Vibration signal reconstruction denoising method for mechanical state evaluation of double-fracture vacuum circuit breaker

Abstract

The embodiment of the invention discloses a vibration signal reconstruction denoising method for mechanical state evaluation of a double-break vacuum circuit breaker, and the method comprises the steps: carrying out the mechanical state detection of the double-break vacuum circuit breaker through an acceleration sensor, and obtaining an original vibration signal containing noise; carrying out ensemble empirical mode decomposition on the noise-containing original vibration signal to obtain a plurality of intrinsic mode function components; performing spatial transformation on an original matrix formed by the plurality of intrinsic mode function components by using a principal component analysis method to obtain a covariance matrix; determining the number of source signals according to the characteristic value of the covariance matrix; constructing a multi-source vibration signal vector according to the number of the source signals; and processing the multi-source vibration signal vector based on a FastICA algorithm to obtain a denoised vibration signal. By means of the method, various kinds of noise in the vibration signals can be effectively removed, and the accuracy of mechanical state recognition and fault diagnosis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vacuum switchgear and state evaluation and fault diagnosis thereof, and in particular to a vibration signal reconstruction and denoising method for mechanical state evaluation of a double-break vacuum circuit breaker. Background Art

[0002] With the development of power systems and technological progress of power equipment, vacuum circuit breakers have taken a dominant position in the medium voltage field with their environmental friendliness, long service life, and easy maintenance. However, the switching of capacitor banks by vacuum circuit breakers is a common problem that needs to be solved at home and abroad, especially the high probability of heavy breakdown of 40.5kV vacuum circuit breakers switching capacitor banks, and the prominent overvoltage problem, which affects the safe and stable operation of the power system. Since there is a saturation effect between the breakdown voltage and gap length of the vacuum gap, in order to make full use of the excellent characteristics of the short vacuum gap, double-break vacuum circuit breakers are an effective way to improve the performance of switching capacitor banks.

[0003] Statistical results show that mechanical failures account for more than half of all failures of high-voltage circuit breakers. Therefore, mechanical status assessment and fault diagnosis of double-break vacuum circuit breakers are of great practical significance for their reliable operation. During the opening and closing process of the circuit breaker, mechanical vibration will occur due to mechanical impact, friction, mechanical force and electromotive force between components. The vibration signal contains rich information, which can reflect various fault characteristics such as tripping failure, jamming, improper component installation position, loose core, broken shaft pin, component deformation, etc., and can be used for mechanical status assessment and fault diagnosis.

[0004] During the opening and closing process, the two vacuum interrupters and their transmission parts of the double-break vacuum circuit breaker will generate multiple vibration sources. The acceleration sensor used to detect the vibration signal is usually installed on the operating mechanism. The single-channel vibration signal obtained by each sensor contains multi-source information. How to analyze and process the signal and realize multi-source signal reconstruction has become the focus of attention. Due to the complex operating environment and the rapid development of communication technology, the obtained vibration signal contains a large amount of environmental white noise, periodic communication noise, and random pulse interference noise generated by the action of the switching device, which seriously affects the detection effect and the accuracy of state assessment and fault diagnosis. Summary of the invention

[0005] Based on this, it is necessary to propose a vibration signal reconstruction and denoising method for mechanical state assessment of double-break vacuum circuit breaker to address the above problems.

[0006] First aspect

[0007] A vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker, the method comprising:

[0008] The mechanical state of the double-break vacuum circuit breaker is detected by using an acceleration sensor to obtain the original vibration signal containing noise;

[0009] Performing collective empirical mode decomposition on the original vibration signal containing noise to obtain multiple intrinsic mode function components;

[0010] Using principal component analysis, spatially transform the original matrix formed by the plurality of intrinsic mode function components to obtain a covariance matrix;

[0011] Determining the number of source signals according to the eigenvalues ​​of the covariance matrix;

[0012] Constructing a multi-source vibration signal vector according to the number of source signals;

[0013] The multi-source vibration signal vector is processed based on the FastICA algorithm to obtain a denoised vibration signal.

