Current signal compression acquisition reconstruction method and device, storage medium and electronic equipment

Abstract

The invention relates to the technical field of current signal processing, in particular to a current signal compression collection reconstruction method and device, a storage medium and electronic equipment, and the method comprises the steps: carrying out the sliding collection of a current signal through a flexible Blackman window, and obtaining a sliding collection signal; constructing a partial orthogonal Gaussian measurement matrix; performing compression acquisition on the sliding acquisition signal by using the partial orthogonal Gaussian measurement matrix to obtain a compressed signal; carrying out sparse transformation on the sliding acquisition signal, and solving to obtain a frequency domain transformation matrix; calculating by using the frequency domain transformation matrix and the partial orthogonal Gaussian measurement matrix to obtain a sensing matrix, and iteratively solving a sparse signal by using an algorithm according to the compressed signal and the sensing matrix; and according to the frequency domain transformation matrix and the sparse signal, calculation is carried out to obtain reconstruction data. By means of compression acquisition and reconstruction, low-volume data acquisition is achieved, high-frequency fault arc characteristics in data are effectively reserved, and the problem of fault arc characteristic deficiency after traditional current signal acquisition is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of current signal processing, and in particular to a current signal compression acquisition and reconstruction method, device, storage medium and electronic equipment. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In modern power systems, arcing is a common and complex form of electrical fault, which is widely present in various power facilities such as high-voltage equipment, substations, and transmission lines. Arcing can not only cause equipment damage and system power outages, but can also cause fires, resulting in casualties and significant economic losses. Therefore, real-time monitoring and analysis of multi-source current signals is of great practical significance.

[0004] Traditional fault arc identification methods usually rely on the monitoring of non-electrical quantities such as light and gas, which requires a large number of monitoring equipment, is complex to install and has high costs. However, by monitoring electrical parameters such as voltage, current, and power for fault arc identification, no additional monitoring equipment is required, reducing hardware costs. However, since the current signal exhibits high-frequency characteristics when a fault arc occurs, and traditional signal monitoring sampling must meet the Nyquist-Shannon theorem, the sampling, transmission and processing costs of high-frequency signals are huge, and higher requirements are placed on the storage space and computing power of on-site monitoring equipment. Therefore, how to efficiently and accurately collect and process multi-source current signals to ensure the reliability of fault arc identification has become a difficult problem that needs to be solved urgently. Summary of the invention

[0005] In order to overcome the shortcomings of the background technology, the present invention provides a current signal compression acquisition and reconstruction method, device, storage medium and electronic device. By compressing, acquiring and reconstructing high-frequency current signals that can characterize the characteristics of fault arcs, low-volume data acquisition is achieved, and the high-frequency fault arc characteristics in the data are effectively retained, solving the problem of missing fault arc characteristics after traditional current signal acquisition.

[0006] To achieve the above object, the present invention provides the following technical solutions: In a first aspect, a current signal compression acquisition and reconstruction method is provided, the method comprising: Using a flexible Blackman window to measure the current signal Perform sliding acquisition to obtain sliding acquisition signal ; Constructing a partially orthogonal Gaussian measurement matrix ; Using the partially orthogonal Gaussian measurement matrix The sliding acquisition signal After compression acquisition, the compressed signal is obtained ; The sliding acquisition signal Perform sparse transformation and solve the frequency domain transformation matrix ; Using the frequency domain transformation matrix and the partially orthogonal Gaussian measurement matrix The perception matrix D is calculated, and then the compression signal is The perception matrix D is solved iteratively using the algorithm for sparse signals ; According to the frequency domain transformation matrix and the sparse signal Calculate the reconstructed data .

[0007] Further, The functional expression of the flexible Blackman window is: (1) In formula (1), t is the current signal acquisition time, T The width of the window can be adjusted.

[0008] Further, The flexible Blackman window is used to analyze the current signal Perform sliding acquisition to obtain sliding acquisition signal , as follows: (2) (3) in, is the number of discrete data points collected, is the sampling interval, is the pulse function, .

[0009] Further, Adjust the window width according to the deviation of the adjacent window reconstruction signal of the current signal T , the deviation of the reconstructed signal of the adjacent window of the current signal for: (4) In formula (4), , They are the h-1th and hth windows respectively; When an arc fault occurs, the current harmonic frequency increases, resulting in Increase, then reduce the window width T ; when When it decreases, the window width is expanded T .

[0010] Further, The partially orthogonal Gaussian measurement matrix is ​​constructed include: Based on the standard Gaussian measurement matrix with independent and identical distribution, a partial orthogonal Gaussian measurement matrix is ​​constructed ; The standard Gaussian measurement matrix contains random variables The following distribution is satisfied: (5) In formula (5), The mean is 0 and the variance is Gaussian distribution.

