High-voltage circuit breaker voiceprint monitoring signal processing method and device and computer readable medium

Abstract

The invention discloses a high-voltage circuit breaker voiceprint monitoring signal processing method and device and a computer readable medium. Comprising the following steps: preprocessing a collected voiceprint monitoring signal, optimizing the number K of decomposition modes and a penalty parameter alpha in a variational mode decomposition algorithm VMD by adopting a snake-egret optimization algorithm SBOA, and decomposing a plurality of signal components from the voiceprint monitoring signal of the high-voltage circuit breaker by adopting the optimized variational mode decomposition algorithm VMD; kurtosis and correlation coefficient indexes are adopted to select key components for the multiple signal components, Hankel matrix construction is carried out on the key components, a singular value decomposition (SVD) algorithm is adopted to carry out further noise reduction on key modal components, and accurate extraction of the voiceprint monitoring signals of the high-voltage circuit breaker is completed. Compared with the prior art, the interference of environmental noise can be reduced, the definition of voiceprint monitoring signals of the high-voltage circuit breaker is improved, and a more accurate basis is provided for analysis and judgment of the equipment state.

Description

Technical Field

[0001] The present invention relates to the technical field of high-voltage circuit breaker voiceprint monitoring, and in particular to a high-voltage circuit breaker voiceprint monitoring signal processing method, device and computer-readable medium. Background Art

[0002] In actual engineering operation, fault diagnosis of high-voltage circuit breakers mainly relies on regular maintenance and manual inspection, which is inefficient and cannot monitor the status of circuit breakers in real time.

[0003] In recent years, voiceprint monitoring technology has gradually become one of the important tools for circuit breaker fault diagnosis. By collecting and analyzing the voiceprint signals generated by the circuit breaker during operation, the abnormal state of the equipment can be identified. However, limited by the influence of environmental noise and the complex acoustic signal transmission path, the existing voiceprint signal processing methods still have great deficiencies in accuracy and stability. Therefore, developing a method that can effectively reduce noise and accurately extract the characteristics of high-voltage circuit breaker voiceprint signals is of great significance to improving the efficiency of fault monitoring, which can not only significantly reduce maintenance costs, but also improve the safety and reliability of the power system. Summary of the invention

[0004] Purpose of the invention: In response to the problems in the background technology, the present invention discloses a method, device and computer-readable medium for processing voiceprint monitoring signals of high-voltage circuit breaker. The number of decomposed modes K and the penalty parameter α in the variational mode decomposition algorithm VMD are optimized through the snake-heron optimization algorithm SBOA, and the key components of multiple signal components are selected in combination with the kurtosis and correlation coefficient indicators, and the final high-voltage circuit breaker voiceprint monitoring signal is determined after further noise reduction.

[0005] Technical solution: The present invention provides a method for processing a high-voltage circuit breaker voiceprint monitoring signal, comprising the following steps:

[0006] Step 1: Collect the voiceprint monitoring signal of the high-voltage circuit breaker to form a data set;

[0007] Step 2: The snake heron optimization algorithm SBOA is used to optimize the number of decomposed modes K and the penalty parameter α in the variational mode decomposition algorithm VMD. Then, the optimized variational mode decomposition algorithm VMD is used to decompose multiple signal components from the voiceprint monitoring signal of the high-voltage circuit breaker;

[0008] Step 3: Use the kurtosis and correlation coefficient indicators to select key components from multiple signal components, construct the Hankel matrix of the main components, and use the singular value decomposition algorithm SVD to further reduce the noise of the key modal components to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal.

[0009] Furthermore, in step 2, the snake heron optimization algorithm SBOA is used to optimize the number of decomposed modes K and the penalty parameter α in the variational mode decomposition algorithm VMD, and then the optimized VMD algorithm is used to decompose multiple signal components from the high-voltage circuit breaker voiceprint monitoring signal. The operation steps are as follows:

[0010] S201, the number of snake egret populations and the maximum number of iterations in the initial SBOA algorithm, and at the same time determine the VMD algorithm network structure, the network parameters to be optimized and the fitness function, and preset the dimension of the parameters to be optimized in the snake egret optimization algorithm;

[0011] S202, initializing the initial value of the snake heron population, calculating the number of decomposition modes K and the initial value of the penalty parameter α, and assigning the initial values ​​of K and α to the VMD algorithm to calculate the initial value of the fitness function;

[0012] S203, based on K and α calculated in S202, updating the position of the snake heron according to the formula, and recalculating the sizes of K and α;

[0013] S204, bringing the recalculated values ​​of K and α into the VMD algorithm to calculate the fitness value, and sorting them with the previous fitness value;

[0014] S205, repeat the above steps for multiple iterations, and select the number of decomposition modes K and the penalty parameter α corresponding to the optimal fitness.

