updating the current dictionary, and other historical partial discharge signals can be used to continue updating the current dictionary; if the change between the current dictionary and the updated dictionary is less than the preset threshold, it means that the elements in the current dictionary can fully represent the historical partial discharge signals, and no further updates are needed, and the updated dictionary at this time is used as a sparse dictionary.

[0093] In addition, the current dictionary can be updated using a preset number of historical partial discharge signals.

[0094] In some embodiments, see Figure 2 The flowchart shown illustrates the implementation of constructing a sparse dictionary. The process of constructing a sparse dictionary can be described in detail as follows:

[0095] Step S201: Obtain multiple historical partial discharge signals and an initial dictionary.

[0096] Here, the historical partial discharge signal can be represented as x i Where represents the i-th element. The initial dictionary can be represented as D0 = [d1, d2, ..., dn]. j ,...,d K ].

[0097] Step S202: Select one historical partial discharge signal from multiple historical partial discharge signals and use it as the current signal, and use the initial dictionary as the current dictionary.

[0098] Step S203: Based on the current dictionary, perform sparse coding on the current signal to determine the sparse features of the current signal and the dictionary matrix related to the current signal in the current dictionary.

[0099] In this embodiment, sparse coding is performed on the current signal, that is, the current signal is represented using the current dictionary, thereby determining the sparse characteristics of the current signal. During the sparse coding process, by using the elements in the current dictionary used to represent the current signal, the dictionary matrix related to the current signal in the current dictionary can be determined.

[0100] Optionally, it can be done via an expression: Solve for the sparse characteristics of the current signal. Where α i x represents the sparse features corresponding to the initial partial discharge signal. i The current signal is represented by the selected historical partial discharge signal, D represents the current dictionary, and α represents the sparse representation. The reconstruction error is represented by ||α||1, which is the difference between the current signal and the signal Dα reconstructed using the current dictionary. It can be calculated using Euclidean distance. ||α||1 represents the sparsity regularization term, which is used to encourage sparsity. λ represents the regularization parameter, which is used to balance reconstruction error and sparsity.

[0101] Step S204: Determine the residual for sparse representation of the current signal based on the current signal, the current dictionary, and the sparse characteristics of the current signal.

[0102] Step S205: Based on the sparse features and residuals of the current signal, as well as the dictionary matrix, the updated dictionary is obtained.

[0103] In this embodiment, the sparse features and residuals can be updated using the dictionary matrix, and then the updated dictionary can be determined using the sparse features and parameters.

[0104] Optionally, it can be done via an expression: Update the current dictionary; where D t+1 D represents the updated dictionary, i.e., the current dictionary in the (t+1)th iteration. t Let η represent the current dictionary, i.e., the dictionary in the t-th iteration, and let α represent the learning rate. t This represents the sparse features corresponding to the current signal, that is, the sparse features of the current signal sparsely encoded using the current dictionary in the t-th iteration, r. t Let T represent the residual corresponding to the current signal, that is, the residual after reconstructing the current signal in the t-th iteration, and let T represent the transpose of the matrix.

[0105] Step S206: Determine whether the change between the current dictionary and the updated dictionary is less than a preset threshold.

[0106] Step S207: If the change between the current dictionary and the updated dictionary is greater than or equal to a preset threshold, then the updated dictionary is used as the current dictionary. One historical partial discharge signal is selected from the remaining historical partial discharge signals as the current signal. The steps of "sparsely encoding the current signal based on the current dictionary, determining the sparse features of the current signal, and the dictionary matrix related to the current signal in the current dictionary" and subsequent steps are re-executed until the change between the current dictionary and the updated dictionary is less than the preset threshold. Then the updated dictionary is used as the sparse dictionary.

[0107] Optionally, this embodiment performs sparse coding on the current signal based on the current dictionary to determine the sparse features of the current signal, as well as the dictionary matrix related to the current signal in the current dictionary. This can be:

[0108] Step 1: Set the relevant parameters for sparse coding; these parameters include the current sparse feature, current residual, preset residual threshold, inactive set, active set, preset number of elements, current iteration count, and maximum iteration count.

[0109] Step 2: Use the current signal as the current residual for initialization, use each dictionary column in the current dictionary as the element of the initial inactive set, and set the initial active set to an empty set.

[0110] Here, the current residual r0 = x i Sparse feature α0=0, activity set The inactive set Γ = {1, 2, ..., K}. Where x... i Indicates the current signal.

[0111] Step 3: Identify the element in the inactive set that has the strongest correlation with the current residual.

[0112] In this embodiment, the dictionary column with the highest correlation to the current residual is determined, which is the element with the highest correlation to the current residual. Here, the correlation coefficient c = D can be used. T r, perform correlation calculation.

[0113] Correspondingly, the expression for determining the element in the inactive set that has the highest correlation with the current residual can be: j = argmax k∈Γ |c k |, where j represents the element in the inactive set that has the greatest correlation with the current residual, k represents an element in the inactive set, and c k This indicates the correlation between the k-th element in the inactive set and the current residual.