[0014] Preferably, performing collective empirical mode decomposition on the original vibration signal containing noise to obtain a plurality of intrinsic mode function components specifically includes:

[0015] Add Gaussian white noise to the original vibration signal containing noise to obtain the target signal;

[0016] Decomposing the target signal using an ensemble empirical mode decomposition method to obtain a decomposition representation of the target signal;

[0017] The Gaussian white noise is updated to Gaussian white noise of different amplitudes, and the above steps are repeated to obtain decomposition representations corresponding to multiple target signals;

[0018] An average operation is performed on the intrinsic mode functions in the decomposition representations corresponding to the multiple target signals to obtain multiple intrinsic mode function components.

[0019] Preferably, the step of decomposing the target signal by using the ensemble empirical mode decomposition method to obtain a decomposition representation of the target signal specifically includes:

[0020] Determine the local maxima and minima of the target signal;

[0021] Based on the local maximum and minimum values, using the spline interpolation method, upper and lower envelopes are obtained;

[0022] Determining an average envelope according to the upper and lower envelopes;

[0023] Subtracting the average envelope from the target signal to obtain a new data sequence;

[0024] Determining whether the new data sequence satisfies an intrinsic mode function condition;

[0025] In the case where the new data sequence satisfies the eigenmode function condition, determining the new data sequence as a first eigenmode function component;

[0026] Subtracting the first intrinsic mode function component from the target signal to obtain a new target signal;

[0027] The above steps are repeated to process the new target signal until all intrinsic mode function components are extracted to obtain a decomposed representation of the target signal.

[0028] Preferably, the satisfying the intrinsic mode function condition specifically includes:

[0029] If the number of extreme points and zero-crossing points in the new data sequence is equal or differs by at most 1, and the average envelope formed by the local maximum envelope and the local minimum envelope at any time is 0, then the intrinsic mode function condition is considered to be met.

[0030] Preferably, the method of using principal component analysis to perform spatial transformation on the original matrix formed by the plurality of intrinsic mode function components to obtain a covariance matrix specifically includes:

[0031] constructing the original matrix using multiple intrinsic mode function components;

[0032] The original matrix is ​​normalized to obtain a normalized matrix;

[0033] The covariance matrix is ​​calculated according to the standardized matrix.

[0034] Preferably, determining the number of source signals according to the eigenvalues ​​of the covariance matrix specifically includes:

[0035] Solve for the eigenvalues ​​of the covariance matrix;

[0036] Sort the eigenvalues ​​in descending order;

[0037] Calculate the cumulative variance contribution rate based on the sorted eigenvalues;

[0038] The number of source signals is determined based on the cumulative variance contribution rate.

[0039] Preferably, constructing a multi-source vibration signal vector according to the number of source signals specifically includes:

[0040] Determine the projection matrix according to the number of source signals;

[0041] The original matrix is ​​multiplied by the projection matrix to construct a multi-source vibration signal vector.

[0042] Preferably, the multi-source vibration signal vector is processed based on the FastICA algorithm to obtain a denoised vibration signal, specifically including:

[0043] Performing a centering operation on the multi-source vibration signal vector to obtain a first vector;

[0044] Performing a whitening operation on the first vector to obtain a second vector;

[0045] Set the number of iterations;

[0046] Randomly select an initialization vector;

[0047] Perform iterative calculation based on the second vector, the number of iterations and the initialization vector;

[0048] Normalize the vector obtained in each iteration to obtain the normalized vector of each iteration;

[0049] Calculate the dot product of the normalized vector of this iteration and the normalized vector of the previous iteration;

[0050] When the dot product is 1, it is considered that the iteration has converged, and the vector obtained by this iteration is multiplied by the multi-source vibration signal vector to obtain a denoised vibration signal.

[0051] Preferably, the performing iterative calculation based on the second vector, the number of iterations and the initialization vector specifically includes:

[0052] Iterative calculation is performed according to the following formula:

[0053] W(p+1)=E[Zg(W T (p)Z)]-E[g'(W T (p)Z)]W(p)

[0054] Where p is the number of iterations, W(·) is the vector obtained in each iteration, E(·) is the mean operation, Z is the second vector, g(·) is the nonlinear function, and W T (·) is the transpose of the vector obtained at each iteration, and g'(·) is the derivative of g(·).

