High-voltage switch cabinet partial discharge signal denoising method based on singular value decomposition and particle swarm optimization wavelet decomposition

Abstract

The invention relates to the technical field of high-voltage switch cabinet partial discharge detection, and discloses a high-voltage switch cabinet partial discharge signal denoising method based on singular value decomposition and particle swarm optimization wavelet decomposition, and the method comprises the steps: firstly analyzing the characteristic parameters of a narrowband interference signal in a partial discharge signal, and extracting a partial discharge signal without narrowband interference; secondly, for the problem of white noise existing in the partial discharge signals, wavelet decomposition is carried out on the partial discharge signals after narrowband interference is eliminated, the partial discharge signals are converted into sub-band signals of different scales and frequencies, wavelet inverse transformation reconstruction is carried out after threshold processing, and denoised signals are obtained; and finally, optimizing the selection of threshold related parameters by using a particle swarm algorithm, solving the number of particles and a population optimal value by using a fitness function, and continuously iterating to finally obtain the optimal threshold related parameters. According to the method, the denoising effect is improved, and the method can be popularized to denoising of partial discharge signals of other devices in combination with performance characteristic parameters of different devices.

Description

Technical Field

[0001] The invention relates to the technical field of partial discharge detection of high-voltage switch cabinets, and in particular to a method for denoising partial discharge signals of high-voltage switch cabinets based on singular value decomposition and particle swarm optimization wavelet decomposition. Background Art

[0002] The environment in which electrical equipment is located usually contains a variety of electromagnetic interference sources, such as electromagnetic fields, electromagnetic radiation, and power supply interference, which can produce noise effects on partial discharge signals. At the same time, the sensors used to detect partial discharge may also introduce various types of noise, including thermal noise, electronic noise, and mechanical vibration, which will be mixed into the partial discharge signal during signal acquisition and transmission. Therefore, in order to improve the signal-to-noise ratio and detection accuracy of the signal, and improve the quality and reliability of the partial discharge signal, the signal must be denoised. The partial discharge signal is a weak pulse signal, which is easily interfered by noise in a complex electromagnetic environment and may even be masked by noise, resulting in the loss of its expression value. The noise interference on site is mainly composed of periodic narrowband interference and white noise. Periodic narrowband interference is a sinusoidal signal, named because its spectrum shows sharp and narrow peaks. When multiple sinusoidal frequencies are mixed, the so-called mixed narrowband interference is formed. This interference usually comes from high-order harmonics inside the power system, communication signals of relay protection, and radio signals in the atmosphere. White noise is a random signal whose performance in both the time domain and the frequency domain cannot be described by a fixed function. It usually originates from the heating of equipment, the coupling of electrical lines and transformer windings, and the normal operation of semiconductor devices. Observing the waveforms in the time domain and frequency domain, periodic narrowband interference and white noise show significant differences. A single noise reduction technology is difficult to effectively deal with complex electromagnetic interference. Summary of the invention

[0003] The purpose of the present invention is to overcome the problem that the denoising method in the prior art is difficult to suppress periodic narrowband interference and white noise at the same time, which affects the accuracy of subsequent partial discharge diagnosis, and provide a high-voltage switch cabinet partial discharge signal denoising method based on singular value decomposition and particle swarm optimization wavelet decomposition. According to the method described in the present invention, firstly, the narrowband interference is eliminated by improved singular value decomposition, then the white noise is removed by wavelet decomposition, and finally the threshold selection of wavelet decomposition is optimized by particle swarm algorithm, which provides an effective solution for efficiently removing noise from the local discharge signal of the high-voltage switch cabinet.

[0004] In order to achieve the above object, the present invention provides a method for denoising partial discharge signals of high-voltage switch cabinets based on singular value decomposition and particle swarm optimization wavelet decomposition, the method comprising the following steps:

[0005] (1) The collected one-dimensional partial discharge signal is converted into a Hankel matrix, and singular value decomposition (SVD) is performed. The effective singular values ​​are extracted by setting a threshold, and the signal is reconstructed and restored to a one-dimensional signal to remove the narrowband interference signal in the signal;

[0006] (2) Decompose the narrowband interference signal by wavelet and transform it into sub-band signals of different scales and frequencies to eliminate the influence of the noise signal as much as possible and retain the effective signal. After threshold processing, reconstruct the signal by inverse wavelet transform to obtain the denoised signal;

[0007] (3) The particle swarm algorithm is used to optimize the selection of threshold-related parameters, and the fitness function is used to obtain the number of particles and the optimal value of the swarm. After continuous iteration, the optimal threshold-related parameters are finally obtained.