[0011] Further, The standard Gaussian measurement matrix is ​​expressed as the following matrix , of size m×n: (6) Extract Matrix The first k columns of the matrix : (7) Pair Matrix Perform orthogonal processing and select the first column as the orthogonal basis vector , the jth column needs to subtract its projection on all previous orthogonal basis vectors and normalize it: (8) Get the orthogonalized matrix: (9) The matrix The first k columns of Q k Matrix, get the partially orthogonal Gaussian matrix : (10).

[0012] Further, In the extraction matrix The first k columns of the matrix When the k value is 0.2×n, according to formula (4), the deviation of the adjacent window reconstruction signal of the current signal is Make dynamic adjustments; when When it increases, the k value is reduced; when When it decreases, increase the k value; The k value selection range is (0.1~0.3)×n.

[0013] Further, According to compressed sensing, the compressed signal The calculation formula is as follows: (11).

[0014] Further, The sliding acquisition signal is processed by discrete wavelet transform Perform sparse transformation and solve the frequency domain transformation matrix .

[0015] Further, The sliding acquisition signal is processed by using discrete wavelet transform Perform sparse transformation and solve the frequency domain transformation matrix ,include: The sliding acquisition signal is analyzed using the Daubechies wavelet basis. Perform N-1 decompositions to obtain 1 low-frequency component and N-1 high-frequency components: (12) In formula (12), is the low frequency component, is the high frequency component, and are the scale function and wavelet function respectively, j and k are the scale and translation numbers respectively; Further transforming equation (12), the sliding acquisition signal The sparse signal under discrete wavelet transform is represented as: (13) In formula (13), is the frequency domain transformation matrix under discrete wavelet transform.

[0016] Further, The calculation formula of the perception matrix D is: (14) In formula (14), is the frequency domain transformation matrix, is a partially orthogonal Gaussian measurement matrix.

[0017] Further, According to the compressed signal and the perception matrix D, and the sparse signal is iteratively solved using the compressed sampling matching pursuit reconstruction algorithm CoSaMP-ESP based on optimal sparsity estimation .

[0018] Further, The compressed signal and the perception matrix D, and the sparse signal is iteratively solved using the compressed sampling matching pursuit reconstruction algorithm CoSaMP-ESP based on optimal sparsity estimation ,include: Input the compressed signal , the perception matrix D and the error threshold , the compressed sampling matching pursuit reconstruction algorithm CoSaMP-ESP based on optimal sparsity estimation includes: ① Initialization, set the index value set E to an empty set, F to an empty set, and the residual v= , sparsity K=5, number of iterations d=0; ②d=d+1, calculate the inner product of the current residual and the perception matrix D, and select the most qualified column of the perception matrix D according to the fuzzy threshold c and store it in the set E: (15) In formula (15), D j is the j-th column of matrix D, v d-1 is the residual of the d-1th iteration; ③According to the set E, we get the support set : (16) ④ According to the support set Find the least squares solution: (17) In formula (17), , Represents the set in the perception matrix D The corresponding columns, I is the identity matrix; ⑤Cut the support set, Select the K items with the largest absolute value and store them in z(k), and update the corresponding , and then update the sparsity estimate K: (18) In formula (18), K is the sparsity, is the support set of the dth iteration, h(m) is the compressed data, D is the perception matrix, z(k) is the sparse signal to be solved, and the adjustable p-norm is used to control the sparsity; ⑥Update residual and relevance : (19) (20) in, is the residual, is the correlation, representing all columns of the perception matrix D D j The column with the highest correlation with the current residual; ⑦If , then stop the iteration and output the sparse signal , otherwise return to step ② to recalculate the residual.

[0019] Further, The value range of the fuzzy threshold c is 0.6~0.8.

[0020] Further, In formula (18), the p-value base is 1, and the p-value range is 0.3~3.

[0021] Further, The reconstructed data The calculation formula is: (twenty one) In formula (21), is the frequency domain transformation matrix, is the sparse signal solved by iteration.

[0022] In a second aspect, a current signal compression acquisition and reconstruction device is also provided, the device comprising: Sliding acquisition module for current signals using a flexible Blackman window Perform sliding acquisition to obtain sliding acquisition signal ; Matrix building module for building partially orthogonal Gaussian measurement matrices ; A compression acquisition module is used to utilize the partially orthogonal Gaussian measurement matrix The sliding acquisition signal After compression acquisition, the compressed signal is obtained ; A sparse transformation module is used to collect the sliding signal Perform sparse transformation and solve the frequency domain transformation matrix ; A calculation and solution module is used to use the frequency domain transformation matrix and the partially orthogonal Gaussian measurement matrix The perception matrix D is calculated, and then the compression signal is The perception matrix D is solved iteratively using the algorithm for sparse signals ; A data reconstruction module is used to transform the frequency domain matrix and the sparse signal Calculate the reconstructed data .

[0023] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing one or more programs, which, when executed, can implement the aforementioned current signal compression acquisition and reconstruction method.

