Leakage acoustic wave denoising method and system based on wavelet packet threshold and VMD

Abstract

The present invention proposes a method and system for reducing the noise of leakage sound waves based on wavelet packet thresholding and VMD, belonging to the field of filtering and noise reduction. The method comprises the following steps: obtaining a leakage sound wave acquisition signal; screening key factors of the wavelet packet thresholding method based on a high signal-to-noise ratio; performing a multi-scale decomposition of the leakage sound wave acquisition signal based on each factor to obtain a preprocessed signal; introducing adaptive weight factors, local perturbations, and global jump mutation strategies into the original whale algorithm to form an enhanced whale algorithm; using minimum envelope entropy as the fitness function, the enhanced whale algorithm is used to optimize the VMD method to obtain the optimal parameter combination; using the optimal parameter combination to optimize variational mode decomposition, the preprocessed signal is decomposed to obtain intrinsic mode function components; the intrinsic mode function components are screened and reconstructed to achieve noise reduction of the leakage information. By combining the wavelet packet thresholding method with variational mode decomposition, efficient noise reduction and accurate extraction of leakage sound wave features can be achieved, thereby improving the quality of signal analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline leakage signal filtering and noise reduction, and in particular to a leakage sound wave noise reduction method and system based on wavelet packet threshold and VMD. Background Art

[0002] There are many leakage monitoring methods that can be applied to oil and gas pipelines. Among them, the acoustic wave method has many advantages compared with the traditional mass balance method, negative pressure wave method, transient model method, etc.: high sensitivity, high positioning accuracy, low false alarm rate, short detection time, strong adaptability; it measures the slight dynamic pressure changes in the pipeline fluid, which is independent of the absolute value of the pipeline operating pressure; the response frequency is wider, the detection range is wider, etc. However, in the actual operation process, the leakage is inevitably affected by the pipeline operation noise, background noise, instrument and equipment noise or other working condition noise, which makes it difficult to extract effective leakage attenuation characteristics from the pipeline signal. In response to this problem, domestic and foreign scholars have conducted a lot of research. According to the survey results, the current domestic and foreign patents involving oil and gas pipeline leakage acoustic wave noise reduction methods are mainly:

[0003] US patent application US6389881 discloses a pipeline leak detection technology based on acoustic wave technology. This technology uses sensors to collect dynamic pressure data within the pipeline and uses pattern matching filtering technology to process the signal to effectively suppress noise and interference, thereby improving the accuracy of leak location.

[0004] Chinese application ZL2022104774909 discloses a pipeline leak acoustic signal detection method optimized for ICEEMDAN. This technology uses the ICEEMDAN algorithm to decompose the collected acoustic signal, calculate the multi-scale permutation entropy (MPE) of each modal function (IMF), and then reconstructs the IMF that meets the requirements based on the MPE value to obtain a denoised acoustic signal.

[0005] Chinese application ZL2015100201556 discloses a method for locating oil and gas pipeline leaks based on acoustic wave amplitude. This method uses low-frequency acoustic wave amplitudes obtained through wavelet analysis to detect and locate leaks, and proposes a leak location method that does not consider sound speed and time difference.

[0006] Chinese application ZL2017114441573 discloses a wavelet-based method for identifying acoustic signals of multiphase pipeline leakage. This technique uses wavelet packet decomposition to decompose the collected acoustic signals of multiphase pipeline leakage to obtain wavelet signals. The energy extreme value curves of the wavelet signals are calculated for each frequency band. Leakage signals are identified by determining whether the energy extreme value on the energy extreme value curves under the same conditions has a sudden change.

[0007] However, existing technologies only apply a single processing method, namely the wavelet packet threshold method or the variational mode decomposition method, and lack the technology to integrate the two methods. At the same time, the traditional wavelet packet threshold method and the variational mode decomposition method cannot eliminate the influence of the empirical setting parameters. Specifically,

[0008] (1) The wavelet packet threshold method can extract high-frequency signals while extracting low-frequency signals, solving the problem of some features being lost in high-frequency components during wavelet decomposition. At the same time, it can set a threshold for the signal to reduce noise. However, if the key factors of the wavelet packet threshold method—wavelet basis function, number of decomposition levels, threshold rule, and threshold function—are improperly selected, it will directly lead to poor decomposition results.

[0009] (2) The variational mode decomposition method can adaptively realize the frequency domain division of the signal and effectively separate the various components, so it can be applied to the noise reduction of oil and gas pipeline leakage signals. In practical applications, the noise intensity is relatively large, and the single use of the variational mode decomposition method will result in a large computational cost.