[0008] Preferably, in step (1), characteristic parameters of narrowband interference signals in partial discharge signals are analyzed, and denoising is performed on the characteristics of the narrowband interference signals. In the process of detecting partial discharge of high-voltage switchgear, it is very easy to be affected by various electromagnetic interferences. Therefore, before performing partial discharge fault diagnosis and fault prediction, the partial discharge signal is denoised.

[0009] Preferably, the periodic narrowband interference in the time domain is usually distributed in the form of sine or cosine, with a high probability of occurrence and usually a large amplitude, which seriously interferes with the detection of partial discharge; the periodic narrowband interference mainly comes from the relay protection communication, infinite carrier communication, etc. of the power system; the formula of periodic narrowband interference is as follows:

[0010]

[0011] Among them, A zi is the amplitude of the narrowband interference signal, f zi is the narrowband interference signal frequency, θ zi is the phase of the narrowband interference signal, m z is the number of narrowband interference signals.

[0012] Preferably, the one-dimensional partial discharge signal is converted into a Hankle matrix, and the one-dimensional signal is expressed as:

[0013] F(t)={f (1) ,f (2) ,f (3) ,...,f (n) ,...,f (N)}

[0014] The transformed Hankel matrix is:

[0015]

[0016] Among them, f (n) represents the nth sampling value in the one-dimensional matrix of the partial discharge signal, N represents the number of sampling points in the one-dimensional matrix, m=N-N+1, n=N / 2.

[0017] Preferably, the converted Hankel matrix is ​​decomposed by SVD into an m*m orthogonal matrix U and an n*n orthogonal matrix, where S is a non-negative diagonal matrix, and the formula is as follows:

[0018] H m*n =U m*m *S m*n *V n*n T

[0019] S m*n =diag(δ1,δ2,...,δ s )

[0020] Among them, δ1,δ2,...,δ s are the singular values ​​of the Hankel matrix and are arranged in descending order.

[0021] Preferably, the larger singular value represents the effective part of the original partial discharge signal. The first r singular values ​​are extracted by setting a threshold and the matrix X' is reconstructed as follows:

[0022] X'=U m*r *S r*r *V n*r T

[0023] Restore the matrix X' to a one-dimensional signal, that is, obtain the effective signal X = {x(1), x(2), ..., x(N)}, where:

[0024]

[0025] Preferably, in step (2), characteristic parameters of white noise signals in partial discharge signals are analyzed, and denoising is performed using wavelet transform; the collected partial discharge signals are expressed by the following formula:

[0026] s(t)=f(t)+ε(t)

[0027] Where s(t) is the partial discharge monitoring signal containing noise, f(t) is the effective part of the signal, and ε(t) is the noise signal.

[0028] Perform discrete wavelet transform on the noisy partial discharge signal and get:

[0029] d j,k =u j,k +e j,k

[0030] Among them, d j,k is the wavelet coefficient of each layer after wavelet transform of the noisy signal s(t), u j,k is the wavelet transform coefficient of the effective part of the signal f(t), e j,k is the wavelet transform coefficient of the signal noise part ε(t).

[0031] Preferably, wavelet denoising is used to eliminate the influence of the noise signal ε(t) as much as possible and retain the effective signal f(t);

[0032] First, the signal to be processed is decomposed into sub-band signals of different scales and frequencies through wavelet transform. The decomposed signals are called approximate coefficients and detail coefficients. The approximate coefficients are the low-frequency part of the signal. The approximate coefficients mainly reflect the overall trend of the signal, while the detail coefficients reflect the details and high-frequency components of the signal. Then the threshold is determined. In the wavelet domain, it is usually assumed that the detail coefficients of the signal mainly contain noise, while the approximate coefficients mainly contain the effective information of the signal. The detail coefficients are threshold processed, and the detail coefficients less than a certain threshold are set to zero, thereby suppressing the noise signal. Finally, the signal is reconstructed. After threshold processing, the processed approximate coefficients and detail coefficients are reconstructed through inverse wavelet transform to obtain the denoised signal.

