Improved CEEMDAN-based power distribution network fault characteristic quantity extraction method and device, and storage medium
Through the improved CEEMDAN and S-G filtering methods, the problem of modal aliasing and white noise residue in traditional EMD algorithms is solved, and the accurate extraction and accurate identification of fault feature quantities are achieved.
Patent Information
- Application Number
- CN202510374929.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-27
- Publication Date
- 2025-07-11
AI Technical Summary
Traditional EMD algorithms have modal aliasing and white noise residue problems in fault feature extraction, resulting in inaccurate extraction of fault feature quantities, affecting the accuracy of fault identification.
The improved CEEMDAN method is used to decompose and denoise the initial fault signal. Combined with S-G filtering, the IMF component with a correlation coefficient greater than 0.5 is retained through correlation analysis, and the CEEMDAN energy ratio and energy entropy are extracted as fault characteristic quantities.
The accuracy and recognition accuracy of fault characteristics are improved, and the noise reduction effect is better than existing methods, ensuring accurate identification of fault types.
Smart Images
Figure CN120296386A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of transmission line fault signal processing, and specifically to a method, device, and storage medium for extracting fault characteristic quantities of a distribution network. Background Art
[0002] In the process of transmission line fault identification, it is necessary to extract fault characteristics. The traditional EMD algorithm has problems such as mode mixing and white noise residue. Although EEMD suppresses the phenomenon of mode mixing, this method requires multiple ensemble averaging coefficients during the decomposition process, resulting in a long calculation time and a large reconstruction error. Therefore, it is difficult to obtain accurate fault characteristics during the extraction of fault characteristic quantities, and the accuracy of subsequent fault identification cannot be guaranteed. Summary of the Invention
[0003] To solve the above problems, the present invention provides a method, device, and storage medium for extracting fault characteristic quantities of a distribution network that can accurately extract fault characteristics.
[0004] To achieve the above object, the present invention is implemented through the following technical solutions:
[0005] The present invention is a method for extracting fault characteristic quantities of a distribution network, including the following steps:
[0006] Step 1, collect the three-phase fault current of the transmission line, and preprocess the fault zero-sequence current of the transmission line to obtain an initial fault signal;
[0007] Step 2, use the improved CEEMDAN to decompose and denoise the initial fault signal to obtain multiple IMF components;
[0008] Step 3, extract the CEEMDAN energy ratio and CEEMDAN energy entropy from each IMF component to complete the extraction of each fault characteristic quantity.
[0009] A further improvement of the present invention lies in: the specific steps of Step 2 are:
[0010] Step 2.1, decompose the initial fault zero-sequence current signal by the CEEMDAN method to obtain n f IMF components, which are respectively and the residual component
[0011] Step 2.2, perform a correlation analysis on the f IMF components obtained by decomposing in Step 2.1 and the initial fault zero-sequence current signal, and retain the f IMF components with a correlation coefficient greater than 0.5, and discard the f IMF components without fault information to obtain m f IMF components;
[0012] Step 2.3, filter the noisy f IMF component by the S-G method to achieve noise reduction of the fault signal. The f IMF component after noise reduction is denoted as IMF l , where l ∈ [1, 2,..., m].
[0013] A further improvement of the present invention lies in that the calculation expression of the correlation coefficient used in the correlation analysis in Step 2.2 is as follows:
[0014]
[0015] where R is the correlation coefficient, X is the initial fault zero-sequence current signal, n is the order of the f IMF component, represents the i-th order f IMF component, where i ∈ [1, 2,..., n].