[0010] (3) The penalty factor α and the modal decomposition number K in the variational mode decomposition method have the most significant impact on the accuracy of signal decomposition. If the parameters are not preset properly, the target feature information will be masked or lost. At the same time, the most commonly used solution is to use a metaheuristic algorithm based on the component eigenvalue as the criterion for judging the decomposition effect and continuously iterate and optimize to obtain the optimal parameter combination. However, this type of algorithm has many problems of its own. For example, the original whale algorithm has problems with low population richness and premature convergence. Summary of the Invention

[0011] In order to solve the above problems, the present invention proposes a leakage sound wave denoising method and system based on wavelet packet thresholding and VMD. The key factors of the wavelet packet thresholding method are screened based on the high signal-to-noise ratio rule, and the original signal is preprocessed using the wavelet packet thresholding method, which can effectively remove the noise components in the signal while retaining the key signal characteristics; the original whale algorithm is introduced with adaptive weight factors, local perturbation and global jump mutation strategies to form an enhanced whale algorithm, and it is introduced into the variational mode decomposition process. With the minimum envelope entropy as the fitness function, the key parameters such as the number of decomposition layers and the penalty factor are adaptively adjusted to ensure that the variational mode decomposition achieves the best performance.

[0012] In order to achieve the above object, the present invention adopts the following technical solutions:

[0013] In a first aspect, the present invention provides a method for reducing noise of leakage sound waves based on wavelet packet threshold and VMD, comprising:

[0014] Acquire leakage acoustic wave acquisition signals;

[0015] Based on a high signal-to-noise ratio, the key factors of the wavelet packet threshold method are screened, including the optimal wavelet packet basis function, the number of decomposition layers, the threshold function, and the threshold rule. The leakage acoustic wave acquisition signal is decomposed at multiple scales based on the various factors in the wavelet packet threshold method, and the node coefficients of the last layer are screened and reconstructed using the correlation coefficient method to obtain a preprocessed signal.

[0016] Taking the minimum envelope entropy as the fitness function, the enhanced whale algorithm is used to optimize the variational mode decomposition method to obtain the optimal parameter combination;

[0017] The optimal parameter combination is used to construct an optimized variational modal decomposition, and the preprocessed signal is decomposed to obtain the intrinsic modal function components; the intrinsic modal function components are screened and reconstructed using the correlation coefficient method to achieve denoising of leakage information.

[0018] Preferably, the component parameters of the wavelet packet threshold method based on high signal-to-noise ratio screening specifically include:

[0019] Obtain the original sample signal, and perform three rounds of wavelet decomposition on the original sample signal based on different wavelet packet basis functions, decomposition levels, threshold functions and threshold rules according to the control variable concept. The optimal parameters are selected based on the maximum signal-to-noise ratio between the reconstructed signal and the original sample signal.

[0020] Among them, in the first round, the number of decomposition layers, threshold function and threshold rule are fixed, and the initial wavelet packet basis function is screened out based on a variety of wavelet packet basis functions; the second round fixes the threshold function and threshold rule, and the optimal wavelet packet basis function and the optimal number of decomposition layers are screened out based on the initial wavelet packet basis function and different decomposition layers; the third round fixes the optimal wavelet packet basis function and the optimal number of decomposition layers, and the optimal threshold function and the optimal threshold rule are screened out based on a variety of threshold functions and a variety of threshold rules.

[0021] Preferably, the multi-scale decomposition of the leakage acoustic wave acquisition signal based on various parameters in the wavelet packet threshold method, and the screening and reconstruction of the last layer node coefficients using the correlation coefficient method to obtain the preprocessed signal specifically include:

[0022] The optimal wavelet packet basis function, decomposition layer number, threshold function and threshold rule are used to perform multi-scale decomposition on the leakage acoustic wave acquisition signal, and the Pearson correlation coefficient between the coefficient of each node in the last layer and the leakage acoustic wave acquisition signal is calculated.

[0023] Nodes with correlation greater than or equal to a preset value are reconstructed to obtain preprocessed signals.

[0024] Preferably, the method of using the minimum envelope entropy as the fitness function and optimizing the variational mode decomposition method using the enhanced whale algorithm to obtain the optimal parameter combination specifically includes:

[0025] Initialize the enhanced whale algorithm and determine the initial whale individual based on the minimum envelope entropy; each initial whale individual represents a set of parameter combinations ;

[0026] The enhanced whale algorithm is used to update the initial individual whale positions;

[0027] Variational modal decomposition is performed on each whale individual at each updated position, the sum of the envelope entropy of the modal components after decomposition of each whale individual is calculated, and the whale individual with the smallest envelope entropy is recorded; when the maximum number of iterations is reached, the whale individual with the smallest sum of envelope entropy is output as the optimal whale individual; the optimal whale individual represents the optimal parameter combination, including the optimal number of decomposition layers and penalty factor.