[0114] Step 4: Add the element to the active set and remove the element from the inactive set to obtain the updated active set and the updated inactive set.

[0115] Step 5: Based on the updated active set and the updated inactive set, determine the direction vector and step size for adjusting the sparse features.

[0116] Here, we can use the expression: Calculate the direction vector; according to the expression: Calculate along direction The step size.

[0117] In the formula, u A D represents the direction vector, indicating the direction in which the sparse feature values ​​should be adjusted at each step. A This represents the dictionary column corresponding to the activity set Λ, 1 ALet c represent a vector of all ones with the same dimension as the number of elements in the activity set, used to calculate the update direction by averaging. γ represents the step size. j c represents the correlation of the element with the highest correlation to the current residual, determined from the inactive set in this iteration. In other words, it represents the correlation between the elements added to the active set in this iteration and the current residual. k This indicates the correlation between each element in the inactive set and the current residual. This indicates the direction of the adjustment step size.

[0118] Step 6: Based on the current sparse features, direction vector, and step size, adjust the sparse features to obtain the updated sparse features, and use the updated sparse features as the current sparse features.

[0119] In this embodiment, it can be based on the expression: α A =α A +γu A This updates the sparse features corresponding to the activity set, thus obtaining the current sparse features. Where α A This represents the sparse features corresponding to the activity set.

[0120] Step 7: Based on the current sparse features and the current dictionary, update the residuals and use the updated residuals as the current residuals.

[0121] In this embodiment, it can be based on the expression: r p+1 =x i -Dα p+1 Update the residual, where r p+1 The α represents the residual determined after the update, i.e., the current residual; D represents the current dictionary; α p+1 This represents the coefficient feature determined after the update, i.e., the current sparse feature, and p represents the current iteration number.

[0122] Step 8: Determine whether the current residual is less than the preset residual threshold, whether the current iteration count is less than the maximum iteration count, and whether the number of elements in the active set is less than the preset number of elements.

[0123] In this embodiment, the termination condition can be determined by the current residual, the current iteration number, and the number of elements in the active set.

[0124] Step 9: If the current residual is equal to or greater than the preset residual threshold, and the current iteration count is less than the maximum iteration count, and the number of elements in the active set is less than the preset number of elements, then update the current iteration count and re-execute the step of "determining the element in the inactive set that has the greatest correlation with the current residual" and subsequent steps until the current residual is less than the preset residual threshold, or the current iteration count is equal to or greater than the maximum iteration count, or the number of elements in the active set is equal to or greater than the preset number of elements. Use the current sparse feature as the sparse feature for determining the current signal, and determine the dictionary matrix in the current dictionary that is related to the current signal based on the elements in the active set.

[0125] In this embodiment, if the current residual is equal to or greater than a preset residual threshold, and the current iteration number is less than the maximum iteration number, and the number of elements in the active set is less than a preset number of elements, it indicates that sparse encoding of the current signal has not yet been achieved, and iteration can continue to update the sparse features. Here, updating the current iteration number can be done by incrementing the iteration number by 1.

[0126] If the current residual is less than the preset residual threshold, or the current iteration count is equal to or greater than the maximum iteration count, or the number of elements in the active set is equal to or greater than the preset number of elements, it indicates that the sparse decomposition of the current signal has been basically achieved, and the currently determined sparse features can be used to represent the current signal. Here, the elements in the active set are all elements related to the current signal; therefore, it can be determined as a dictionary matrix related to the current signal.

[0127] The previous section introduced the process of constructing a sparse dictionary. The following section will introduce the process of constructing a denoised signal generation model.

[0128] In some embodiments, see Figure 3 The diagram shows the structure of a Generative Adversarial Network (GAN), which includes a generator and a discriminator. The generator (G) uses the input data to generate high-fidelity, noise-free signals, improving signal quality and consistency through residual connections, skip connections, and self-attention mechanisms. The discriminator distinguishes the generated signal from the real signal, helping the generator improve the realism of the generated signal.

[0129] In this embodiment, before inputting the frequency domain features, sparse features, and reconstructed signal into a preset denoising signal generation model to obtain the denoised partial discharge signal output by the denoising signal generation model, multiple historical partial discharge signals and multiple real denoised partial discharge signals can be acquired first. The frequency domain features of each historical partial discharge signal are extracted, and based on a pre-built sparse dictionary, each historical partial discharge signal is sparsely decomposed to obtain the sparse features and reconstructed signal of each historical partial discharge signal. Then, the frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal are input into the generator of a generative adversarial network to obtain the noiseless generated signal output by the generator. Afterward, multiple real denoised partial discharge signals and noiseless generated signals are input into the discriminator of the generative adversarial network to obtain the probability that each noiseless generated signal and each real denoised partial discharge signal output by the discriminator is a real signal. Finally, based on the probability, the model parameters of the discriminator and the model parameters of the generator are adjusted to obtain the denoised signal generation model.

[0130] In this embodiment, the frequency domain features, sparse features, and reconstructed signals of historical partial discharge signals, as well as real denoised partial discharge signals, are used to train the generative adversarial network model. This enables the generator to capture details in the signal and maintain its high fidelity, thereby generating a high-fidelity noiseless generated signal. By using a discriminator to focus on the features in the signal, the noiseless generated signal and the real denoised partial discharge signal are identified, thus improving the quality of the generated signal.