[0024] Based on the same inventive concept, the present invention also provides an electronic device, including a processor, a communication interface, a computer-readable storage medium as described above, and a communication bus; wherein the processor, the communication interface, and the computer-readable storage medium communicate with each other via the communication bus; and the processor is used to execute the program stored in the computer-readable storage medium.

[0025] Compared with the prior art, the present invention has the following beneficial effects: 1. It can effectively break through the limitation of Nyquist high-frequency sampling, realize low-speed dynamic compression sampling and accurate reconstruction of high-frequency current signals of fault arcs, retain the effective characteristics of fault arcs in current signals, provide data acquisition technology support for fault arc identification in power grid operation, and has good engineering application prospects; 2. Use a flexible Blackman window to perform sliding acquisition of current signals, with lower sidelobe peaks and faster sidelobe attenuation rates, and higher data spectrum acquisition. Use a flexible Blackman window to adjust the window width according to the degree of electrical fluctuations, which can effectively collect the original information of the current signal and retain the high-frequency characteristics of the fault arc. 3. Using the CoSaMP-ESP reconstruction algorithm, the reconstruction speed is fast and the reconstruction accuracy is high, which can provide strong technical support for fault arc identification; 4. Use compressed sensing technology to compress and collect current signals, reducing the amount of collected current signals and speeding up the transmission efficiency of power data.

[0026] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings.

[0027] The present invention will be further described below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0029] Figure 1 A schematic diagram of the steps of a current signal compression acquisition and reconstruction method according to an embodiment of the present invention; Figure 2 A schematic diagram of a flow chart of a current signal compression acquisition and reconstruction method according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a current signal compression acquisition and reconstruction device according to an embodiment of the present invention; Figure 4 A schematic diagram comparing the reconstruction error MSE of the reconstruction algorithm of the embodiment of the present invention and other algorithms with the change of compression ratio CR; Figure 5 A schematic diagram comparing the reconstruction time T of the reconstruction algorithm of the embodiment of the present invention and other algorithms with the change of the signal sampling volume; Figure 6 The figure is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the implementation of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] like Figure 1 to Figure 2 As shown, the first embodiment of the present invention provides a current signal compression acquisition and reconstruction method, the method comprising the following steps: S1. Use flexible Blackman window to analyze the current signal Perform sliding acquisition to obtain sliding acquisition signal ; This step is the sliding acquisition of current signals. Specifically, a flexible Blackman window is used to slide the current signals and the window adjustment parameters are introduced to acquire data. .

[0032] S2. Constructing a partially orthogonal Gaussian measurement matrix ; This step is to construct a partially orthogonal Gaussian measurement matrix. Specifically, according to the restricted isometry property (RIP) in compressed sensing technology and the requirement that the correlation between the measurement matrix and the sparse basis is very low, a partially orthogonal Gaussian measurement matrix with independent and identical distribution and orthogonal characteristics is constructed. .

[0033] S3, using the partially orthogonal Gaussian measurement matrix The sliding acquisition signal After compression acquisition, the compressed signal is obtained ; This step is the compressed sensing of the current signal. Specifically, the partially orthogonal Gaussian measurement matrix obtained in step S2 is used to compress and collect the windowed current signal according to the compressed sensing technology to obtain the compressed data: .

[0034] S4, collecting the sliding signal Perform sparse transformation and solve the frequency domain transformation matrix ; This step is a signal sparse transformation. Specifically, the current signal after windowing in step S1 is Perform sparse transformation and solve the frequency domain transformation matrix under DWT transformation (Discrete Wavelet Transform) .

[0035] S5, using the frequency domain transformation matrix and the partially orthogonal Gaussian measurement matrix The perception matrix D is calculated, and then the compression signal is The perception matrix D is solved iteratively using the algorithm for sparse signals ; This step is to iteratively solve the sparse signal. Specifically, the frequency domain transformation matrix in step S4 is used and partially orthogonal Gaussian measurement matrices The perception matrix D is calculated, and then the CoSaMP-ESP improved compressed sampling matching pursuit reconstruction algorithm is used to solve the sparse signal .

[0036] S6, according to the frequency domain transformation matrix and the sparse signal Calculate the reconstructed data .

[0037] This step is to reconstruct the current signal. Specifically, according to the sparse signal obtained by iterative solution in step S5 , calculate the reconstructed data .

[0038] In the above technical solution, by compressing and collecting and reconstructing the high-frequency current signal that can characterize the fault arc characteristics, low-volume data collection is achieved, and the high-frequency fault arc characteristics in the data are effectively retained, solving the problem of missing fault arc characteristics after traditional current signal collection.

[0039] The embodiment of the present invention actually provides a current signal compression acquisition and reconstruction method taking into account fault arc identification. The method has the following characteristics: in view of the problem of difficulty in feature acquisition and monitoring when a fault arc occurs in various power facilities such as high-voltage equipment, substations, and transmission lines in modern power grids, the embodiment of the present invention proposes a current signal compression acquisition and reconstruction method taking into account fault arc identification, aiming at the high-frequency characteristics of the current signal before and after the arc fault occurs, so as to effectively break through the limitation of Nyquist high-frequency sampling, realize low-speed dynamic compression sampling and accurate reconstruction of the high-frequency current signal of the fault arc, retain the effective characteristics of the fault arc in the current signal, provide data acquisition technology support for fault arc identification in power grid operation, and have good engineering application prospects.