[0033] The threshold is calculated as follows:

[0034] T m =c m *max(d n (j,L))

[0035] Among them, c m is the threshold adjustment factor, d n (j,L) represents the detail coefficient after wavelet decomposition, and L is the number of decomposition levels.

[0036] Preferably, in step (3), to optimize the parameters L, m and c m , we first define an optimization boundary for each parameter, and then generate a set of particles P with random positions and velocities i ; Each particle corresponds to the denoising parameters L, m, c m a selection within the corresponding boundaries;

[0037] Each initial particle P i The fitness function F is used to evaluate its position. The fitness function is as follows:

[0038]

[0039] Where l is the length of the signal, p is the number of delay cycles set, is the k-th delayed autocorrelation coefficient of the noise residual.

[0040] Preferably, after each iteration, the position and velocity of the particle are updated according to its fitness function as follows:

[0041]

[0042]

[0043] Among them, ω represents the inertia factor, r1 and r2 are the individual learning factor and the group learning factor; c1 and c2 are two random numbers; Represents particle P i The individual historical optimal position of Represents the global optimal position of the particle swarm; at the end of the iteration, the two optimal positions remain stable, and the parameter selection at this time is considered to be optimal.

[0044] Compared with the prior art, the present invention has the following advantages:

[0045] (1) A narrowband interference removal method based on singular value decomposition is adopted. By selecting a suitable threshold and then selecting the singular value corresponding to the narrowband interference, the narrowband interference can be removed. Compared with the traditional method, this method is simple to operate, fast in operation, and has strong denoising ability, thus solving the problem of slow algorithm operation speed and unsatisfactory denoising effect.

[0046] (2) The particle swarm optimization algorithm is combined with wavelet threshold denoising to achieve adaptive parameter selection in the wavelet denoising process to obtain the best denoising effect. Compared with the traditional soft and hard threshold functions, the new threshold function method can adapt to various noise conditions more flexibly. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flowchart of the algorithm for removing narrowband noise based on singular value decomposition;

[0048] Figure 2 This is the flow chart of the white noise suppression algorithm based on particle swarm algorithm. DETAILED DESCRIPTION

[0049] The present invention is described in detail below with reference to the accompanying drawings and specific implementation methods. For the narrowband interference in the noisy partial discharge signal, the noisy partial discharge signal is subjected to singular value decomposition, a reasonable threshold is set to suppress the narrowband interference, and a partial discharge signal without narrowband interference is extracted. Figure 1As shown in Figure 1, in order to solve the white noise problem in the partial discharge signal, the partial discharge signal after removing the narrowband interference is transformed into sub-band signals of different scales and frequencies by wavelet decomposition, and then reconstructed by inverse wavelet transform after threshold processing to obtain the denoised signal, and the particle swarm algorithm is used to find the optimal threshold-related parameter selection, as shown in Figure 1. Figure 2 shown.

[0050] Specifically, the method for denoising the local discharge signal of a high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition of the present invention comprises the following steps:

[0051] (1) The collected one-dimensional partial discharge signal is converted into a Hankel matrix, and singular value decomposition (SVD) is performed. The effective singular values ​​are extracted by setting a threshold, and the signal is reconstructed and restored to a one-dimensional signal to remove the narrowband interference signal in the signal;

[0052] (2) Decompose the narrowband interference signal by wavelet and transform it into sub-band signals of different scales and frequencies to eliminate the influence of the noise signal as much as possible and retain the effective signal. After threshold processing, reconstruct the signal by inverse wavelet transform to obtain the denoised signal;

[0053] (3) The particle swarm algorithm is used to optimize the selection of threshold-related parameters, and the fitness function is used to obtain the number of particles and the optimal value of the swarm. After continuous iteration, the optimal threshold-related parameters are finally obtained.

[0054] In step (1), characteristic parameters of narrowband interference signals in partial discharge signals are analyzed, and denoising is performed based on the characteristics of the narrowband interference signals.