[0015] Furthermore, the specific process of the snake heron optimization algorithm SBOA is as follows:

[0016] In the initial stage, the position of the snake heron in the search space is randomly initialized using the following formula:

[0017] X i,j =lb j +r×(ub j -lb j ),i=1,2,...,N,j=1,2,...,Dim

[0018] Among them, X i,j represents the value of the i-th snake heron in the j-th dimension, N represents the number of snake herons, Dim is the dimension of the solution space, lb j Represents the global worst solution, ub j is the optimal position, r represents the random number of direction;

[0019] The hunting behavior of snake herons can be divided into three stages: finding prey, consuming prey, and attacking prey. The entire predation process is divided into three equal time intervals, corresponding to the three stages. The modeling of each stage in SBOA is as follows:

[0020] Phase 1: Finding prey: The position update of the snake heron in this phase is mathematically modeled according to the following equation:

[0021]

[0022] Among them, t represents the current number of iterations, T represents the maximum number of iterations, represents the j-th new state of the i-th snake heron in stage 1, X random_1 and X random_2 is a random candidate solution of the first stage iteration, R1 represents a randomly generated array of dimension 1×Dim in the interval [0,1], Represents the fitness value of the objective function update, F i Represents the fitness value before the objective function is updated;

[0023] Phase 2: Consumption of prey: Brownian motion RB is introduced to simulate the process of snake egret consuming the physical strength of prey. The position update of snake egret in this phase is modeled according to the following equation:

[0024] RB=randn(1,Dim)

[0025]

[0026] In the formula, randn(1,Dim) represents an array of dimension 1×Dim randomly generated from a standard normal distribution, X best Indicates the current optimal value;

[0027] Stage 3: Attacking prey: Introduce the Levy flight strategy, denoted as RL, enhance the global search capability of the optimizer to imitate the flying ability of the snake heron, and introduce a nonlinear perturbation factor to improve the algorithm performance. In this stage, the position root of the snake heron is modeled according to the following equation:

[0028]

[0029] There are two strategies for snake herons to avoid predators. One is to blend into the surroundings when the surroundings are conducive to camouflage; otherwise, they choose to fly or walk quickly.

[0030] The formula corresponding to the camouflage strategy is:

[0031]

[0032] in, represents the new candidate solution of stage three, Represents the fitness value of the objective function after the update in stage three;

[0033] When choosing the flight or escape strategy, a dynamic disturbance factor is introduced, corresponding to the following:

[0034]

[0035] J = round(1+rand(1,1))

[0036] In the formula, C1 corresponds to camouflage, C2 corresponds to escape, r = 0.5, R2 represents a randomly generated array of dimension 1×Dim from a normal distribution, X random represents a random candidate solution for the current iteration, and J represents a random selection of an integer 1 or 2.

[0037] Furthermore, the kurtosis and correlation coefficient indicators are used to select key components, and the singular value decomposition algorithm SVD is used to further reduce the noise of the key modal components to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal, which specifically includes the following processes:

[0038] S301, calculating the kurtosis value of each modal component to measure its non-Gaussianity, presetting a kurtosis threshold, and screening preliminary candidate modal components according to the threshold;

[0039] S302, calculating the correlation coefficient between each candidate modal component and the original high-voltage circuit breaker voiceprint monitoring signal, presetting a correlation coefficient threshold, and further screening key modal components;

[0040] S303, using the singular value decomposition algorithm SVD to perform noise reduction processing on the selected key modal components, eliminating redundant noise components, so that the optimized components can better represent the main features of the voiceprint signal;

[0041] S304, reconstruct the key modal components after noise reduction processing to complete the accurate extraction of the high-voltage circuit breaker soundprint monitoring signal.