[0055] Second aspect

[0056] A vibration signal reconstruction and denoising system for mechanical state assessment of a double-break vacuum circuit breaker, the system comprising:

[0057] The original vibration signal acquisition module is used to detect the mechanical state of the double-break vacuum circuit breaker using an acceleration sensor to obtain the original vibration signal containing noise;

[0058] An intrinsic mode function component acquisition module is used to perform collective empirical mode decomposition on the original vibration signal containing noise to obtain multiple intrinsic mode function components;

[0059] A covariance matrix acquisition module is used to perform spatial transformation on the original matrix composed of the plurality of intrinsic mode function components using a principal component analysis method to obtain a covariance matrix;

[0060] A source signal quantity determination module, used to determine the source signal quantity according to the eigenvalue of the covariance matrix;

[0061] A multi-source vibration signal vector construction module, used to construct a multi-source vibration signal vector according to the number of source signals;

[0062] The denoised vibration signal acquisition module is used to process the multi-source vibration signal vector based on the FastICA algorithm to obtain a denoised vibration signal.

[0063] The embodiments of the present invention have the following beneficial effects:

[0064] The method provided by the present invention can utilize a single-channel vibration signal to obtain a multi-source vibration signal inside a double-break vacuum circuit breaker, thereby effectively removing various noises in the vibration signal and improving the accuracy of mechanical state recognition and fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] 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 work.

[0066] in:

[0067] Figure 1 A flowchart of a vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker in one embodiment;

[0068] Figure 2 is an original vibration signal containing noise in one embodiment;

[0069] Figure 3 is an eigenmode function component in one embodiment;

[0070] Figure 4 is a flow chart of the FastICA algorithm in one embodiment;

[0071] Figure 5 is a vibration signal after denoising in one embodiment;

[0072] Figure 6 It is a structural block diagram of a vibration signal reconstruction and denoising system for mechanical state assessment of a double-break vacuum circuit breaker in one embodiment. DETAILED DESCRIPTION

[0073] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of 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.

[0074] In view of the problems existing in the prior art, a vibration signal denoising method with good denoising effect and small calculation amount is needed. The present invention proposes a vibration signal reconstruction denoising method for mechanical state assessment of double-break vacuum circuit breakers. The original vibration signal containing noise is subjected to collective empirical mode decomposition to obtain multiple intrinsic mode function components, and then the principal component analysis is used to extract features, so as to estimate the number of source signals and construct multi-source vibration signals, and the FastICA algorithm is used to separate the signals to obtain the denoised vibration signals.

[0075] Figure 1 A flowchart of a vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker in one embodiment. Figure 1 As shown, the method specifically comprises the following steps:

[0076] Step 202: Use an acceleration sensor to detect the mechanical state of the double-break vacuum circuit breaker to obtain an original vibration signal containing noise.

[0077] Figure 2 is an original vibration signal containing noise in one embodiment. Figure 2 As shown in the figure, the original signal is strongly interfered by the periodic narrowband signal and the environmental white noise signal. The method of this embodiment performs a collective empirical mode decomposition on the noisy signal to obtain the intrinsic mode function, and then uses principal component analysis to extract features, estimates the number of source signals and constructs multi-source vibration signals, and uses the FastICA algorithm to separate the signals to obtain the denoised vibration signal, which can effectively remove various noises in the vibration signal in mechanical state detection and improve the accuracy of state recognition and fault diagnosis.

[0078] Step 204 , performing collective empirical mode decomposition on the original vibration signal containing noise to obtain a plurality of intrinsic mode function components.

[0079] The single-channel vibration signal is subjected to ensemble empirical mode decomposition (EEMD) to obtain a finite number of intrinsic mode function (IMF) components.

[0080] Step 204 performs collective empirical mode decomposition on the original vibration signal containing noise to obtain multiple intrinsic mode function components, which specifically includes steps 2041 to 2044:

[0081] Step 2041, adding Gaussian white noise to the original vibration signal containing noise to obtain a target signal.