[0055] In the process of detecting partial discharge of high-voltage switchgear, it is very easy to be affected by various electromagnetic interferences. Therefore, before performing partial discharge fault diagnosis and fault prediction, the partial discharge signal must be denoised. Periodic narrowband interference in the time domain is usually distributed in the form of sine or cosine, with a high probability of occurrence and usually a large amplitude, which seriously interferes with the detection of partial discharge. Periodic narrowband interference mainly comes from relay protection communication, infinite carrier communication, etc. of the power system. The formula for periodic narrowband interference is as follows:

[0056]

[0057] Among them, A zi is the amplitude of the narrowband interference signal, f zi is the narrowband interference signal frequency, θ zi is the phase of the narrowband interference signal, m z is the number of narrowband interference signals.

[0058] The one-dimensional partial discharge signal is converted into a Hankle matrix. The one-dimensional signal can be expressed as:

[0059] F(t)={f (1) ,f (2) ,f (3) ,...,f (n) ,...,f (N)}

[0060] The transformed Hankel matrix is:

[0061]

[0062] Among them, f (n) represents the nth sampling value in the one-dimensional matrix of the partial discharge signal, N represents the number of sampling points in the one-dimensional matrix, m=N-N+1, n=N / 2.

[0063] Perform SVD decomposition on the transformed Hankel matrix and decompose it into an m*m orthogonal matrix U and an n*n orthogonal matrix. S is a non-negative diagonal matrix. The formula is as follows:

[0064] H m*n =U m*m *S m*n *V n*n T

[0065] S m*n =diag(δ1,δ2,...,δ s )

[0066] Among them, δ1,δ2,...,δ s are the singular values ​​of the Hankel matrix and are arranged in descending order.

[0067] The larger singular value can represent the effective part of the original partial discharge signal. By setting the threshold, the first r singular values ​​are extracted and the matrix X' is reconstructed. The formula is as follows:

[0068] X'=U m*r *S r*r *V n*r T

[0069] Restoring the matrix X' to a one-dimensional signal, we can get the effective signal X = {x(1), x(2), ..., x(N)}, where:

[0070]

[0071] In step (2), the characteristic parameters of the white noise signal in the partial discharge signal are analyzed and denoising is performed using wavelet transform.

[0072] The collected partial discharge signal can be expressed by the following formula:

[0073] s(t)=f(t)+ε(t)

[0074] Where s(t) is the partial discharge monitoring signal containing noise, f(t) is the effective part of the signal, and ε(t) is the noise signal.

[0075] By performing discrete wavelet transform on the noisy partial discharge signal, we can get:

[0076] d j,k =u j,k +e j,k

[0077] Among them, d j,k is the wavelet coefficient of each layer after wavelet transform of the noisy signal s(t), u j,k is the wavelet transform coefficient of the effective part of the signal f(t), e j,k is the wavelet transform coefficient of the signal noise part ε(t).

[0078] Wavelet denoising is used to eliminate the influence of the noise signal ε(t) as much as possible and retain the effective signal f(t). First, the signal to be processed is decomposed into sub-band signals of different scales and frequencies through wavelet transform. The decomposed signals are called approximate coefficients and detail coefficients. The approximate coefficients are the low-frequency part of the signal, the approximate coefficients mainly reflect the overall trend of the signal, and the detail coefficients reflect the details and high-frequency components of the signal. Then the threshold is determined. In the wavelet domain, it is usually assumed that the detail coefficients of the signal mainly contain noise, while the approximate coefficients mainly contain the effective information of the signal. Therefore, the detail coefficients can be thresholded and the detail coefficients less than a certain threshold are set to zero, thereby suppressing the noise signal. The threshold calculation method is as follows:

[0079] T m =c m *max(d n (j,L))

[0080] Among them, c m is the threshold adjustment factor, d n (j,L) represents the detail coefficient after wavelet decomposition, and L is the number of decomposition levels.

[0081] Finally, the signal is reconstructed. After threshold processing, the processed approximate coefficients and detail coefficients are reconstructed through inverse wavelet transform to obtain the denoised signal.

[0082] In step (3), the particle swarm algorithm is used to optimize and iterate the parameters to find the optimal particle swarm parameter set.