[0042] Furthermore, the specific process of kurtosis, correlation coefficient and singular value decomposition in step 3 is as follows:

[0043] The calculation formula for the correlation coefficient between sample X and sample Y is:

[0044]

[0045] Where R is the correlation coefficient; Cov(X,Y) is the covariance of sample X and sample Y; is the variance of sample X, The variance of sample Y, sample X and sample Y are modal components;

[0046] The kurtosis calculation formula is:

[0047]

[0048] Where G is the kurtosis value; x Ave is the mean of signal x; σ is the standard deviation of signal x; n represents the number of modal components;

[0049] Perform singular value decomposition on matrix A and get the following formula:

[0050]

[0051] In the formula, u i ∈Q m×1 , v i ∈Q 1×n , i=1,2,3,…,q.

[0052] The present invention also discloses a device based on the above-mentioned high-voltage circuit breaker voiceprint monitoring signal processing method, comprising:

[0053] A signal acquisition module is used to collect the voiceprint monitoring signal of the high-voltage circuit breaker;

[0054] Model optimization module, used to optimize the number of decomposed modes K and penalty parameter α in the variational mode decomposition algorithm VMD using the snake heron optimization algorithm SBOA;

[0055] A signal extraction module is used to decompose multiple signal components from the voiceprint monitoring signal of the high-voltage circuit breaker using an optimized variational mode decomposition algorithm VMD;

[0056] The signal screening and denoising module is used to select key components from multiple signal components using the kurtosis and correlation coefficient indicators, and further denoise the key modal components using the singular value decomposition algorithm SVD to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal.

[0057] The present invention also discloses a computer-readable medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for processing a voiceprint monitoring signal of a high-voltage circuit breaker as described above is implemented.

[0058] Beneficial effects:

[0059] 1. The present invention uses the snake heron optimization algorithm (SBOA) to optimize the variational mode decomposition method VMD, which not only solves the parameter optimization problem of the traditional VMD algorithm, but also solves the problems of mode aliasing, endpoint effect, parameter setting, etc. existing in the traditional time-frequency processing method.

[0060] 2. The present invention utilizes the combination of the kurtosis-correlation coefficient criterion to better extract key information of the high-voltage circuit breaker voiceprint monitoring signal, effectively screen the inherent modal components decomposed by VMD, and has good practicality.

[0061] 3. The present invention adopts Hankel matrix reconstruction and singular value decomposition for noise reduction, which can achieve the effect of removing noise and purifying effective information very well. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of the present invention.

[0063] Figure 2 This is the optimization process of SBOA for VMD of the present invention.

[0064] Figure 3 It is the time-frequency waveform of the original soundprint monitoring signal of the high-voltage circuit breaker of the present invention.

[0065] Figure 4 The present invention optimizes the VMD algorithm to decompose the time-frequency waveform of the voiceprint monitoring signal.

[0066] Figure 5 A comparative decomposition algorithm is used to verify the effectiveness of the proposed algorithm in this invention.

[0067] Figure 6 This is the correlation coefficient and kurtosis index calculation waveform proposed by the present invention.

[0068] Figure 7 This is the sensitive component waveform selected by the present invention.

[0069] Figure 8 This is the voiceprint monitoring signal waveform with further noise reduction by singular value decomposition proposed in the present invention.

[0070] Fig. 9 This is the waveform of the voiceprint monitoring signal after noise reduction after reconstruction in the present invention. DETAILED DESCRIPTION

[0071] Reference Figures 1 to 9 As shown, as a specific solution of this application, it is described as follows:

[0072] The present invention discloses a method, device and computer-readable medium for processing a voiceprint monitoring signal of a high-voltage circuit breaker. The specific process of the method for processing a voiceprint monitoring signal of a high-voltage circuit breaker is as follows:

[0073] Step 1: Collect the voiceprint monitoring signals of high-voltage circuit breakers to form a data set.

[0074] Step 2: Use the snake heron optimization algorithm (SBOA) algorithm to optimize the parameters (K, α) in the variational mode decomposition algorithm (VMD). Then, use the optimized VMD algorithm to decompose multiple signal components from the high-voltage circuit breaker voiceprint monitoring signal.