[0131] The historical partial discharge signals here can be the same as or different from the historical partial discharge signals in the above embodiments. Furthermore, the number of actual denoised partial discharge signals can be the same as the number of noise-free generated signals.

[0132] Optionally, in this embodiment, the model parameters of the discriminator and the generator are adjusted according to probability to obtain a denoised signal generation model. This can be achieved by first calculating the first loss function value of the discriminator based on probability; then calculating the second loss function value of the generator based on probability, the noiseless generated signal, and the real denoised partial discharge signal; finally, adjusting the model parameters of the discriminator based on the first loss function value and the model parameters of the generator based on the second loss function value to obtain a trained generative adversarial network model, and obtaining the denoised signal generation model based on the trained generative adversarial network model.

[0133] In this embodiment, the first loss function value of the discriminator can be calculated first, the generator can be fixed, and the discriminator can be trained. Then, the second loss function value of the generator can be calculated, the discriminator can be fixed, and the generator can be trained. In this way, the discriminator and the generator can be repeatedly trained adversarially to obtain a trained generative adversarial network model.

[0134] Here, the denoising signal generation model is used to obtain the denoised generated signal, so the trained generator can be directly used as the denoising signal generation model.

[0135] Alternatively, the generative adversarial network can also adopt HiFi-GAN, with improvements made to the generator. The generator can include an input layer, a multi-scale convolutional layer, multiple residual blocks, and an output layer, and also incorporates a self-attention mechanism and skip connections.

[0136] In the generator, the input layer receives the frequency domain features, sparse features, and reconstructed signal from the input partial discharge signal. Multi-scale convolutional layers use kernels of different sizes to extract multi-scale features, capturing different frequency components and temporal features of the reconstructed signal. Each residual block contains two convolutional layers, each followed by batch normalization and an activation function (such as ReLU). The output layer generates a high-fidelity, noise-free signal.

[0137] Residual connections can alleviate the vanishing gradient problem by directly adding the input to the output through skip connections. Furthermore, introducing self-attention mechanisms can capture long-range dependencies, enhancing the generator's ability to model complex signal patterns.

[0138] The discriminator in a generative adversarial network can be a multi-scale discriminator. Multiple discriminators can be used independently, or a single discriminator can contain both time-domain and frequency-domain discriminators.

[0139] Here, a multi-scale discriminator can be designed with multiple independent discriminators, each focusing on signal features at different scales. For example, three discriminators can be designed to process the original signal and downsampled signals at different scales, respectively. A frequency domain discriminator can also be added to perform STFT or wavelet transform on the input signal before discrimination, enhancing the detection capability of time-domain and frequency-domain features.

[0140] A multi-scale discriminator can be represented as: D m (x)=[D1(x),D2(x),...,D N [(x)], where D i Let D be the discriminator at the i-th scale. If the n-th discriminator is a frequency domain discriminator, then its formula can be expressed as: D n (X(f,t))=Discriminator(X(f,t)).

[0141] In the discriminator, each discriminator consists of multiple convolutional layers and fully connected layers. Different features of the signal are extracted through convolutional operations, and then classification is performed through fully connected layers.

[0142] In this embodiment, the multi-scale convolutional layers of the generator can decompose the reconstructed signal into multiple time-scale features, obtaining decomposed features. By utilizing the sparse features, frequency domain features, and decomposed features of the reconstructed signal, the transient features, spectral characteristics, and non-stationary characteristics of the partial discharge signal can be captured. Simultaneously, the generator structure is improved: residual networks and skip connections are used in the generator to enhance the model's ability to capture signal details and maintain high signal fidelity; a self-attention mechanism is introduced to enhance the model's ability to model long-distance dependencies and improve the understanding of complex patterns in the partial discharge signal. Considering the characteristics of the partial discharge signal, the discriminator structure is improved: multiple discriminators are designed, each focusing on signal features at different scales, enhancing the discriminator's ability to detect various noises. In addition to the time-domain discriminator, a frequency-domain discriminator is introduced specifically for detecting the authenticity of frequency domain features, further improving the denoising effect.

[0143] Optionally, the second loss function corresponding to the second loss function value includes an adversarial loss function, a feature matching loss function, and a spectrum reconstruction loss function.

[0144] In this embodiment, the second loss function value of the generator is calculated based on probability, the noiseless generated signal, and the real denoised partial discharge signal. This can be achieved by first calculating the adversarial loss function value of the generator based on probability; then calculating the feature matching loss function value of the generator based on the intermediate features of the noiseless generated signal and the real denoised partial discharge signal in the discriminator; finally calculating the spectrum reconstruction loss function value of the generator based on the spectrum of the noiseless generated signal and the real denoised partial discharge signal; and finally calculating the second loss function value of the generator based on the adversarial loss function value, the feature matching loss function value, and the spectrum reconstruction loss function value.