[0040] Each step of the above method will be further described below.

[0041] 1. Current signal sliding acquisition The embodiment of the present invention uses a flexible Blackman window to detect the current signal representing the fault arc characteristics Sliding window acquisition is performed. Compared with traditional orthogonal Gaussian windows, Hanning windows and other windows, the Blackman window has a lower sidelobe peak and a faster sidelobe decay rate, and has the advantage of higher data spectrum acquisition, which can effectively retain the original characteristics of the current signal. The flexible Blackman window function is specifically expressed as: (1) In formula (1), t is the current signal acquisition time, T The width of the window can be adjusted.

[0042] The above window function is used to analyze the current signal Collect and obtain the current signals of different time windows to obtain the collected signals , as follows: (2) (3) in, is the number of discrete data points collected, is the sampling interval, is the pulse function, .

[0043] Adjust the window width according to the deviation of the adjacent window reconstruction signal of the current signal T , neighborhood reconstruction bias for: (4) In formula (4), , They are the h-1th and hth windows respectively.

[0044] The above-mentioned flexible Blackman window utilizes flexible feedback modulation technology. The window width of the traditional Blackman window is fixed, but the embodiment of the present invention can adjust the window width according to the deviation of the adjacent window reconstruction signal of the current signal. T , so it is called a flexible Blackman window. The advantage of using a flexible Blackman window is that when an arc fault occurs, the current harmonic frequency increases, resulting in When the value increases, the window is reduced, which can effectively capture the instantaneous change of current and improve the arc fault feature acquisition capability; When it decreases, the window is expanded to effectively retain the signal spectrum characteristics. Using a flexible acquisition window (i.e., a flexible Blackman window) and adjusting the window width according to the degree of electrical fluctuation can effectively collect the original information of the current signal and retain the high-frequency characteristics of the fault arc.

[0045] 2. Construct a partially orthogonal Gaussian measurement matrix The measurement matrix in the compressed sensing technology needs to satisfy the restricted isometry property (RIP) condition, which requires that the correlation between the measurement matrix and the sparse basis is very low. Therefore, the embodiment of the present invention is based on a Gaussian measurement matrix with independent and identical distribution, and constructs a partially orthogonal Gaussian measurement matrix, so as to retain the characteristics of the standard Gaussian random matrix, and at the same time improve the correlation control ability of the matrix through partial orthogonalization, thereby reducing the amount of calculation during reconstruction.

[0046] Random variables in the selected Gaussian measurement matrix The following distribution is satisfied: (5) In formula (5), The mean is 0 and the variance is Gaussian distribution.

[0047] Standard Gaussian measurement matrix As follows, the size is m×n: (6) Extract Matrix The first k columns of the matrix , the k value benchmark is 0.2×n, according to formula (4) the neighborhood reconstruction deviation Dynamic adjustment, when When it increases, the k value decreases; when When decreasing, increase the k value, and the k value selection range is (0.1~0.3)×n: (7) Pair Matrix Perform orthogonal processing and select the first column as the orthogonal basis vector , the jth column needs to subtract its projection on all previous orthogonal basis vectors and normalize it: (8) Get the orthogonalized matrix: (9) The orthogonalized k-column matrix Q k and the remaining non-orthogonalized columns A k+1:n Combined together, the first k columns of matrix A are replaced by Q k Matrix, get the partially orthogonal Gaussian matrix , which can improve the correlation control ability of the matrix and reduce the amount of calculation during reconstruction: (10).

[0048] 3. Compressed sensing of current signals The signal processing of the embodiment of the present invention adopts compressed sensing technology. Compressed sensing (CS), as an emerging signal processing technology, provides a new solution. The theory of compressed sensing is based on the sparsity of the signal, and the original signal can be reconstructed through a small amount of non-correlated measurement data. This theory breaks through the limitations of the traditional Nyquist sampling theorem and can significantly reduce the amount of data in the sampling stage while ensuring the integrity and accuracy of the signal at the reconstruction end. Compressed sensing technology provides an efficient solution for the compressed acquisition of current signals, which can significantly improve the ability to monitor and diagnose current signals before and after arc faults occur.

[0049] Using the partially orthogonal Gaussian measurement matrix obtained in step 2 above, the windowed current signal is compressed according to the compressed sensing technology to obtain the compressed data: , after compression, the initial n data are compressed into m data, the specific formula is as follows: (11).

[0050] The compressed sensing technology is used to compress and collect current signals, which reduces the amount of collected current signals and speeds up the transmission efficiency of power data.