[0082] Step 2042, decompose the target signal using the ensemble empirical mode decomposition method to obtain a decomposed representation of the target signal.

[0083] In step 2042, the target signal is decomposed by using the ensemble empirical mode decomposition method to obtain a decomposition representation of the target signal, which specifically includes steps 20421 to 20428:

[0084] Step 20421, determine the local maximum and minimum of the target signal.

[0085] Step 20422, based on the local maximum and minimum values, use the spline interpolation method to obtain the upper and lower envelopes.

[0086] Step 20423, determine the average envelope based on the upper and lower envelope lines.

[0087] Step 20424, subtract the average envelope from the target signal to obtain a new data sequence.

[0088] Step 20425, determine whether the new data sequence meets the intrinsic mode function condition.

[0089] Step 20426, when the new data sequence satisfies the eigenmode function condition, the new data sequence is determined as the first eigenmode function component.

[0090] Satisfy the eigenmode function conditions, including:

[0091] If the number of extreme points and zero-crossing points in the new data sequence is equal or differs by at most 1, and the average envelope formed by the local maximum envelope and the local minimum envelope at any time is 0, then the intrinsic mode function condition is considered to be met.

[0092] Step 20427, subtract the first intrinsic mode function component from the target signal to obtain a new target signal.

[0093] Step 20428, repeat the steps to process the new target signal until all the intrinsic mode function components are extracted to obtain the decomposed representation of the target signal.

[0094] Step 2043, updating the Gaussian white noise to Gaussian white noise of different amplitudes, repeating the above steps, and obtaining decomposition representations corresponding to multiple target signals.

[0095] Step 2044, performing an average operation on the intrinsic mode functions in the decomposition representations corresponding to the multiple target signals to obtain multiple intrinsic mode function components.

[0096] For example, (2.1) for the signal to be decomposed x(t), the target signal x1(t) is obtained by adding Gaussian white noise ω(t):

[0097] x1(t)=x(t)+ω(t)

[0098] (2.2) The target signal x1(t) is decomposed using the ensemble empirical mode decomposition method, and the decomposition of the target signal x1(t) is expressed as:

[0099]

[0100] Where imf i (t) is the i-th intrinsic mode function component obtained by the ensemble empirical mode decomposition, r n (t) is the residual value, n is the number of eigenmode function components;

[0101] (2.3) Repeat steps (2.1) and (2.2) m times, adding Gaussian white noise of different amplitudes to the signal to be decomposed x(t) each time. The target signal x obtained in the mth time is m The decomposition of (t) is expressed as:

[0102]

[0103] Number component, r mn (t) is the residual value obtained by repeating steps (2.1) and (2.2) for the mth time.

[0104] (2.4) The eigenmode function sequences in the decomposition representation of the m target signal sequences are averaged and taken as the final result of each eigenmode function component:

[0105]

[0106] Where imf ik (t) is the i-th intrinsic mode function component obtained by repeating steps (2.1) and (2.2) for the kth time.

[0107] The detailed steps of step (2.2) include:

[0108] (2.2.1) For the signal to be decomposed x(t), the spline interpolation method is used to obtain the upper and lower envelopes, and the average envelope ml is obtained;

[0109] (2.2.2) Subtract the average envelope ml from the signal to be decomposed x(t) to obtain a new data sequence h;

[0110] (2.2.3) If the number of extreme points and zero-crossing points in the new data sequence h is equal or differs by at most 1, and the average envelope formed by the local maximum envelope and the local minimum envelope at any time is 0, then it is considered that the intrinsic mode function condition is met, otherwise repeat step (2.2.1) to obtain the first intrinsic mode function component of the signal to be decomposed x(t);

[0111] (2.2.4) Subtract the obtained intrinsic mode function components from the signal to be decomposed x(t), and repeat steps (2.2.1), (2.2.2), and (2.2.3) again until all the intrinsic mode function components are extracted, and the residual value r obtained by subtracting n intrinsic mode function components from the signal to be decomposed x(t) is n (t) is less than a preset amount;

[0112] (2.2.5) The final signal to be decomposed is expressed as:

[0113]

[0114] Averaging the intrinsic mode function components decomposed after each data processing can eliminate the Gaussian white noise originally introduced, and at the same time suppress the modal aliasing effect existing in the ensemble empirical mode decomposition itself, and obtain a more reasonable intrinsic mode function. Figure 3 is an eigenmode function component in one embodiment.