[0145] In this embodiment, the adversarial loss function, the feature matching loss function, and the spectrum reconstruction loss function can be used to form a weighted loss function.

[0146] The expression for the weighted loss function is:

[0147]

[0148] In the formula, L G L represents the weighted loss function. adv (G;D k Let λ represent the adversarial loss function when training the generator with the discriminator fixed in the k-th training iteration. fm L represents the weight hyperparameters before feature matching loss, which control the importance of feature matching loss in the total loss. fm (G;D k ) represents the feature matching loss in the k-th training iteration, λ melL represents the weighting hyperparameters before Mel spectral loss, which controls the importance of Mel spectral loss in the total loss. mel (G) represents the spectral reconstruction loss function in the k-th training iteration, k represents the number of times the discriminator and generator are trained alternately, and K represents the total number of times the discriminator and generator are trained alternately.

[0149] Wherein, with the discriminator fixed, the adversarial loss function during generator training can be expressed as:

[0150] L adv (G;D)=Ex n [(D(G(x n ))-1) 2 ];

[0151] In the formula, L adv (G;D) represents the adversarial loss function when training the generator with a fixed discriminator, Ex n This indicates that for all data samples x n The expected value of the sample data, which is the mean value of the sample data, D(G(x) n )) represents the discriminator's response to the data G(x) generated by the generator. n The output probability of ).

[0152] Furthermore, with the generator fixed, the adversarial loss function during discriminant training, i.e., the first loss function, can be expressed as:

[0153] L adv (D;G)=E(x r ,x n )[(D(x r )-1) 2 +(D(G(x n ))) 2 ];

[0154] In the formula, L adv (D; G) denotes the adversarial loss function E(x) when training the generator with a fixed discriminator. r ,x n ) represents all real data samples x r and noisy data sample x n The expected value of the sample data, i.e., the mean value of the sample data, D(x r ) indicates that the discriminator evaluates the real data sample x. r The output probability, ideally, should be close to 1, because the discriminator should identify true data as real, D(G(x)). n )) indicates that the discriminator evaluates the generated data sample G(x) n The output probability of the data should ideally be close to 0, because the discriminator should identify the generated data as fake.

[0155] The feature matching loss function is a similarity measure learned by measuring the feature differences between the discriminator and the generated samples. Each intermediate feature of the discriminator is extracted, and the L1 distance between the real sample and the conditionally generated sample in each feature space is calculated.

[0156] The expression for the feature matching loss function is:

[0157]

[0158] In the formula, L fm (G;D) represents the feature matching loss function, where D... i (x r D represents the feature of the real sample in the i-th layer discriminator. i (G(x n )) represents the feature of the generated sample in the i-th layer of the discriminator, where i represents the i-th layer in the discriminator, T represents the total number of layers in the discriminator, and N i This represents the number of features in the discriminator of the i-th layer.

[0159] The Mel spectrum loss function is the loss calculated by reconstructing the generated signal back into the Mel spectrum.

[0160] The expression for the Mel spectrum loss function is: L mel (G)=||Mel(x r )-Mel(G(x n ))||1.

[0161] In the formula, L mel (G) represents the Mel spectrum loss function, Mel(x) r Mel(G(x)) represents the Mel spectrum corresponding to the real sample. n )) indicates the generated sample's corresponding Mel spectrum.

[0162] In some embodiments, after inputting the frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal into the generator of the generative adversarial network to obtain the noiseless generated signal output by the generator, the first feature of the noiseless generated signal and the second feature of the real denoised partial discharge signal can be extracted based on a preset feature extraction network model; and then the perceptual loss function value of the generator can be determined according to the first feature and the second feature.

[0163] Accordingly, in this embodiment, after calculating the second loss function value of the generator based on the probability, the noiseless generated signal, and the real denoised partial discharge signal, the second loss function value can also be updated based on the perception loss function value.

[0164] In this embodiment, the generator's loss function also includes a perceptual loss function. The perceptual loss function can improve the subjective quality of the generated samples. The difference between the generated samples and the real samples in the feature space is calculated by a pre-trained feature extraction network to ensure that the generated samples are perceptually similar to the real samples.

[0165] The expression for the perceptual loss function is: In the formula, L p Let φ represent the perceptual loss function. i (x) represents the features extracted from the real sample by the i-th layer of the feature extraction network model, φ i (G(z)) represents the feature extracted by the i-th layer of the feature extraction network model from the generated sample G(z).

[0166] Here, optimization is achieved through a loss function. By introducing a weighted loss function combined with perceptual loss, and using a pre-trained feature extraction network (VGG) to calculate the perceptual difference between the generated signal and the real signal, the subjective quality of the generated signal can be improved.

[0167] In some embodiments, before extracting the frequency domain features of the initial partial discharge signal, the initial partial discharge signal may also be subjected to DC component removal and normalization processing.

[0168] Correspondingly, after inputting the frequency domain features, sparse features, and reconstructed signal into the preset denoising signal generation model to obtain the denoised partial discharge signal output by the denoising signal generation model, the denoised partial discharge signal can also be subjected to inverse normalization processing.