[0016] A further improvement of the present invention lies in that the specific operation of Step 2.3 is as follows:
[0017] Let r and w be the width of the noise reduction window and the length of the sliding array respectively, where
[0018] w = 2r + 1
[0019] The data points x = (-r, -r + 1, 0, 1, r, -1, -r) in the sliding window are fitted by a high-order polynomial. The expression of the high-order polynomial is:
[0020]
[0021] where q is the degree of the polynomial; a g is the coefficient of the g-th data; x g is the g-th data point in the window, where g ∈ [1, 2,..., q - 1];
[0022] Each movement of the window gives w equations:
[0023]
[0024] In the formula, r is the width of the noise reduction window; q is the degree of the polynomial; e is the least squares fitting residual; f is the f IMF component before noise reduction; Y is the first term in the above formula, A is the second term in the above formula, and F is the third term in the above formula;
[0025] f = YA + F
[0026] The least squares solution of A is:
[0027] A′ = (Y T Y)-1 Y T f
[0028] Wherein, A′ is the least squares solution of A, Y T is the transpose of Y, and f is f before noise reduction IMF component, and after noise reduction, it is:[[]]
[0029] f′ = YA′ = Y(Y T Y)[[]] -1 Y T f.[[]]
[0030] A further improvement of the present invention lies in: The specific operation steps of step 3 are:[[]]
[0031] Step 3.1, extract the CEEMDAN energy ratio, specifically:[[]]
[0032] The energy of CEEMDAN at the sampling node k is:[[]]
[0033]
[0034] The total energy value of the l-th order IMF component is:[[]]
[0035]
[0036] Normalize the energy value of each node and the total energy value of this layer, and the CEEMDAN energy ratio is:[[]]
[0037]
[0038] Wherein, A lk is the amplitude of the IMF l at node k, m is the total order of the IMF components, l ∈ [1, 2,..., m], k is the sampling node in the IMF l component, and i is the order of the f IMF component, i ∈ [1, 2,..., n];[[]]
[0039] Step 3.2, extract the CEEMDAN energy entropy, specifically:[[]]
[0040] The total energy of each IMF component after noise reduction is:[[]]
[0041]
[0042] Wherein, m is the total order of the IMF, and the CEEMDAN energy entropy is:[[]]
[0043]
[0044] Wherein, p l = E l / E represents the ratio of the l-th order IMF component to the total energy of the signal.
[0045] The electronic device of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method as described above are implemented.
[0046] The computer-readable storage medium of the present invention stores a computer program. When the computer program is executed by a processor, the steps of the method as described above are implemented.
[0047] The beneficial effects of the present invention are as follows: Based on the CEEMDAN method, the present invention combines the S-G filtering and noise reduction method to perform noise reduction processing on the faulty zero-sequence current signal, improving the accuracy of fault feature extraction. By combining the energy ratio and energy entropy as the fault feature quantity, the present invention can greatly improve the accuracy of fault type recognition. Description of the Drawings
[0048] Figure 1 is a flowchart for extracting the fault feature quantity of the present invention;
[0049] Figure 2 is a schematic diagram of the fault signal noise reduction process of the present invention;
[0050] Figure 3 is a schematic diagram of the noisy fault signal in the embodiment of the present invention;
[0051] Figure 4 are the IMFs obtained by performing CEEMDAN decomposition on the noisy fault signal in the embodiment of the present invention. Detailed Embodiments
[0052] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the following further elaborates on the present invention in conjunction with the drawings and embodiments. It should be noted that the embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0053] It should also be understood that the following embodiments are only used to further illustrate the present invention and cannot be construed as limiting the protection scope of the present invention. Non-essential improvements and adjustments made by those skilled in the art based on the above content of the present invention fall within the protection scope of the present invention. The parameter settings in the following specific embodiments only represent a feasible example, and those skilled in the art can make targeted modifications according to specific business scenarios.
[0054] As Figure 1 shown, the method for extracting the fault feature quantity of the distribution network of the present invention includes the following steps:
[0055] Step 1, collect the three-phase fault currents of the transmission line, and synthesize the three-phase fault currents of the transmission line to obtain an initial fault zero-sequence current signal, where,
[0056]
[0057] where, I0 is the initial fault zero-sequence current signal, I a 、I b 、I c are the a, b, and c phase currents at the fault point of the transmission line respectively.
[0058] Step 2, decompose the fault zero-sequence current by CEEMDAN, calculate the correlation coefficients of each IMF component, screen out the IMF containing fault characteristics, and filter it by S-G. S-G can meet various different noise reduction requirements, and it is compatible with smoothing and edge-preserving noise reduction to improve the signal-to-noise ratio of the fault zero-sequence current waveform.
[0059] As Figure 2 shown, the specific steps are as follows:
[0060] Step 2.1, decompose the initial fault zero-sequence current signal by the CEEMDAN method to obtain n f IMF components which are components and the residual component
[0061] Step 2.2, perform a correlation analysis on the f IMF components decomposed in Step 2.1 and the initial fault zero-sequence current signal, retain the f IMF components with a correlation coefficient greater than 0.5, discard the f IMF components without fault information, and obtain m f IMF components;
[0062] Step 2.3, filter the noisy f IMF components by the S-G method to achieve fault signal noise reduction, and the filtered f IMF components are denoted as IMF l , where l ∈ [1, 2,..., m].
[0063] The calculation expression of the correlation coefficient used in the correlation analysis in Step 2.2 is as follows:
[0064]
[0065] where, R is the correlation coefficient, X is the initial fault zero-sequence current signal, n is the order of the f IMF component, represents the i-th order f IMF component, where, i ∈ [1, 2,..., n].