[0051] 4. Data sparse transformation For the current signal after windowing in step 1 Perform sparse transformation. Perform DWT transformation (discrete wavelet transform) and use the Daubechies wavelet basis to perform N-1 decompositions to obtain 1 low-frequency component (approximate coefficient) and N-1 high-frequency components (detail coefficients): (12) In formula (12), is the low frequency component, is the high frequency component, and are the scaling function and wavelet function respectively, j and k are the scaling number and translation number respectively.

[0052] The right side of equation (12) is sorted out, and the matrix is ​​decomposed according to the principle of signal sparsity to obtain the sparse signal z(k) and the frequency domain transformation matrix , and then further invert both sides of the matrix, the windowed current signal The sparse signal under DWT transformation is expressed as: (13) In formula (13), is the frequency domain transformation matrix under DWT transformation, according to The matrix on the left side of the expression is inverted. Here, The matrix is ​​obtained by transforming formula (12).

[0053] In this embodiment, step 4 belongs to the compression end, and the frequency domain transformation matrix is ​​obtained by representing the sparse signal , which is used for subsequent current signal reconstruction; Step 5 belongs to the reconstruction end, z(k) is unknown, and only the compressed data h(m) and the sensing matrix D are obtained. It is necessary to iteratively solve z(k) first, and then according to the matrix The original signal x(n) is obtained by summing up equation (13).

[0054] 5. Iterative solution of sparse signals First, use the DWT frequency domain transformation matrix in the above steps and partially orthogonal Gaussian measurement matrices The perception matrix D is calculated: (14) Then, the sparse signal is solved using the Compressive Sampling Matching Pursuit based on Optimal Sparsity Estimation (CoSaMP-ESP) algorithm. , can effectively reconstruct the original fault arc characteristics in the current signal. Using the CoSaMP-ESP reconstruction algorithm, the reconstruction speed is fast and the reconstruction accuracy is high, which can provide strong technical support for fault arc identification.

[0055] Input compressed current signal , perception matrix D and error threshold The CoSaMP-ESP algorithm process is as follows: ① Initialization, set the index value set E to an empty set, F to an empty set, and the residual v= , sparsity K=5, number of iterations d=0; here K=5 is an empirical value, and the K value is iteratively updated in the subsequent step ⑤, so the initial value of K has little effect on the final algorithm reconstruction effect.

[0056] ②d=d+1, calculate the inner product of the current residual and the perception matrix D, and select the most qualified column of the matrix D and store it in E according to the fuzzy threshold c (the value in this embodiment is 0.6~0.8), thereby reducing the possibility of the wrong column being included in the support set, thereby reducing the computational complexity and reconstruction error: (15) In formula (15), D j is the j-th column of matrix D, v d-1 is the residual of the d-1th iteration.

[0057] Among them, "fuzzy threshold" refers to an adjustment mechanism introduced in the candidate atom selection process to reduce the possibility of erroneous atoms (i.e. inaccurate atoms) being included in the support set. The setting of fuzzy threshold can reduce the possible erroneous selection problems in the candidate atom set. The traditional CoSaMP algorithm selects the 2K atoms with the largest absolute value of the inner product of the maximum residual and the perception matrix each time. This method of fixing the number of selections is prone to erroneous atoms when the sparsity is large, thereby increasing the computational complexity and reconstruction error.

[0058] ③According to the set E calculated by the fuzzy threshold in the previous step, the support set is obtained : (16) ④ According to the support set Solve the least squares solution and find a solution that minimizes the error by minimizing the sum of squares of the error, that is, find the estimated value of the reconstructed signal so that the error between this estimated value and the actual observed data is minimized: (17) In formula (17), , Represents the matrix D and the set The corresponding column represents the current collection The coefficient vector in , the dimension of this vector is the same as the set The number of elements in . When operating the perception matrix, the unit matrix I is introduced to solve the problem that the inverse of the matrix cannot be found during reconstruction.

[0059] ⑤Cut the support set, Select the K items with the largest absolute value and store them in z(k) to ensure that the support set contains only those columns of D that are most important for signal reconstruction, and update the corresponding The difference between the embodiment of the present invention and the traditional reconstruction algorithm is that the sparsity estimation value K is updated to dynamically control the support set The size of D is adapted to the sparse characteristics of the signal, improving the accuracy and sparse effect of signal reconstruction, while avoiding too many irrelevant columns of D from entering the support set and reducing the reconstruction error: (18) In formula (18), K is the sparsity, is the support set of the dth iteration, h(m) is the compressed data, D is the perception matrix, z(k) is the sparse signal to be solved, and the adjustable p-norm is used to control the sparsity. When p<1, the norm tends to promote sparsity, making more small coefficients tend to zero; when p>1, the norm tends to reduce sparsity. In practical applications, different current data frequencies have different sparse structures. When the current data frequency is high, the p-value is increased, and when the current data frequency is reduced, the p-value is reduced. By adjusting the appropriate p-value, the model can better adapt to the sparse structure of the data and improve the performance and generalization ability of the model. The p-value benchmark is 1, and the range of variation is generally 0.3~3.