[0115] Step 206 , using principal component analysis, spatially transforms the original matrix composed of multiple intrinsic mode function components to obtain a covariance matrix.

[0116] By analyzing the changing trend of the eigenvalues, when the sum of the current N eigenvalues ​​accounts for more than 98% of the total sum of all eigenvalues, the number of source signals is determined to be N, and a multi-source vibration signal vector X is constructed in a new eigenspace with the number of source signals being N.

[0117] The principal component analysis (PCA) is used to perform spatial transformation on the matrix composed of the intrinsic mode function components, and its eigenvalues ​​are obtained and arranged in order from large to small.

[0118] In step 206, the original matrix composed of multiple intrinsic mode function components is spatially transformed by principal component analysis to obtain a covariance matrix, which specifically includes steps 2061 to 2063:

[0119] Step 2061, constructing an original matrix using multiple intrinsic mode function components.

[0120] Step 2062, normalize the original matrix to obtain a normalized matrix.

[0121] Step 2063, calculating the covariance matrix according to the standardized matrix.

[0122] Step 208: determine the number of source signals according to the eigenvalues ​​of the covariance matrix.

[0123] In step 208, the number of source signals is determined according to the eigenvalues ​​of the covariance matrix, which specifically includes steps 2081 to 2084:

[0124] Step 2081, solving the eigenvalues ​​of the covariance matrix.

[0125] Step 2082, sort the eigenvalues ​​in descending order.

[0126] Step 2083, calculating the cumulative variance contribution rate according to the sorted eigenvalues.

[0127] Step 2084, determining the number of source signals based on the cumulative variance contribution rate.

[0128] Step 210: construct a multi-source vibration signal vector according to the number of source signals.

[0129] In step 210, a multi-source vibration signal vector is constructed according to the number of source signals, which specifically includes steps 2101 to 2102:

[0130] Step 2101, determining a projection matrix according to the number of source signals.

[0131] Step 2102, multiplying the original matrix and the projection matrix to construct a multi-source vibration signal vector.

[0132] For example, (3.1) assumes that the intrinsic mode function obtained by ensemble empirical mode decomposition has n samples, each sample has m feature quantities, and constructs the original matrix:

[0133] X=[X1,X2,…,X m ] n×m

[0134] (3.2) Normalize the original matrix:

[0135]

[0136] In the formula, μ j and σ j are the jth sample X j The mean and variance of

[0137] (3.3) Calculate the covariance matrix XX of the standardized matrix T And solve its eigenvalue λ1≥λ2≥…≥λ m and the corresponding eigenvectors [R1,R2,…,R m ];

[0138] (3.4) Calculate the cumulative variance contribution rate:

[0139]

[0140] When the cumulative contribution rate reaches 85% to 95%, the first k principal components contain most of the information that the m original features can provide, and the number of principal components is k;

[0141] (3.5) The eigenvectors corresponding to the first k eigenvalues ​​form a projection matrix. Multiplying the original matrix with the projection matrix yields the expression matrix in the new space and the corresponding eigenvalues.

[0142] After the fourth eigenvalue, the eigenvalue size is basically stable and close to 0. The first four eigenvalues ​​account for as much as 98% of the total eigenvalues, which can reflect all the information of the signal, and the number of source signals is determined to be 4.

[0143] Step 212: Process the multi-source vibration signal vector based on the FastICA algorithm to obtain a denoised vibration signal.

[0144] The FastICA algorithm based on independent component analysis (ICA) is used to process the constructed multi-source vibration signal vector X and obtain the denoised vibration signal. The vibration signal reconstruction and denoising method for mechanical state assessment of double-break vacuum circuit breaker in this embodiment uses single-channel vibration signal data.