[0066] Step 2.3 is specifically operated as follows:
[0067] Let r and w be the width of the noise reduction window and the length of the sliding array respectively, where
[0068] w = 2r + 1
[0069] The data points x = (-r, -r + 1, 0, 1, r, -1, -r) in the sliding window are fitted by a high-order polynomial, and the high-order polynomial expression is:
[0070]
[0071] where q is the degree of the polynomial; a g is the coefficient of the g-th data; x g is the g-th data point in the window, where g ∈ [1, 2,..., q - 1];
[0072] Each movement of the window gives w equations:
[0073]
[0074] In the formula, r is the width of the noise reduction window; q is the degree of the polynomial; e is the least squares fitting residual; f is the f IMF component before noise reduction; Y is the first term in the above formula, A is the second term in the above formula, and F is the third term in the above formula;
[0075] f = YA + F
[0076] The least squares solution of A is:
[0077] A′ = (Y T Y) -1 Y T f
[0078] In the formula, A' is the least squares solution of A, Y T is the transpose of Y, f is the f IMF component before noise reduction, and after noise reduction is:
[0079] f′ = YA′ = Y(Y T Y) -1 Y T f.
[0080] In order to verify the effect of the fault signal noise reduction method based on CEEMDAN and S-G, white noise is added to the high-resistance arc grounding fault waveform in the present invention, and the noisy fault signal is as Figure 3 shown. In Figure 3It can be seen that after adding white noise, the smoothness of the signal becomes weaker and a large number of 'glitches' appear. By decomposing the signal using the method proposed in the present invention, the obtained f IMF components are as Figure 4 shown.
[0081] After decomposition by CEEMDAN, 7th-order f IMF components and 1 residual component are obtained. In order to accurately extract the effective f IMF components containing fault information, first calculate the correlation coefficients between each f IMF component and the noise-added signal. The solution results are shown in Table 1;
[0082] Table 1 Correlation coefficients between each f IMF component and the noise-added signal
[0083]
[0084] The f IMF components with R value greater than 0.5 are effective signals. According to Table 1, in the noise-added signal is the effective signal. However, there is noise interference in the effective f IMF components and noise reduction processing is required. In order to verify the superiority of the method proposed in the present invention, the noise reduction results of the method of the present invention are compared with those of several existing methods. The experimental results are shown in Table 2;
[0085] Table 2 Comparison results of noise reduction effects
[0086] indicators CEEMDAN The literature
[20] In this paper SNR 9.326 10.854 11.430 RMSE 2.347 1.823 1.641
[0087] SNR and RMSE are indicators for measuring the signal quality. The larger the SNR value and the smaller the RMSE value, the better the noise reduction effect. It can be seen from Table 2 that the noise reduction effect of the method proposed in the present invention is better than that of the existing methods, proving that the combination of CEEMDAN and S-G has a good noise reduction effect.
[0088] In order to further verify the noise reduction effect of the present invention, comparative experiments are carried out under four signal-to-noise ratios. The experimental results are shown in Table 3;
[0089] Table 3 Comparison results of noise reduction effects under different signal-to-noise ratios
[0090]
[0091] It can be seen from Table 3 that under different signal-to-noise ratios, the noise reduction effect of the present invention is higher than that of the existing noise reduction methods, further verifying the superiority of the method of the present invention.
[0092] Based on the above data, after denoising the noise signal using the fault signal denoising method based on CEEMDAN and S-G, the denoising effect has been significantly improved, and the effect and performance are stable and reliable. The denoising effect is better than the existing technology. Under different signal-to-noise ratios, the method proposed in the present invention performs excellently, ensuring the accurate extraction of subsequent fault features.
[0093] Step 3: Extract the CEEMDAN energy ratio and CEEMDAN energy entropy from each IMF component to complete the extraction of each fault feature quantity.