[0060] ⑥Update residual and relevance : (19) (20) in, is the residual, is the relevance, Represents all the columns of matrix D D jThe column with the highest correlation with the current residual in reflects the reconstructed signal and D j The maximum correlation between them is used to guide the column selection of the matrix D for the next iteration. D j This indicates that they are more correlated with the residuals, so these can be given priority in subsequent iterations. D j The support set can be entered to reduce the residual error faster and more effectively and improve the reconstruction accuracy.

[0061] ⑦If , then stop the iteration and output the sparse signal , otherwise return to step ② and recalculate the residual.

[0062] In some embodiments, the error threshold =10 -3 , this value can better meet the calculation requirements of the CoSaMP-ESP algorithm.

[0063] 6. Current signal reconstruction The coefficient signal obtained by iterative estimation according to the above steps , calculate the reconstructed data : (twenty one) In formula (21), is the frequency domain transformation matrix, is the sparse signal solved by iteration.

[0064] In summary, the embodiment of the present invention first uses the Blackman window to perform sliding acquisition on the current signal that can characterize the fault arc characteristics, and introduces flexible feedback modulation technology to adjust the window width according to the neighborhood reconstruction deviation, so as to realize effective monitoring of high-frequency fluctuations of the current signal; secondly, an improved partially orthogonal Gaussian measurement matrix is ​​introduced, and compressed sensing technology is used to realize compressed acquisition of current signals; then, DWT discrete wavelet transform technology is used to realize sparse transformation of the collected current signal; finally, a compressed sampling matching pursuit reconstruction algorithm (CoSaMP-ESP) based on optimal sparsity estimation is proposed to realize accurate iterative reconstruction of low-dimensional data.

[0065] The second embodiment of the present invention also provides a current signal compression acquisition and reconstruction device, such as Figure 3 As shown, the device comprises: Sliding acquisition module for current signals using a flexible Blackman window Perform sliding acquisition to obtain sliding acquisition signal ; Matrix building module for building partially orthogonal Gaussian measurement matrices ; A compression acquisition module is used to utilize the partially orthogonal Gaussian measurement matrix The sliding acquisition signal After compression acquisition, the compressed signal is obtained ; A sparse transformation module is used to collect the sliding signal Perform sparse transformation and solve the frequency domain transformation matrix ; A calculation and solution module is used to use the frequency domain transformation matrix and the partially orthogonal Gaussian measurement matrix The perception matrix D is calculated, and then the compression signal is The perception matrix D is solved iteratively using the algorithm for sparse signals ; A data reconstruction module is used to transform the frequency domain matrix and the sparse signal Calculate the reconstructed data .

[0066] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0067] The following will further explain in conjunction with specific embodiments.

[0068] According to a specific embodiment, the historical data of current signals generated by normal operation and arc fault of a substation in a certain city in China is used as a validation data set for calculation, the initial sampling time window width is 100ms, and the time window width range is set to 1 power frequency cycle to 20 power frequency cycles, that is, the window adjustment range is 20ms to 400ms, the original data sampling frequency is 150kHz, and the sampling time is 5s. The initial window width (that is, the initial sampling time window width) and the width adjustment range (that is, the window width range) here are the flexible Blackman window width in step 1 T The initial width and adjustment range of ; the number of discrete points n collected in step 1 can be calculated based on the sampling frequency and sampling time.

[0069] In order to verify the effect of the method of the embodiment of the present invention on the compressed acquisition and reconstruction of the current signal when an arc fault occurs, the compressed sampling matching pursuit (CoSaMP) algorithm, the variable step sparsity self-estimating subspace pursuit (VSSESP) algorithm and the CoSaMP-ESP algorithm proposed in the embodiment of the present invention are used to reconstruct and compare the collected signals. Taking the A phase current signal data as an example, the reconstruction accuracy and efficiency are as follows: Figure 4 and Figure 5 The CoSaMP algorithm and the VSSESP algorithm are both prior art and will not be described in detail here.

[0070] Depend on Figure 4 It can be seen that when the compression ratio CR (compressed data / original data) is between 0.9 and 0.1, the mean square error (mean square error between the original signal and the reconstructed signal) MSE of the CoSaMP-ESP algorithm proposed in the embodiment of the present invention is smaller than that of the CoSaMP and VSSESP algorithms, which proves that the reconstruction accuracy of the method proposed in the embodiment of the present invention is higher; Figure 5 It can be seen that under the same data volume, the time required for the CoSaMP-ESP reconstruction proposed in the embodiment of the present invention is less than that of the other two methods, indicating that the reconstruction efficiency of the method in the embodiment of the present invention is higher.