[0145] In step 212, the multi-source vibration signal vector is processed based on the FastICA algorithm to obtain a denoised vibration signal, which specifically includes steps 2121 to 2128:

[0146] Step 2121, performing a centering operation on the multi-source vibration signal vector to obtain a first vector.

[0147] Step 2122, performing a whitening operation on the first vector to obtain a second vector.

[0148] Step 2123, set the number of iterations.

[0149] Step 2124, randomly select an initialization vector.

[0150] Step 2125, performing iterative calculation based on the second vector, the number of iterations and the initialization vector.

[0151] Based on the second vector, the number of iterations and the initialization vector, an iterative calculation is performed, specifically including:

[0152] Iterative calculation is performed according to the following formula:

[0153] W(p+1)=E[Zg(W T (p)Z)]-E[g'(W T (p)Z)]W(p)

[0154] Where p is the number of iterations, W(·) is the vector obtained in each iteration, E(·) is the mean operation, Z is the second vector, g(·) is the nonlinear function, and W T (·) is the transpose of the vector obtained at each iteration, and g'(·) is the derivative of g(·).

[0155] Step 2126, normalize the vector obtained in each iteration to obtain a normalized vector for each iteration.

[0156] Step 2127, calculate the dot product of the normalized vector of this iteration and the normalized vector of the previous iteration.

[0157] Step 2128, when the dot product is 1, it is considered that the iteration has converged, and the vector obtained by this iteration is multiplied by the multi-source vibration signal vector to obtain a denoised vibration signal.

[0158] Exemplarily, (5.1) the constructed multi-source vibration signal vector X is centered, that is, each data in the multi-source vibration signal vector X is subtracted from the mean value E[X] of all its data to obtain a vector X′, and the mean value of all data in the vector X′ is 0, and then the vector X′ is linearly transformed so that its components are uncorrelated and have unit variance, that is, whitened into a new vector Z;

[0159] (5.2) Set the number of iterations p = 0 and randomly select an initial vector W(0);

[0160] (5.3)p=p+1;

[0161] (5.4) Let W(p+1)=E[Zg(W T (p)Z)]-E[g'(W T (p)Z)]W(p), where E[·] is the mean operation; g(·) is a nonlinear function. In the nonlinear function g1(y)=tanh(a1y), g3(y)=y 3 Choose one of them, 1≤a1≤2;

[0162] (5.5) Normalize W(p+1), W * (p+1)=W(p+1) / ||W(p+1)||, if W * (p+1) and W obtained in the previous iteration * If the dot product of (p) is 1, it is considered that W(p+1) has reached convergence. If it does not converge, return to step (5.3);

[0163] (5.6) The denoised vibration signal S = W(p+1)·X is obtained.

[0164] Figure 4 FIG. 4 is a flow chart of the FastICA algorithm in one embodiment. Figure 5 is a vibration signal after denoising in one embodiment. The constructed multi-source vibration signal is subjected to source signal separation, and the correlation coefficient and relative mean square error parameters are introduced to evaluate the final suppression effect of the signal noise. Finally, the denoised vibration signal is obtained by separation. Figure 5 shown.

[0165] The vibration signal reconstruction and denoising method for mechanical status assessment of double-break vacuum circuit breaker in this embodiment fully considers the multi-source vibration information generated by the two vacuum interrupters and their transmission parts, can effectively filter out environmental white noise and periodic communication noise, and can effectively improve the accuracy of vibration signal detection.

[0166] In this embodiment, the mechanical state assessment of the double-break vacuum circuit breaker uses a vibration signal reconstruction and denoising method to perform collective empirical mode decomposition on the single-channel vibration signal, suppresses the modal aliasing effect of the collective empirical mode decomposition, and uses principal component analysis to reduce the dimension of the decomposition result. The number of source signals is estimated based on the size of the eigenvalues ​​after dimensionality reduction and the change in their proportion.

[0167] In this embodiment, the vibration signal reconstruction and denoising method for the mechanical state assessment of the double-break vacuum circuit breaker performs collective empirical mode decomposition of the noisy signal to obtain a finite number of intrinsic mode function components, and then uses principal component analysis to extract features to estimate the number of source signals and construct multi-source vibration signals.