[0094] Since the distribution of zero-sequence current in different fault types is different, resulting in different IMF components where the fault features are located after CEEMDAN decomposition, the present invention calculates the ratio of the energy of each IMF component to the total energy of the signal as the fault feature quantity to improve the accuracy of fault type recognition. The specific operation is as follows:
[0095] The specific operation steps of Step 3 are as follows:
[0096] Step 3.1: Extract the CEEMDAN energy ratio, specifically:
[0097] The energy of CEEMDAN at sampling node k is:
[0098]
[0099] The total energy value of the l-th order IMF component is:
[0100]
[0101] Normalize the energy value of each node with the total energy value of this layer. The CEEMDAN energy ratio is:
[0102]
[0103] where A lk is the amplitude of the IMF l at node k, m is the total order of the IMF components, l ∈ [1, 2,..., m], k is the sampling node in the IMF l component, i is the order of the f IMF component, i ∈ [1, 2,..., n];
[0104] Step 3.2: Extract the CEEMDAN energy entropy, specifically:
[0105] The total energy of each IMF component after denoising is:
[0106]
[0107] where m is the total order of the IMF, and the CEEMDAN energy entropy is:
[0108]
[0109] wherein, p l = E l / E represents the ratio of the l-th order IMF component to the total energy of the signal.
[0110] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce means for realizing the functions specified in one Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.
[0111] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that realizes the functions specified in one Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.
[0112] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable devices provide steps for realizing the functions specified in one Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.
[0113] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.
[0114] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.
Claims
1. Method for extracting fault characteristic quantities of distribution network, characterized in that: It includes the following steps: Step 1, collect the three-phase fault currents of the transmission line, and synthesize the three-phase fault currents of the transmission line to obtain an initial fault zero-sequence current signal; Step 2, use the improved CEEMDAN to decompose and denoise the initial fault signal to obtain multiple IMF components; Step 3, extract the CEEMDAN energy ratio and CEEMDAN energy entropy from each IMF component to complete the extraction of each fault feature quantity.
2. The method for extracting fault feature quantities of a distribution network according to claim 1, wherein: The specific steps of Step 2 are as follows: Step 2.1, decompose the initial fault zero-sequence current signal by the CEEMDAN method to obtain n f IMF components, and the n f IMF components include the component and the residual component Step 2.2, perform a correlation analysis on the f components obtained by decomposing in Step 2.1 with the initial fault zero-sequence current signal, and retain the f components with a correlation coefficient greater than 0.5, while discarding the f components without fault information, to obtain m f components; IMF IMF IMF IMF Step 2.3, filter the noisy f IMF component by the S-G method to achieve noise reduction of the fault signal. The denoised f IMF component is denoted as IMF l , where l ∈ [1, 2,..., m].
3. The method for extracting fault feature quantities of a distribution network according to claim 2, wherein: The calculation expression of the correlation coefficient used in the correlation analysis in Step 2.2 is as follows: Among them, R is the correlation coefficient, X is the initial zero-sequence current signal of the fault, and n is the number of f IMF component quantities, indicating the i-th order f IMF component, where i ∈ [1, 2,..., n].
4. The extraction method of the fault characteristic quantity of the distribution network according to claim 2, characterized in that: The specific operation of Step 2.3 is: Let r and w be the width of the denoising window and the length of the sliding array respectively, where w＝2r+1 The data points x = (-r, -r + 1, 0, 1, r, -1, -r) in the sliding window are fitted by a high-order polynomial, and the high-order polynomial expression is: where q is the degree of the polynomial; a g is the coefficient of the g-th data; x g is the g-th data point in the window, where g ∈ [1, 2,..., q - 1]; Each movement of the window gives w equations: Where r is the width of the noise reduction window; q is the degree of the polynomial; e is the least squares fitting residual; f is the f before noise reduction; IMF component; Y is the first term in the above formula, A is the second term in the above formula, and F is the third term in the above formula; f = YA + F The least squares solution of A is: A′ = (Y T Y) -1 Y T f where A′ is the least-squares solution of A, Y T is the transpose of Y, and f is the f IMF component before noise reduction, and after noise reduction it is: f′ = YA′ = Y(Y T Y) -1 Y T f。 5. The extraction method of the fault characteristic quantity of the distribution network according to claim 1, characterized in that: The specific operation steps of Step 3 are: Step 3.1, extract the CEEMDAN energy ratio, specifically: The energy of CEEMDAN at the sampling node k is: The total energy value of the l-th order IMF component is: Normalize the energy value of each node and the total energy value of this layer, and the CEEMDAN energy ratio is: where A lk is the amplitude of the IMF at node k, m is the total order of the IMF components, l ∈ [1, 2,..., m], k is the sampling node in the IMF l component, i is the order of the f l component, i ∈ [1, 2,..., n]; IMF Step 3.2, extract the CEEMDAN energy entropy, specifically: The total energy of each IMF component after denoising is: Among them, m is the total order of the IMF, and the CEEMDAN energy entropy is: where p l = E l / E, representing the ratio of the l-th order IMF component to the total energy of the signal.
6. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 5.
7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the method described in any one of claims 1 to 5.