[0075] Each snake heron represents a candidate solution to an optimization problem. We can evaluate the objective function based on the values ​​proposed by each snake heron for the problem variables. By comparing the obtained objective function values, we can effectively analyze the quality of the corresponding candidate solutions and determine the best candidate solution to the problem.

[0076] In the initial stage, the position of the snake heron in the search space is randomly initialized using the following formula.

[0077] X i,j =lb j +r×(ub j -lb j ),i=1,2,...,N,j=1,2,...,Dim

[0078] Among them, X i，j represents the value of the i-th snake heron in the j-th dimension, N represents the number of snake herons, and Dim is the dimension of the solution space.

[0079] Since the hunting behavior of snake herons is usually divided into three stages: finding prey, consuming prey, and attacking prey, we divide the entire predation process into three equal time intervals, corresponding to the three stages. In SBOA, each stage is modeled as follows:

[0080] Phase 1: Finding prey: The position update of the snake heron in this phase can be mathematically modeled according to the following equation:

[0081]

[0082] Among them, t represents the current number of iterations, T represents the maximum number of iterations, represents the j-th new state of the i-th snake heron in stage 1, X random_1 and X random_2 is a random candidate solution of the first stage iteration, R1 represents a randomly generated array of dimension 1×Dim in the interval [0,1], Represents the fitness value of the objective function update, F i Represents the fitness value before the objective function is updated.

[0083] Phase 2: Consumption of prey: Brownian motion (RB) is introduced to simulate the process of snake egrets consuming the physical strength of their prey. The position update of the snake egret in this phase can be modeled according to the following equation:

[0084] RB=randn(1,Dim)

[0085]

[0086] In the formula, randn(1,Dim) represents an array of dimension 1×Dim randomly generated from a standard normal distribution, X best Indicates the current optimal value.

[0087] Stage 3: Attacking prey: Introduce the Levy flight strategy (denoted as RL) to enhance the global search capability of the optimizer to imitate the flying ability of the snake heron, and introduce a nonlinear perturbation factor to improve the algorithm performance. The position root of the snake heron in this stage can be modeled according to the following equation:

[0088]

[0089] The snake heron's intelligence is reflected in its two strategies for avoiding predators. One is that when the surrounding environment is conducive to camouflage, the snake heron will blend into the surrounding environment; otherwise, it will choose to fly or walk quickly.

[0090] The formula corresponding to the camouflage strategy is:

[0091]

[0092] When choosing a flight or escape strategy, we introduce a dynamic perturbation factor to repeatedly explore and apply the balance, and the corresponding work is as follows:

[0093]

[0094] J = round(1+rand(1,1))

[0095] In the formula, C1 corresponds to camouflage, C2 corresponds to escape, r = 0.5, R2 represents a randomly generated array of dimension 1×Dim from a normal distribution, X random represents a random candidate solution for the current iteration, and J represents a random selection of an integer 1 or 2.

[0096] After completing VMD optimization, the voiceprint data is decomposed by variational mode decomposition (VMD) according to the two parameters found in step 2 to achieve noise reduction, laying the foundation for further processing.

[0097] The specific principle of the variational mode decomposition (VMD) algorithm in step 2 is as follows: The mathematical principle of the VMD algorithm is shown in formula (1):

[0098]

[0099] In the formula, is the partial derivative with time, δ(t) is the impulse function, μ k (t) is the k-order intrinsic modal component (IMF) obtained after the original signal is decomposed by the VMD algorithm, and f k is the center frequency of the original data signal.

[0100] The constraint expressions it satisfies are as follows:

[0101]

[0102] Where s(t) is the input data signal and K is the total number of IMFs.

[0103] VMD abandons the recursive framework of EMD and transforms the signal decomposition problem into the solution of the variational problem. By introducing the penalty factor α and the Lagrangian multiplier λ, the constrained variational problem is further transformed into an unconstrained variational problem. The extended Lagrangian function obtained is:

[0104]

[0105] In the formula, the penalty factor α is introduced to represent the reconstruction accuracy of the signal, λ is the Lagrange multiplier, {μ k} is the kth order modal component obtained after VMD decomposition, {ω k} is {μ k} center frequency.