[0168] The vibration signal reconstruction and denoising method for mechanical condition assessment of double-break vacuum circuit breakers in this embodiment uses the FastICA algorithm to separate the source signals of the constructed multi-source vibration signals. Compared with other denoising methods, the denoised signal is closer to the noise-free source signal and has a small amount of calculation, which can better meet the requirements of on-site assessment and diagnosis.

[0169] Figure 6This is a structural block diagram of a vibration signal reconstruction and denoising method system for mechanical state assessment of a double-break vacuum circuit breaker in one embodiment. Figure 6 As shown, the system includes:

[0170] The original vibration signal acquisition module 6100 is used to perform mechanical state detection on the double-break vacuum circuit breaker using an acceleration sensor to obtain an original vibration signal containing noise.

[0171] The intrinsic mode function component acquisition module 6200 is used to perform collective empirical mode decomposition on the original vibration signal containing noise to obtain multiple intrinsic mode function components.

[0172] The covariance matrix acquisition module 6300 is used to perform spatial transformation on the original matrix composed of multiple intrinsic mode function components using the principal component analysis method to obtain the covariance matrix.

[0173] The source signal quantity determination module 6400 is used to determine the source signal quantity according to the eigenvalues ​​of the covariance matrix.

[0174] The multi-source vibration signal vector construction module 6500 is used to construct a multi-source vibration signal vector according to the number of source signals.

[0175] The denoised vibration signal acquisition module 6600 is used to process the multi-source vibration signal vector based on the FastICA algorithm to obtain a denoised vibration signal.

[0176] It should be pointed out here that the description of the above embodiment of the vibration signal reconstruction and denoising system for mechanical state evaluation of double-break vacuum circuit breakers is similar to the above method description and has the same beneficial effects as the method embodiment. For technical details not disclosed in the embodiment of the vibration signal reconstruction and denoising system for mechanical state evaluation of double-break vacuum circuit breakers of the present invention, those skilled in the art should refer to the description of the method embodiment of the present invention for understanding.

[0177] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0178] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0179] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker, characterized in that: The method comprises: The mechanical state of the double-break vacuum circuit breaker is detected by using an acceleration sensor to obtain the original vibration signal containing noise. Performing collective empirical mode decomposition on the original vibration signal containing noise to obtain multiple intrinsic mode function components; Using principal component analysis, spatially transform the original matrix formed by the plurality of intrinsic mode function components to obtain a covariance matrix; Determining the number of source signals according to the eigenvalues ​​of the covariance matrix; Constructing a multi-source vibration signal vector according to the number of source signals; The multi-source vibration signal vector is processed based on the FastICA algorithm to obtain a denoised vibration signal.

2. The vibration signal reconstruction and denoising method for mechanical state assessment of double-break vacuum circuit breaker according to claim 1, characterized in that: The step of performing collective empirical mode decomposition on the original vibration signal containing noise to obtain a plurality of intrinsic mode function components specifically includes: Add Gaussian white noise to the original vibration signal containing noise to obtain the target signal; Decomposing the target signal using an ensemble empirical mode decomposition method to obtain a decomposition representation of the target signal; The Gaussian white noise is updated to Gaussian white noise of different amplitudes, and the above steps are repeated to obtain decomposition representations corresponding to multiple target signals; An average operation is performed on the intrinsic mode functions in the decomposition representations corresponding to the multiple target signals to obtain multiple intrinsic mode function components.

3. The vibration signal reconstruction and denoising method for mechanical state assessment of double-break vacuum circuit breaker according to claim 2, characterized in that: Decomposing the target signal by using the ensemble empirical mode decomposition method to obtain a decomposition representation of the target signal specifically includes: Determine the local maxima and minima of the target signal; Based on the local maximum and minimum values, using the spline interpolation method, upper and lower envelopes are obtained; Determining an average envelope according to the upper and lower envelopes; Subtracting the average envelope from the target signal to obtain a new data sequence; Determining whether the new data sequence satisfies an intrinsic mode function condition; In the case where the new data sequence satisfies the eigenmode function condition, determining the new data sequence as a first eigenmode function component; Subtracting the first intrinsic mode function component from the target signal to obtain a new target signal; Repeat the above steps to process the new target signal until all intrinsic mode function components are extracted to obtain a decomposed representation of the target signal.