[0169] Correspondingly, when constructing the sparse dictionary and the denoised signal generation model, the DC component can be removed and the normalization process can be performed on the historical partial discharge signal and the real denoised partial discharge signal.

[0170] Here, we can use an expression: Perform DC component removal processing, where x d (t) represents the signal after removing the DC component, x(t) represents the original signal, x(i) represents the signal, N represents the total number of signal samples, and i represents the i-th signal in the signal.

[0171] It can also be done through expressions: Normalization is performed, where x n (t) represents the normalized signal, x d (t) represents the signal after removing the DC component, max(|x d (t)|) represents the maximum absolute value of the signal over the entire time range, and is used to represent the maximum amplitude of the signal within a certain time period.

[0172] This invention employs a pre-constructed sparse dictionary to sparsely decompose the initial partial discharge signal, obtaining its sparse features and reconstructed signal. This allows for the extraction of key features from the initial partial discharge signal, dimensionality reduction, and improved denoising performance. By inputting the extracted frequency domain features, sparse features, and reconstructed signal of the initial partial discharge signal into a pre-defined denoising signal generation model, the model outputs a denoised partial discharge signal. This allows the model to focus on key components of the initial partial discharge signal during processing, effectively restoring a high-fidelity signal and improving the generation effect of the denoised signal, thus achieving efficient and accurate denoising. Furthermore, by determining the frequency domain features, sparse features, and reconstructed signal of the initial partial discharge signal and inputting them into the denoising signal generation model, redundant information in the signal can be removed, reducing unnecessary computational complexity and enabling convenient and efficient signal generation, thereby improving the speed and efficiency of generating denoised partial discharge signals. By combining sparse dictionary learning with generative adversarial network (GAN) models, a multi-layered signal processing architecture can be formed, which can more effectively separate signals and noise and improve denoising performance. Using a sparse dictionary to extract sparse features from the signal and then inputting them into the GAN model allows the GAN model to focus more on the key parts of the signal, reducing unnecessary computational complexity and improving training efficiency and generation performance. Furthermore, HiFi-GAN can be used to process partial discharge signals, removing noise and effectively restoring high-fidelity, high-quality partial discharge signals. By dynamically adjusting the sparse coding strategy based on the characteristics of the input signal and combining it with the GAN model, the adaptive learning capability of the GAN can be further utilized to accurately denoise different types of partial discharge signals. The sparse dictionary is used for efficient feature extraction and dimensionality reduction in the early stages of signal processing, removing redundant information, making subsequent GAN model processing simpler and more efficient. While ensuring processing performance, it can also significantly improve processing speed and efficiency.

[0173] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0174] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0175] Figure 4 A schematic diagram of a partial discharge signal denoising device provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below:

[0176] like Figure 4As shown, the noise reduction device 40 for partial discharge signals includes:

[0177] Acquisition module 41 is used to acquire the initial partial discharge signal of the power equipment;

[0178] Extraction module 42 is used to extract the frequency domain features of the initial partial discharge signal;

[0179] The reconstruction module 43 is used to perform sparse decomposition on the initial partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features of the initial partial discharge signal and the reconstructed signal.

[0180] The denoising module 44 is used to input frequency domain features, sparse features and reconstructed signals into a preset denoising signal generation model to obtain a denoised partial discharge signal output by the denoising signal generation model; wherein, the denoising signal generation model is obtained by training a generative adversarial network model based on the frequency domain features, sparse features and reconstructed signals of multiple historical partial discharge signals, as well as multiple real denoised partial discharge signals.

[0181] In one possible implementation, the partial discharge signal denoising device 40 further includes a building module for:

[0182] Acquire multiple historical partial discharge signals and an initial dictionary;

[0183] Select one historical partial discharge signal from multiple historical partial discharge signals and use it as the current signal, and use the initial dictionary as the current dictionary;

[0184] Based on the current dictionary, sparse coding is performed on the current signal to determine the sparse features of the current signal, as well as the dictionary matrix related to the current signal in the current dictionary.

[0185] The sparse dictionary is determined based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix.

[0186] In one possible implementation, the building module is specifically used for:

[0187] Based on the current signal, the current dictionary, and the sparse characteristics of the current signal, determine the residual for sparse representation of the current signal;

[0188] Based on the sparse features and residuals of the current signal, as well as the dictionary matrix, the updated dictionary is obtained;

[0189] Determine whether the change between the current dictionary and the updated dictionary is less than a preset threshold;

[0190] If the change between the current dictionary and the updated dictionary is greater than or equal to a preset threshold, the updated dictionary is used as the current dictionary. One historical partial discharge signal is selected from the remaining historical partial discharge signals as the current signal, and the steps of "sparsely encoding the current signal based on the current dictionary, determining the sparse features of the current signal, and the dictionary matrix related to the current signal in the current dictionary" and subsequent steps are repeated until the change between the current dictionary and the updated dictionary is less than the preset threshold, and the updated dictionary is used as the sparse dictionary.