[0071] The reconstructed current signal is divided into a training set and a test set according to the random sampling principle at a ratio of 7:3. The data set is trained and verified using the typical fault arc recognition model-CNN model. The CNN network structure is set to input layer, fully connected layer, convolution layer (2×1×1×16, 1 channel, 16 convolution kernels), batch normalization layer, activation layer (Rule function), convolution layer (2×1×16×32, 16 channels, 32 convolution kernels), batch normalization layer, activation layer (Rule function), fully connected layer, Softmax layer and output layer; the optimizer uses the Adam algorithm, the regularization parameter is set to 1e-1, the learning rate is 1e-3, and the number of iterations is 3000. Training and testing are performed using the Adam algorithm, and the recognition accuracy results of the test sets of different algorithms are shown in Table 1. The results show that the arc fault recognition accuracy is higher under the reconstruction method of the embodiment of the present invention, that is, the arc fault feature retention in the current signal before and after data compression and reconstruction is better.

[0072] Table 1 Comparison of arc fault identification accuracy under different reconstruction algorithms

[0073] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing one or more programs, which, when executed, can implement the aforementioned current signal compression acquisition and reconstruction method.

[0074] Based on the same inventive concept, the present invention also provides an electronic device, such as Figure 6 As shown, it includes a processor, a communication interface, a computer-readable storage medium as described above, and a communication bus; wherein the processor, the communication interface, and the computer-readable storage medium communicate with each other via the communication bus; and the processor is used to execute the program stored in the computer-readable storage medium.

[0075] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0076] The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, the functional modules in the various embodiments of the present invention may be integrated into a processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0077] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0078] It should be noted that, for the convenience of description, the aforementioned method embodiments are all described as a series of action combinations, but those skilled in the art should be aware that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0079] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0080] Parts not involved in the above embodiments are the same as the prior art or can be implemented by using the prior art, and will not be further described here.

[0081] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A current signal compression acquisition and reconstruction method, characterized in that: The method comprises: Using a flexible Blackman window to measure the current signal Perform sliding acquisition to obtain sliding acquisition signal ; Constructing a partially orthogonal Gaussian measurement matrix ; Using the partially orthogonal Gaussian measurement matrix The sliding acquisition signal After compression acquisition, the compressed signal is obtained ; The sliding acquisition signal Perform sparse transformation and solve the frequency domain transformation matrix ; Using the frequency domain transformation matrix and the partially orthogonal Gaussian measurement matrix The perception matrix D is calculated, and then the compression signal is The perception matrix D is solved iteratively using the algorithm for sparse signals ; According to the frequency domain transformation matrix and the sparse signal Calculate the reconstructed data .

2. The current signal compression acquisition and reconstruction method according to claim 1 is characterized in that: The functional expression of the flexible Blackman window is: （1） In formula (1), t is the current signal acquisition time, T The width of the window can be adjusted.

3. The current signal compression acquisition and reconstruction method according to claim 2 is characterized in that: The flexible Blackman window is used to analyze the current signal Perform sliding acquisition to obtain sliding acquisition signal , as follows: （2） （3） in, is the number of discrete data points collected, is the sampling interval, is the pulse function, .

4. The current signal compression acquisition and reconstruction method according to claim 2 is characterized in that: Adjust the window width according to the deviation of the adjacent window reconstruction signal of the current signal T , the deviation of the reconstructed signal of the adjacent window of the current signal for: （4） In formula (4), , Respectively h -1. h windows; When an arc fault occurs, the current harmonic frequency increases, resulting in Increase, then reduce the window width T ; when When it decreases, the window width is expanded T .

5. The current signal compression acquisition and reconstruction method according to claim 4 is characterized in that: The partially orthogonal Gaussian measurement matrix is ​​constructed include: Based on the standard Gaussian measurement matrix with independent and identical distribution, a partial orthogonal Gaussian measurement matrix is ​​constructed ; The standard Gaussian measurement matrix contains random variables The following distribution is satisfied: (5) In formula (5), The mean is 0 and the variance is Gaussian distribution.

6. The current signal compression acquisition and reconstruction method according to claim 5, characterized in that: The standard Gaussian measurement matrix is ​​expressed as the following matrix , of size m×n: (6) Extract Matrix The first k columns of the matrix : (7) Pair Matrix Perform orthogonal processing and select the first column as the orthogonal basis vector , No. j The column needs to subtract its projection onto all previous orthogonal basis vectors and normalize it: (8) Get the orthogonalized matrix: (9) The matrix Before k Replace the column with Q k Matrix, get the partially orthogonal Gaussian matrix : (10)。 7. The current signal compression acquisition and reconstruction method according to claim 6 is characterized in that: In the extraction matrix Before k Columns form a matrix hour, k The value benchmark is 0.2×n. According to formula (4), the signal deviation of the adjacent window reconstruction of the current signal Make dynamic adjustments; when When increasing, decrease k value; when When decreasing, increasing k value; The k value selection range is (0.1~0.3)×n.