[0106] Step 3: Considering the redundancy of the acquired modal components, the key components are selected using the kurtosis and correlation coefficient indicators. The Hankel matrix is ​​constructed for the principal components, and then the singular value decomposition algorithm (SVD) is used to further reduce the noise of the key modal components to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal.

[0107] The calculation formula of the correlation coefficient of sample X and sample Y is:

[0108]

[0109] Where R is the correlation coefficient; Cov(X,Y) is the covariance of sample X and sample Y; is the variance of sample X, The variance of the sample Y.

[0110] The kurtosis calculation formula is:

[0111]

[0112] Where K is the kurtosis value; x Ave is the mean of signal x; σ is the standard deviation of signal x.

[0113] Perform singular value decomposition on matrix A and get the following formula:

[0114]

[0115] In the formula, u i ∈Q m×1 , v i ∈Q 1×n , i=1,2,3,…,q.

[0116] In order to verify the effectiveness of the VMD algorithm based on SBOA optimization proposed in this invention, the voiceprint monitoring signal is used for experimental verification, and it is compared with the EMD commonly used in the existing industry and the EEMD and CEEMD algorithms based on the EMD improvement. The experimental results are as follows:

[0117] First, the VMD parameters are optimized using the SBOA algorithm. The optimization process is as follows: Figure 2 As shown in the figure. The results show that the SBOA algorithm converges to the global optimal solution of 7.294 at the 6th iteration, which is better than other optimization algorithms, proving the efficiency of the SBOA algorithm in achieving VMD parameter optimization. The final optimized parameters are the number of modes K = 5 and the penalty factor α = 160.

[0118] Based on the optimized VMD algorithm, the original voiceprint monitoring signal is as follows Figure 3 As shown, it is decomposed into 5 narrowband signal components, and its spectrum diagram is as follows Figure 4 As shown in the figure, it can be seen that there is no characteristic frequency overlap in the component frequency domain diagram, no modal aliasing occurs, and the decomposition process retains the main frequency components of the signal, effectively suppresses weak noise, and reflects excellent decomposition performance.

[0119] Under the same experimental conditions, EMD, EEMD, and CEEMD algorithms were used to decompose the first five components, and the spectrum diagrams were obtained as follows: Figure 5 The results show that within the same time scale, EMD and its improved algorithm have obvious modal aliasing and endpoint effects, with more overlapping characteristic frequencies and poor decomposition effect.

[0120] In summary, through comparison, it can be seen that the VMD algorithm based on SBOA optimization proposed in the present invention has better decomposition performance than the traditional method when processing nonlinear and non-stationary signals, can effectively solve the problem of modal aliasing, and improve the accuracy of signal processing.

[0121] The present invention selects components with large kurtosis values ​​and high correlation coefficients in the inherent modal components as principal components by calculating kurtosis and correlation coefficients; based on the verification results, the kurtosis index and correlation coefficient are calculated to obtain the kurtosis and correlation coefficient calculation curve, such as Figure 6 As shown. Figure 6 The modal components with high correlation coefficient and kurtosis in the actual natural modal components of the optimized VMD decomposition are selected as the main components of the high-voltage circuit breaker voiceprint monitoring signal. Figure 7 shown.

[0122] Reconstruct the waveform, construct the Hankel matrix of the principal component, and denoise the principal component through singular value decomposition (SVD), and reconstruct the time-frequency waveform of the high-voltage circuit breaker voiceprint monitoring signal after denoising; since the high-voltage circuit breaker voiceprint monitoring signal is a one-dimensional signal, singular value decomposition cannot be directly applied, so it is necessary to construct a multi-dimensional matrix and construct the Hankel matrix of the principal component. Perform singular value decomposition (SVD), and the singular values ​​after decomposition will be arranged from large to small. Since the backbone signal is mainly concentrated in the first few larger singular values, the smaller singular values ​​can be set to 0 and then the signal can be reconstructed to achieve the noise reduction goal. Finally, implement singular value decomposition to denoise the principal component, such as Figure 8 As shown, the signal is reconstructed to obtain the time-frequency waveform of the voiceprint signal after noise reduction, as shown Fig. 9 At this point, the main process of the combined processing method of the present invention ends.