4. The vibration signal reconstruction and denoising method for mechanical state assessment of double-break vacuum circuit breaker according to claim 3, characterized in that: The satisfying of the eigenmode function condition specifically includes: If the number of extreme points and zero-crossing points in the new data sequence is equal or differs by at most 1, and the average envelope formed by the local maximum envelope and the local minimum envelope at any time is 0, then the intrinsic mode function condition is considered to be met.

5. The vibration signal reconstruction and denoising method for mechanical state assessment of double-break vacuum circuit breaker according to claim 1, characterized in that: The method of using the principal component analysis method to perform spatial transformation on the original matrix formed by the multiple intrinsic mode function components to obtain a covariance matrix specifically includes: constructing the original matrix using multiple intrinsic mode function components; Standardizing the original matrix to obtain a standardized matrix; The covariance matrix is ​​calculated according to the standardized matrix.

6. The vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker according to claim 5, characterized in that: Determining the number of source signals according to the eigenvalues ​​of the covariance matrix specifically includes: Solve for the eigenvalues ​​of the covariance matrix; Sort the eigenvalues ​​in descending order; Calculate the cumulative variance contribution rate based on the sorted eigenvalues; The number of source signals is determined based on the cumulative variance contribution rate.

7. The vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker according to claim 6, characterized in that: The constructing of a multi-source vibration signal vector according to the number of source signals specifically includes: Determine the projection matrix according to the number of source signals; The original matrix is ​​multiplied by the projection matrix to construct a multi-source vibration signal vector.

8. The vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker according to claim 1, characterized in that: The step of processing the multi-source vibration signal vector based on the FastICA algorithm to obtain a denoised vibration signal specifically includes: Performing a centering operation on the multi-source vibration signal vector to obtain a first vector; Performing a whitening operation on the first vector to obtain a second vector; Set the number of iterations; Randomly select an initialization vector; Perform iterative calculation based on the second vector, the number of iterations and the initialization vector; Normalize the vector obtained in each iteration to obtain the normalized vector of each iteration; Calculate the dot product of the normalized vector of this iteration and the normalized vector of the previous iteration; When the dot product is 1, it is considered that the iteration has converged, and the vector obtained by this iteration is multiplied by the multi-source vibration signal vector to obtain a denoised vibration signal.

9. The vibration signal reconstruction and denoising method for mechanical state assessment of a double-break vacuum circuit breaker according to claim 8, characterized in that: The performing iterative calculation based on the second vector, the number of iterations and the initialization vector specifically includes: Iterative calculation is performed according to the following formula: W(p+1)=E[Zg(W T (p)Z)]-E[g'(W T (p)Z)]W(p) Where p is the number of iterations, W(·) is the vector obtained in each iteration, E(·) is the mean operation, Z is the second vector, g(·) is the nonlinear function, and W T (·) is the transpose of the vector obtained at each iteration, and g'(·) is the derivative of g(·).

10. A vibration signal reconstruction and denoising system for mechanical state assessment of double-break vacuum circuit breaker, characterized in that: The system comprises: The original vibration signal acquisition module is used to detect the mechanical state of the double-break vacuum circuit breaker using an acceleration sensor to obtain the original vibration signal containing noise; An intrinsic mode function component acquisition module is used to perform collective empirical mode decomposition on the original vibration signal containing noise to obtain multiple intrinsic mode function components; A covariance matrix acquisition module is used to perform spatial transformation on the original matrix composed of the plurality of intrinsic mode function components using a principal component analysis method to obtain a covariance matrix; A source signal quantity determination module, used to determine the source signal quantity according to the eigenvalue of the covariance matrix; A multi-source vibration signal vector construction module, used to construct a multi-source vibration signal vector according to the number of source signals; The denoised vibration signal acquisition module is used to process the multi-source vibration signal vector based on the FastICA algorithm to obtain a denoised vibration signal.

Images