[0191] In one possible implementation, the building module is specifically used for:

[0192] Set the relevant parameters for sparse coding; among which, the relevant parameters include the current sparse feature, the current residual, the preset residual threshold, the inactive set, the active set, the preset number of elements, the current iteration count, and the maximum iteration count;

[0193] Use the current signal as the current residual for initialization, use each dictionary column in the current dictionary as the element of the initial inactive set, and set the initial active set to an empty set;

[0194] Identify the element in the inactive set that has the strongest correlation with the current residual;

[0195] Add the element to the active set and remove the element from the inactive set to obtain the updated active set and the updated inactive set;

[0196] Based on the updated active set and the updated inactive set, determine the direction vector and step size for adjusting the sparse features;

[0197] Based on the current sparse features, direction vector and step size, adjust the sparse features to obtain the updated sparse features, and use the updated sparse features as the current sparse features;

[0198] Based on the current sparse features and the current dictionary, update the residual and use the updated residual as the current residual;

[0199] Determine whether the current residual is less than the preset residual threshold, whether the current iteration count is less than the maximum iteration count, and whether the number of elements in the active set is less than the preset number of elements;

[0200] If the current residual is equal to or greater than the preset residual threshold, and the current iteration count is less than the maximum iteration count, and the number of elements in the active set is less than the preset number of elements, then update the current iteration count and re-execute the step of "determining the element in the inactive set that has the greatest correlation with the current residual" and subsequent steps until the current residual is less than the preset residual threshold, or the current iteration count is equal to or greater than the maximum iteration count, or the number of elements in the active set is equal to or greater than the preset number of elements. Use the current sparse feature as the sparse feature for determining the current signal, and determine the dictionary matrix in the current dictionary that is related to the current signal based on the elements in the active set.

[0201] In one possible implementation, the partial discharge signal denoising device 40 further includes a training module for:

[0202] Acquire multiple historical partial discharge signals and multiple real denoised partial discharge signals;

[0203] The frequency domain features of each historical partial discharge signal are extracted, and based on the pre-built sparse dictionary, each historical partial discharge signal is sparsely decomposed to obtain the sparse features and reconstructed signal of each historical partial discharge signal.

[0204] The frequency domain characteristics, sparse characteristics, and reconstructed signal of each historical partial discharge signal are input into the generator of the generative adversarial network to obtain the noiseless generated signal output by the generator.

[0205] Multiple real denoised partial discharge signals and noiseless generated signals are input into the discriminator of the generative adversarial network to obtain the probability that each noiseless generated signal and each real denoised partial discharge signal output by the discriminator is a real signal.

[0206] Based on the probability, the model parameters of the discriminator and the generator are adjusted to obtain the denoised signal generation model.

[0207] In one possible implementation, the training module is specifically used for:

[0208] Calculate the first loss function value of the discriminator based on the probability;

[0209] The second loss function value of the generator is calculated based on the probability, the noiseless generated signal, and the real denoised partial discharge signal.

[0210] Based on the first loss function value, the model parameters of the discriminator are adjusted, and based on the second loss function value, the model parameters of the generator are adjusted to obtain the trained generative adversarial network model. Based on the trained generative adversarial network model, a denoised signal generation model is obtained.

[0211] In one possible implementation, the second loss function corresponding to the second loss function value includes an adversarial loss function, a feature matching loss function, and a spectrum reconstruction loss function;

[0212] The training module is specifically used for:

[0213] Calculate the adversarial loss function value of the generator based on the probability;

[0214] The feature matching loss function value of the generator is calculated based on the intermediate features of the noiseless generated signal in the discriminator and the intermediate features of the real denoised partial discharge signal in the discriminator.

[0215] Based on the spectrum diagram of the noiseless generated signal and the spectrum diagram of the real denoised partial discharge signal, calculate the spectrum reconstruction loss function value of the generator;

[0216] The second loss function value of the generator is calculated based on the adversarial loss function value, the feature matching loss function value, and the spectrum reconstruction loss function value.

[0217] In one possible implementation, the training module is also used for:

[0218] Based on a preset feature extraction network model, the first feature of the noiseless generated signal and the second feature of the real denoised partial discharge signal are extracted.

[0219] The perceptual loss function value of the generator is determined based on the first and second features;

[0220] The training module is also used for:

[0221] Update the second loss function value based on the perceptual loss function value.

[0222] Figure 5 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Figure 5 As shown, the electronic device 50 of this embodiment includes: a processor 51, a memory 52, and a computer program 53 stored in the memory 52 and executable on the processor 51. When the processor 51 executes the computer program 53, it implements the steps in the above-described embodiments of the methods for denoising partial discharge signals, for example... Figure 1 Steps S101 to S104 are shown. Alternatively, when processor 51 executes computer program 53, it implements the functions of each module in the above-described device embodiments, for example... Figure 4 The functions of modules 41 to 44 are shown.

[0223] For example, computer program 53 can be divided into one or more modules / units, one or more of which are stored in memory 52 and executed by processor 51 to complete the present invention. One or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 53 in electronic device 50. For example, computer program 53 can be divided into... Figure 4 Modules 41 to 44 are shown.