[0123] The high-voltage circuit breaker voiceprint monitoring signal processing method of the present invention is also provided in a computer-readable medium, storing a computer program, and when the computer program is executed by a processor, the high-voltage circuit breaker voiceprint monitoring signal processing method as described above is implemented. In addition, the present invention can also be based on the above-mentioned high-voltage circuit breaker voiceprint monitoring signal processing method and can be written into a code to form a modular structure, forming a high-voltage circuit breaker voiceprint monitoring signal processing device, including:

[0124] The signal acquisition module is used to collect the voiceprint monitoring signal of the high-voltage circuit breaker.

[0125] The model optimization module is used to optimize the number of decomposed modes K and the penalty parameter α in the variational mode decomposition algorithm VMD using the snake heron optimization algorithm SBOA.

[0126] The signal extraction module is used to decompose multiple signal components from the voiceprint monitoring signal of the high-voltage circuit breaker by using the optimized variational mode decomposition algorithm VMD.

[0127] The signal screening and denoising module is used to select key components from multiple signal components using the kurtosis and correlation coefficient indicators, and further denoise the key modal components using the singular value decomposition algorithm SVD to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal.

[0128] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for processing voiceprint monitoring signals of a high-voltage circuit breaker, characterized in that: The steps include: Step 1: Collect the voiceprint monitoring signal of the high-voltage circuit breaker to form a data set; Step 2: The snake heron optimization algorithm SBOA is used to optimize the number of decomposed modes K and the penalty parameter α in the variational mode decomposition algorithm VMD. Then, the optimized variational mode decomposition algorithm VMD is used to decompose multiple signal components from the voiceprint monitoring signal of the high-voltage circuit breaker; Step 3: Use the kurtosis and correlation coefficient indicators to select key components from multiple signal components, construct the Hankel matrix for the key components, and use the singular value decomposition algorithm SVD to further reduce the noise of the key modal components to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal.

2. A method for processing high-voltage circuit breaker voiceprint monitoring signals according to claim 1, characterized in that: In step 2, the snake heron optimization algorithm SBOA is used to optimize the number of decomposed modes K and the penalty parameter α in the variational mode decomposition algorithm VMD. Then, the optimized VMD algorithm is used to decompose multiple signal components from the high-voltage circuit breaker voiceprint monitoring signal. The operation steps are as follows: S201, the number of snake egret populations and the maximum number of iterations in the initial SBOA algorithm, and at the same time determine the VMD algorithm network structure, the network parameters to be optimized and the fitness function, and preset the dimension of the parameters to be optimized in the snake egret optimization algorithm; S202, initializing the initial value of the snake heron population, calculating the number of decomposition modes K and the initial value of the penalty parameter α, and assigning the initial values ​​of K and α to the VMD algorithm to calculate the initial value of the fitness function; S203, based on K and α calculated in S202, updating the position of the snake heron according to the formula, and recalculating the sizes of K and α; S204, bringing the recalculated values ​​of K and α into the VMD algorithm to calculate the fitness value, and sorting them with the previous fitness value; S205, repeat the above steps for multiple iterations, and select the number of decomposition modes K and the penalty parameter α corresponding to the optimal fitness.