8. The current signal compression acquisition and reconstruction method according to claim 1 is characterized in that: According to compressed sensing, the compressed signal The calculation formula is as follows: (11)。 9. The current signal compression acquisition and reconstruction method according to claim 1, characterized in that: The sliding acquisition signal is processed by discrete wavelet transform Perform sparse transformation and solve the frequency domain transformation matrix .

10. The current signal compression acquisition and reconstruction method according to claim 9, characterized in that: The sliding acquisition signal is processed by using discrete wavelet transform Perform sparse transformation and solve the frequency domain transformation matrix ,include: The sliding acquisition signal is analyzed using the Daubechies wavelet basis. conduct N -1 decomposition, get 1 low-frequency component and N -1 high frequency component: (12) In formula (12), is the low frequency component, is the high frequency component, and are the scaling function and wavelet function respectively, j , k are the scale number and translation number respectively; Further transforming equation (12), the sliding acquisition signal The sparse signal under discrete wavelet transform is represented as: (13) In formula (13), is the frequency domain transformation matrix under discrete wavelet transform.

11. The current signal compression acquisition and reconstruction method according to claim 1, characterized in that: The calculation formula of the perception matrix D is: (14) In formula (14), is the frequency domain transformation matrix, is a partially orthogonal Gaussian measurement matrix.

12. The current signal compression acquisition and reconstruction method according to claim 1, characterized in that: According to the compressed signal and the perception matrix D, and the sparse signal is iteratively solved using the compressed sampling matching pursuit reconstruction algorithm CoSaMP-ESP based on optimal sparsity estimation .

13. A current signal compression acquisition and reconstruction method according to claim 12, characterized in that: The compressed signal and the perception matrix D, and the sparse signal is iteratively solved using the compressed sampling matching pursuit reconstruction algorithm CoSaMP-ESP based on optimal sparsity estimation ,include: Input the compressed signal , the perception matrix D and the error threshold , the compressed sampling matching pursuit reconstruction algorithm CoSaMP-ESP based on optimal sparsity estimation includes: ① Initialization, set the index value collection E is an empty set, F is an empty set, the residual v= , sparsity K =5, number of iterations d =0; ② d = d +1, calculate the inner product of the current residual and the perception matrix D, and select the most qualified column of the perception matrix D according to the fuzzy threshold c and store it in the set E : (15) In formula (15), D j is the matrix D j List, v d-1 For the d -1 iteration residual; ③According to the collection E , and get the support set : (16) ④ According to the support set Solve for the least squares solution: (17) In formula (17), , Represents the set in the perception matrix D The corresponding columns, I is the identity matrix; ⑤Cut the support set, Select the one with the largest absolute value K Item Deposit z ( k ) and update the corresponding , and then update the sparsity estimate K : (18) In formula (18), K is the sparsity, For the d The support set of the iteration, h(m) is the compressed data, D is the perception matrix, z ( k ) is the sparse signal to be solved, and an adjustable p The norm controls sparsity; ⑥Update residual and relevance : (19) (20) in, is the residual, is the correlation, representing all columns of the perception matrix D D j The column with the highest correlation with the current residual; ⑦If , then stop the iteration and output the sparse signal , otherwise return to step ② to recalculate the residual.

14. A current signal compression acquisition and reconstruction method according to claim 13, characterized in that: The value range of the fuzzy threshold c is 0.6~0.

8.

15. The current signal compression acquisition and reconstruction method according to claim 13, characterized in that: In formula (18), p The value base is 1, p The value range is 0.3~3.

16. A current signal compression acquisition and reconstruction method according to any one of claims 1 to 15, characterized in that: The reconstructed data The calculation formula is: (21) In formula (21), is the frequency domain transformation matrix, is the sparse signal solved by iteration.

17. A current signal compression acquisition and reconstruction device, characterized in that: The device comprises: Sliding acquisition module for current signals using flexible Blackman windows Perform sliding acquisition to obtain sliding acquisition signal ; Matrix building module for building partially orthogonal Gaussian measurement matrices ; A compression acquisition module is used to utilize the partially orthogonal Gaussian measurement matrix The sliding acquisition signal After compression acquisition, the compressed signal is obtained ; A sparse transformation module is used to collect the sliding signal Perform sparse transformation and solve the frequency domain transformation matrix ; A calculation and solution module is used to use the frequency domain transformation matrix and the partially orthogonal Gaussian measurement matrix The perception matrix D is calculated, and then the compression signal is The perception matrix D is solved iteratively using the algorithm for sparse signals ; A data reconstruction module is used to transform the frequency domain matrix and the sparse signal Calculate the reconstructed data .

18. A computer-readable storage medium storing one or more programs, characterized in that: When the one or more programs are executed, the current signal compression acquisition and reconstruction method described in any one of claims 1 to 16 can be implemented.

19. An electronic device comprising a processor, a communication interface, the computer-readable storage medium of claim 18, and a communication bus; wherein: The processor, the communication interface, and the computer-readable storage medium communicate with each other via a communication bus; characterized in that the processor is used to execute the program stored in the computer-readable storage medium.

Images