[0224] Electronic device 50 may include, but is not limited to, processor 51 and memory 52. ​​Those skilled in the art will understand that... Figure 5 This is merely an example of electronic device 50 and does not constitute a limitation on electronic device 50. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0225] The processor 51 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0226] The memory 52 can be an internal storage unit of the electronic device 50, such as a hard disk or RAM of the electronic device 50. The memory 52 can also be an external storage device of the electronic device 50, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the electronic device 50. Furthermore, the memory 52 can include both internal and external storage units of the electronic device 50. The memory 52 is used to store computer programs and other programs and data required by the electronic device. The memory 52 can also be used to temporarily store data that has been output or will be output.

[0227] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0228] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0229] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0230] In the embodiments provided by this invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For instance, the division of modules or 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; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0231] 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.

[0232] 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.

[0233] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0234] 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, and should all be included within the protection scope of the present invention.

Claims

1. A method for denoising partial discharge signals, characterized in that, include: Acquire the initial partial discharge signal of the power equipment; Extract the frequency domain features of the initial partial discharge signal; Based on a pre-constructed sparse dictionary, the initial partial discharge signal is sparsely decomposed to obtain sparse features and a reconstructed signal; the reconstructed signal is obtained by reconstructing the signal using the sparse features and the sparse dictionary. The frequency domain features, the sparse features, and the reconstructed signal are input into a preset denoising signal generation model to obtain a denoised partial discharge signal output by the denoising signal generation model; wherein, the denoising signal generation model is obtained by training a generative adversarial network model based on the frequency domain features, sparse features, and reconstructed signals of multiple historical partial discharge signals, as well as multiple real denoised partial discharge signals. Before inputting the frequency domain features, the sparse features, and the reconstructed signal into a preset denoising signal generation model to obtain the denoised partial discharge signal output by the denoising signal generation model, the method further includes: Acquire multiple historical partial discharge signals and multiple real denoised partial discharge signals; The frequency domain features of each historical partial discharge signal are extracted, and based on the pre-built sparse dictionary, each historical partial discharge signal is sparsely decomposed to obtain the sparse features and reconstructed signal of each historical partial discharge signal. The frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal are input into the generator of the generative adversarial network to obtain the noiseless generated signal output by the generator. Multiple real denoised partial discharge signals and the noiseless generated signal are input into the discriminator of the generative adversarial network to obtain the probability that each noiseless generated signal and each real denoised partial discharge signal output by the discriminator is a real signal. Based on the probability, adjust the model parameters of the discriminator and the model parameters of the generator to obtain a denoised signal generation model; Based on the probability, the model parameters of the discriminator and the model parameters of the generator are adjusted to obtain a denoised signal generation model, including: Calculate the first loss function value of the discriminator based on the probability; Based on the probability, the noiseless generated signal, and the real denoised partial discharge signal, the second loss function value of the generator is calculated; the second loss function corresponding to the second loss function value includes an adversarial loss function, a feature matching loss function, and a spectrum reconstruction loss function. Based on the first loss function value, the model parameters of the discriminator are adjusted, and based on the second loss function value, the model parameters of the generator are adjusted to obtain a trained generative adversarial network model. Based on the trained generative adversarial network model, a denoised signal generation model is obtained.

2. The method for denoising partial discharge signals according to claim 1, characterized in that, Before performing sparse decomposition on the initial partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features and reconstructed signal of the initial partial discharge signal, the method further includes: Acquire multiple historical partial discharge signals and an initial dictionary; Select one historical partial discharge signal from the plurality of historical partial discharge signals and use it as the current signal, and use the initial dictionary as the current dictionary; Based on the current dictionary, sparse coding is performed on the current signal to determine the sparse features of the current signal, as well as the dictionary matrix related to the current signal in the current dictionary. The sparse dictionary is determined based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix.

3. The method for denoising partial discharge signals according to claim 2, characterized in that, The step of determining the sparse dictionary based on the current signal, the current dictionary, the sparse features of the current signal, and the dictionary matrix includes: Based on the current signal, the current dictionary, and the sparse characteristics of the current signal, determine the residual for sparse representation of the current signal; Based on the sparse features and residuals of the current signal, as well as the dictionary matrix, the updated dictionary is obtained; Determine whether the change between the current dictionary and the updated dictionary is less than a preset threshold; If the change between the current dictionary and the updated dictionary is greater than or equal to a preset threshold, the updated dictionary is used as the current dictionary. One historical partial discharge signal is selected from the remaining historical partial discharge signals as the current signal, and the steps of "sparsely encoding the current signal based on the current dictionary, determining the sparse features of the current signal, and the dictionary matrix related to the current signal in the current dictionary" and subsequent steps are repeated until the change between the current dictionary and the updated dictionary is less than the preset threshold, and the updated dictionary is used as the sparse dictionary.