3. A method for processing high-voltage circuit breaker voiceprint monitoring signals according to claim 2, characterized in that: The specific process of the snake heron optimization algorithm SBOA is as follows: In the initial stage, the position of the snake heron in the search space is randomly initialized using the following formula: X i,j =lb j +r×(ub j -lb j ),i=1,2,...,N,j=1,2,...,Dim Among them, X i,j represents the value of the i-th snake heron in the j-th dimension, N represents the number of snake herons, Dim is the dimension of the solution space, lb j Represents the global worst solution, ub j is the optimal position, r represents the random number of direction; The hunting behavior of snake herons can be divided into three stages: finding prey, consuming prey, and attacking prey. The entire predation process is divided into three equal time intervals, corresponding to the three stages. The modeling of each stage in SBOA is as follows: Phase 1: Finding prey: The position update of the snake heron in this phase is mathematically modeled according to the following equation: Among them, t represents the current number of iterations, T represents the maximum number of iterations, represents the j-th new state of the i-th snake heron in stage 1, X random_1 and X random_2 is the random candidate solution of the first stage iteration, R1 represents a randomly generated array of dimension 1×Dim in the interval [0,1], and F i new,P1 Represents the fitness value of the objective function update, F i Represents the fitness value before the objective function is updated; Phase 2: Consumption of prey: Brownian motion RB is introduced to simulate the process of snake egret consuming the physical strength of prey. The position update of snake egret in this phase is modeled according to the following equation: RB=randn(1,Dim) In the formula, randn(1,Dim) represents an array of dimension 1×Dim randomly generated from a standard normal distribution, X best Indicates the current optimal value; Stage 3: Attacking prey: Introduce the Levy flight strategy, denoted as RL, enhance the global search capability of the optimizer to imitate the flying ability of the snake heron, and introduce a nonlinear perturbation factor to improve the algorithm performance. In this stage, the position root of the snake heron is modeled according to the following equation: There are two strategies for snake herons to avoid predators. One is to blend into the surroundings when the surroundings are conducive to camouflage; otherwise, they choose to fly or walk quickly. The formula corresponding to the camouflage strategy is: in, represents the new candidate solution of stage three, Represents the fitness value of the objective function after the update in stage three; When choosing the flight or escape strategy, a dynamic disturbance factor is introduced, corresponding to the following: J = round(1+rand(1,1)) In the formula, C1 corresponds to camouflage, C2 corresponds to escape, r = 0.5, R2 represents a randomly generated array of dimension 1×Dim from a normal distribution, X random represents a random candidate solution for the current iteration, and J represents a random selection of an integer 1 or 2.

4. A method for processing high-voltage circuit breaker voiceprint monitoring signals according to claim 1, characterized in that: The kurtosis and correlation coefficient indicators are used to select key components, and the singular value decomposition algorithm SVD is used to further reduce the noise of the key modal components to complete the accurate extraction of the high-voltage circuit breaker soundprint monitoring signal, which specifically includes the following processes: S301, calculating the kurtosis value of each modal component to measure its non-Gaussianity, presetting a kurtosis threshold, and screening preliminary candidate modal components according to the threshold; S302, calculating the correlation coefficient between each candidate modal component and the original high-voltage circuit breaker voiceprint monitoring signal, presetting a correlation coefficient threshold, and further screening key modal components; S303, using the singular value decomposition algorithm SVD to perform noise reduction processing on the selected key modal components, eliminating redundant noise components, so that the optimized components can better represent the main features of the voiceprint signal; S304, reconstruct the key modal components after noise reduction processing to complete the accurate extraction of the high-voltage circuit breaker soundprint monitoring signal.

5. A method for processing high-voltage circuit breaker voiceprint monitoring signals according to claim 4, characterized in that: The specific process of kurtosis, correlation coefficient and singular value decomposition in step 3 is as follows: The calculation formula for the correlation coefficient between sample X and sample Y is: Where R is the correlation coefficient; Cov(X,Y) is the covariance of sample X and sample Y; is the variance of sample X, The variance of sample Y, sample X and sample Y are modal components; The kurtosis calculation formula is: Where G is the kurtosis value; x Ave is the mean of signal x; σ is the standard deviation of signal x; n represents the number of modal components; Perform singular value decomposition on matrix A and get the following formula: In the formula, u i ∈Q m×1 , v i ∈Q 1×n , i=1,2,3,…,q.

6. A device based on the high-voltage circuit breaker voiceprint monitoring signal processing method according to any one of claims 1 to 5, characterized in that: include: A signal acquisition module is used to collect the voiceprint monitoring signal of the high-voltage circuit breaker; Model optimization module, used to optimize the number of decomposed modes K and penalty parameter α in the variational mode decomposition algorithm VMD using the snake heron optimization algorithm SBOA; A signal extraction module is used to decompose multiple signal components from the voiceprint monitoring signal of the high-voltage circuit breaker using an optimized variational mode decomposition algorithm VMD; The signal screening and denoising module is used to select key components from multiple signal components using the kurtosis and correlation coefficient indicators, and further denoise the key modal components using the singular value decomposition algorithm SVD to complete the accurate extraction of the high-voltage circuit breaker voiceprint monitoring signal.

7. A computer readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the high-voltage circuit breaker voiceprint monitoring signal processing method as described in any one of claims 1 to 5 is implemented.

Images