[0083] To optimize the parameters L, m and cm , we first define an optimization boundary for each parameter, and then generate a set of particles P with random positions and velocities i Each particle corresponds to the denoising parameters L, m, c m A selection within the corresponding boundary. Each initial particle P i It is evaluated using the fitness function F according to its position.

[0084] The fitness function is as follows:

[0085]

[0086] Where l is the length of the signal, p is the number of delay cycles set, is the k-th delayed autocorrelation coefficient of the noise residual.

[0087] After each iteration, the position and velocity of the particle are updated according to its fitness function as follows:

[0088]

[0089]

[0090] Among them, ω represents the inertia factor, r1 and r2 are the individual learning factor and the group learning factor. c1 and c2 are two random numbers. Represents particle P i The individual historical optimal position of Represents the global optimal position of the particle swarm. At the end of the iteration, the two optimal positions remain stable, and the parameter selection at this time is considered to be optimal.

[0091] In summary, the present invention proposes a method for denoising the local discharge signal of a high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition. First, the characteristic parameters of the narrowband interference signal in the local discharge signal are analyzed to extract the local discharge signal without narrowband interference; secondly, the local discharge signal after removing the narrowband interference is subjected to wavelet decomposition, and the inverse wavelet transform is reconstructed after threshold processing to obtain the denoised signal; finally, the particle swarm algorithm is used to optimize the selection of threshold-related parameters, and the number of particles and the optimal value of the group are obtained by the fitness function, and the optimal threshold-related parameters are obtained by continuous iteration. The present invention not only improves the denoising effect compared with the traditional method, but can also be combined with the performance characteristic parameters of different devices to be extended to the denoising of local discharge signals of other devices.

[0092] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited thereto. Within the technical concept of the present invention, the technical solution of the present invention can be subjected to a variety of simple modifications, including the combination of various technical features in any other suitable manner, and these simple modifications and combinations should also be regarded as the contents disclosed by the present invention and belong to the protection scope of the present invention.

Claims

1. A method for denoising partial discharge signals of high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition, characterized in that: The method comprises the following steps: (1) The collected one-dimensional partial discharge signal is converted into a Hankel matrix, and singular value decomposition (SVD) is performed. The effective singular values ​​are extracted by setting a threshold, and the signal is reconstructed and restored to a one-dimensional signal to remove the narrowband interference signal in the signal; (2) Decompose the narrowband interference signal by wavelet and transform it into sub-band signals of different scales and frequencies to eliminate the influence of the noise signal as much as possible and retain the effective signal. After threshold processing, reconstruct the signal by inverse wavelet transform to obtain the denoised signal; (3) The particle swarm algorithm is used to optimize the selection of threshold-related parameters, and the fitness function is used to obtain the number of particles and the optimal value of the swarm. After continuous iteration, the optimal threshold-related parameters are finally obtained.

2. The method for denoising the local discharge signal of a high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 1 is characterized in that: In step (1), characteristic parameters of narrowband interference signals in partial discharge signals are analyzed, and denoising is performed on the characteristics of the narrowband interference signals. In the process of detecting partial discharge of high-voltage switchgear, it is very easy to be affected by various electromagnetic interferences. Therefore, before performing partial discharge fault diagnosis and fault prediction, the partial discharge signal is denoised.

3. The method for denoising the local discharge signal of a high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 2 is characterized in that: In the time domain, periodic narrowband interference is usually distributed in the form of sine or cosine, with a high probability of occurrence and usually a large amplitude, which seriously interferes with the detection of partial discharge. Periodic narrowband interference mainly comes from relay protection communication, infinite carrier communication, etc. of the power system. The formula for periodic narrowband interference is as follows: Among them, A zi is the amplitude of the narrowband interference signal, f zi is the narrowband interference signal frequency, θ zi is the phase of the narrowband interference signal, m z is the number of narrowband interference signals.

4. The method for denoising the local discharge signal of a high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 3 is characterized in that: The one-dimensional partial discharge signal is converted into a Hankle matrix, and the one-dimensional signal is expressed as: F(t)={f (1) ,f (2) ,f (3) ,...,f (n) ,...,f (N) } The transformed Hankel matrix is: Among them, f (n) represents the nth sampling value in the one-dimensional matrix of the partial discharge signal, N represents the number of sampling points in the one-dimensional matrix, m=N-N+1, n=N / 2.