4. The method for denoising partial discharge signals according to claim 2, characterized in that, Based on the current dictionary, sparse coding is performed on the current signal to determine its sparse features, as well as the dictionary matrix related to the current signal, including: Set relevant parameters for sparse coding; wherein, the relevant parameters include the current sparse feature, the current residual, the preset residual threshold, the inactive set, the active set, the preset number of elements, the current iteration count, and the maximum iteration count; Use the current signal as the current residual for initialization, use each dictionary column in the current dictionary as the element of the initial inactive set, and set the initial active set to an empty set; Identify the element in the inactive set that has the strongest correlation with the current residual; Add the element to the active set and remove the element from the inactive set to obtain the updated active set and the updated inactive set; Based on the updated active set and the updated inactive set, determine the direction vector and step size for adjusting the sparse features; Based on the current sparse features, the direction vector, and the step size, the sparse features are adjusted to obtain updated sparse features, and the updated sparse features are used as the current sparse features. Based on the current sparse features and the current dictionary, update the residual and use the updated residual as the current residual; Determine whether the current residual is less than the preset residual threshold, whether the current iteration count is less than the maximum iteration count, and whether the number of elements in the active set is less than the preset number of elements; If the current residual is equal to or greater than the preset residual threshold, and the current iteration count is less than the maximum iteration count, and the number of elements in the active set is less than the preset number of elements, then update the current iteration count and re-execute the step of "determining the element in the inactive set that has the greatest correlation with the current residual" and subsequent steps until the current residual is less than the preset residual threshold, or the current iteration count is equal to or greater than the maximum iteration count, or the number of elements in the active set is equal to or greater than the preset number of elements. Use the current sparse feature as the sparse feature for determining the current signal, and determine the dictionary matrix in the current dictionary that is related to the current signal based on the elements in the active set.

5. The method for denoising partial discharge signals according to any one of claims 1 to 4, characterized in that, Based on the probability, the noiseless generated signal, and the actual denoised partial discharge signal, the second loss function value of the generator is calculated, including: Calculate the adversarial loss function value of the generator based on the probability; The feature matching loss function value of the generator is calculated based on the intermediate features of the noiseless generated signal in the discriminator and the intermediate features of the real denoised partial discharge signal in the discriminator. Based on the spectrum diagram of the noiseless generated signal and the spectrum diagram of the real denoised partial discharge signal, calculate the spectrum reconstruction loss function value of the generator; The second loss function value of the generator is calculated based on the adversarial loss function value, the feature matching loss function value, and the spectrum reconstruction loss function value.

6. The method for denoising partial discharge signals according to any one of claims 1 to 4, characterized in that, After inputting the frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal into the generator of the generative adversarial network to obtain the noiseless generated signal output by the generator, the process further includes: Based on a preset feature extraction network model, the first feature of the noiseless generated signal and the second feature of the real denoised partial discharge signal are extracted. The perceptual loss function value of the generator is determined based on the first feature and the second feature; After calculating the second loss function value of the generator based on the probability, the noiseless generated signal, and the real denoised partial discharge signal, the method further includes: Update the second loss function value based on the perceived loss function value.

7. A noise reduction device for partial discharge signals, characterized in that, include: The acquisition module is used to acquire the initial partial discharge signal of the power equipment; An extraction module is used to extract the frequency domain features of the initial partial discharge signal; The reconstruction module is used to perform sparse decomposition on the initial partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features and reconstructed signal of the initial partial discharge signal; the reconstructed signal is obtained by reconstructing the signal using the sparse features and the sparse dictionary. A denoising module is used to input the frequency domain features, the sparse features, and the reconstructed signal into a preset denoising signal generation model to obtain a denoised partial discharge signal output by the denoising signal generation model; wherein, the denoising signal generation model is obtained by training a generative adversarial network model based on the frequency domain features, sparse features, and reconstructed signals of multiple historical partial discharge signals, as well as multiple real denoised partial discharge signals. A training module is used to acquire multiple historical partial discharge signals and multiple real denoised partial discharge signals; extract the frequency domain features of each historical partial discharge signal, and perform sparse decomposition on each historical partial discharge signal based on a pre-built sparse dictionary to obtain the sparse features and reconstructed signal of each historical partial discharge signal; input the frequency domain features, sparse features, and reconstructed signal of each historical partial discharge signal into the generator of a generative adversarial network to obtain the noiseless generated signal output by the generator; input the multiple real denoised partial discharge signals and the noiseless generated signal into the discriminator of the generative adversarial network to obtain the probability that each noiseless generated signal and each real denoised partial discharge signal output by the discriminator is a real signal; adjust the model parameters of the discriminator and the model parameters of the generator according to the probability to obtain the denoised signal generation model; The training module is specifically used for: Calculate the first loss function value of the discriminator based on the probability; Based on the probability, the noiseless generated signal, and the real denoised partial discharge signal, the second loss function value of the generator is calculated; the second loss function corresponding to the second loss function value includes an adversarial loss function, a feature matching loss function, and a spectrum reconstruction loss function. Based on the first loss function value, the model parameters of the discriminator are adjusted, and based on the second loss function value, the model parameters of the generator are adjusted to obtain a trained generative adversarial network model. Based on the trained generative adversarial network model, a denoised signal generation model is obtained.

8. An electronic device comprising a memory and a processor, the memory for storing a computer program, the processor for calling and running the computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6 above.

Images