5. The method for denoising partial discharge signals of high-voltage switch cabinets based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 4 is characterized in that: Perform SVD decomposition on the transformed Hankel matrix and decompose it into an m*m orthogonal matrix U and an n*n orthogonal matrix. S is a non-negative diagonal matrix. The formula is as follows: H m*n =U m*m *S m*n *V n*n T S m*n =diag(δ1,δ2,...,δ s ) Among them, δ1,δ2,...,δ s are the singular values ​​of the Hankel matrix and are arranged in descending order.

6. The method for denoising partial discharge signals of high-voltage switch cabinets based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 5 is characterized in that: The larger singular value represents the effective part of the original partial discharge signal. The first r singular values ​​are extracted by setting the threshold and the matrix X' is reconstructed as follows: X'=U m*r *S r*r *V n*r T Restore the matrix X' to a one-dimensional signal, that is, obtain the effective signal X = {x(1), x(2), ..., x(N)}, where:

7. The method for denoising partial discharge signals of high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 1 is characterized in that: In step (2), the characteristic parameters of the white noise signal in the partial discharge signal are analyzed, and denoising is performed using wavelet transform; the collected partial discharge signal is expressed by the following formula: s(t)=f(t)+ε(t) Where s(t) is the partial discharge monitoring signal containing noise, f(t) is the effective part of the signal, and ε(t) is the noise signal; Perform discrete wavelet transform on the noisy partial discharge signal and get: d j,k =u j,k +e j,k Among them, d j,k is the wavelet coefficient of each layer after wavelet transform of the noisy signal s(t), u j,k is the wavelet transform coefficient of the effective part of the signal f(t), e j,k is the wavelet transform coefficient of the signal noise part ε(t).

8. The method for denoising partial discharge signals of high-voltage switch cabinets based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 7 is characterized in that: Use wavelet denoising to eliminate the influence of noise signal ε(t) as much as possible and retain the effective signal f(t); First, the signal to be processed is decomposed into sub-band signals of different scales and frequencies through wavelet transform. The decomposed signals are called approximate coefficients and detail coefficients. The approximate coefficients are the low-frequency part of the signal. The approximate coefficients mainly reflect the overall trend of the signal, while the detail coefficients reflect the details and high-frequency components of the signal. Then the threshold is determined. In the wavelet domain, it is usually assumed that the detail coefficients of the signal mainly contain noise, while the approximate coefficients mainly contain the effective information of the signal. The detail coefficients are threshold processed, and the detail coefficients less than a certain threshold are set to zero, thereby suppressing the noise signal. Finally, the signal is reconstructed. After threshold processing, the processed approximate coefficients and detail coefficients are reconstructed through inverse wavelet transform to obtain the denoised signal. The threshold is calculated as follows: T m =c m *max(d n (j,L)) Among them, c m is the threshold adjustment factor, d n (j,L) represents the detail coefficient after wavelet decomposition, and L is the number of decomposition levels.

9. The method for denoising partial discharge signals of high-voltage switch cabinets based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 1 is characterized in that: In step (3), to optimize the parameters L, m and c m , we first define an optimization boundary for each parameter, and then generate a set of particles P with random positions and velocities i ; Each particle corresponds to the denoising parameters L, m, c m a selection within the corresponding boundaries; Each initial particle P i The fitness function F is used to evaluate its position. The fitness function is as follows: Where l is the length of the signal, p is the number of delay cycles set, is the k-th delayed autocorrelation coefficient of the noise residual.

10. The method for denoising partial discharge signals of high-voltage switch cabinet based on singular value decomposition and particle swarm optimization wavelet decomposition according to claim 9 is characterized in that: After each iteration, the position and velocity of the particle are updated according to its fitness function as follows: Among them, ω represents the inertia factor, r1 and r2 are the individual learning factor and the group learning factor; c1 and c2 are two random numbers; Represents particle P i The individual historical optimal position of Represents the global optimal position of the particle swarm; at the end of the iteration, the two optimal positions remain stable, and the parameter selection at this time is considered to be optimal.

Images