[0028] Preferably, the step of initializing the enhanced whale algorithm and determining the initial whale individual based on minimum envelope entropy includes:

[0029] Set the number of whale populations, generate whale individuals, and optimize the target parameters Range restriction; the individual whales represent a group Parameter combination, where is the number of decomposition layers, is the penalty factor;

[0030] Initialize the parameters of the variational mode decomposition method, perform variational mode decomposition on each whale individual, calculate the sum of the envelope entropies of the modal components of the decomposed signal, and record and save the whale individual with the minimum envelope entropy as the initial whale individual.

[0031] Preferably, the use of the enhanced whale algorithm to update the initial individual whale positions specifically includes:

[0032] Generate a random number p, and determine whether the random number p is greater than or equal to 0.5;

[0033] If so, a weight factor is introduced and a spiral update mechanism is selected to update the individual positions of the whales;

[0034] If not, continue to judge Is it greater than 1? If so, a random search mechanism is selected to update the individual positions of the whales. If not, a differential mutation perturbation factor is introduced and a shrinkage and encirclement mechanism is selected to update the individual positions of the whales.

[0035] Preferably, the optimized variational mode decomposition is constructed by using the optimal parameter combination to decompose the preprocessed signal to obtain the intrinsic mode function components; the intrinsic mode function components are screened and reconstructed by using the correlation coefficient method to achieve noise reduction of leakage information, specifically including:

[0036] Inputting the optimal parameter combination into a variational mode decomposition algorithm, decomposing the preprocessed signal, and calculating the Pearson correlation coefficient between each intrinsic mode function component and the leakage acoustic wave acquisition signal;

[0037] The modal function components with correlation greater than or equal to a preset value are reconstructed to obtain the noise-reduced leakage sound wave signal.

[0038] In a second aspect, the present invention provides a leakage acoustic wave noise reduction system based on wavelet packet threshold and VMD, comprising:

[0039] The original signal acquisition module is used to obtain the leakage sound wave acquisition signal;

[0040] The signal preprocessing module is used to screen the key factors of the wavelet packet threshold method based on a high signal-to-noise ratio. The key factors include the optimal wavelet packet basis function, the number of decomposition layers, the threshold function, and the threshold rule. The module performs multi-scale decomposition of the leakage acoustic wave acquisition signal based on the various factors in the wavelet packet threshold method, and uses the correlation coefficient method to screen and reconstruct the node coefficients of the last layer to obtain a preprocessed signal.

[0041] Algorithm optimization module, which uses the minimum envelope entropy as the fitness function and adopts the enhanced whale algorithm to optimize the variational mode decomposition method to obtain the optimal parameter combination;

[0042] The denoising module is used to construct an optimized variational mode decomposition using the optimal parameter combination, decompose the preprocessed signal to obtain intrinsic mode function components; and use the correlation coefficient method to screen and reconstruct the intrinsic mode function components to achieve denoising of leakage information.

[0043] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the leakage acoustic wave denoising method based on wavelet packet threshold and VMD described in the first aspect.

[0044] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the leakage acoustic wave denoising method based on wavelet packet threshold and VMD described in the first aspect are implemented.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. This invention is based on a method that combines the wavelet packet threshold method with optimized variational mode decomposition. It processes target signals with high accuracy, achieving a correlation of 99.9% with the original components. It can estimate and identify oil and gas pipeline leakage signals, providing a new approach to noise reduction for leakage acoustic signals.

[0047] 2. The improved enhanced whale algorithm of this invention better balances global exploration and search capabilities with local development and search capabilities, effectively solving the problems of low swarm richness and premature convergence. At the same time, the enhanced whale algorithm only needs to be applied to one set of parameters to obtain the optimal parameters, which can significantly reduce the time required to find the optimal parameters.

[0048] 3. Compared with the traditional variational mode decomposition method, the variational mode decomposition method after enhanced whale optimization in the present invention can adaptively select the optimal parameter combination, better divide the time domain and frequency domain of the signal, and obtain the effective decomposition components of the leakage acquisition signal;

[0049] 4. The signal preprocessing method based on the wavelet packet threshold method provided by the present invention solves the difficult problems of determining the wavelet packet basis function, the number of decomposition layers, the threshold function and the threshold rule, while freeing up a large amount of data costs for the optimized variational mode decomposition method.

[0050] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their description are used to explain the present invention but do not constitute a limitation of the present invention.

[0052] Figure 1 A main flow chart of a leakage acoustic wave denoising method based on wavelet packet thresholding and VMD provided by an embodiment of the present invention;

[0053] Figure 2 A detailed flow chart of a leakage acoustic wave denoising method based on wavelet packet thresholding and VMD provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0055] Example 1

[0056] like Figure 1 As shown, this embodiment discloses a leakage sound wave denoising method based on wavelet packet threshold and VMD, comprising the following steps:

[0057] S1: Acquire leakage acoustic wave acquisition signal;

[0058] S2: Filtering key factors of the wavelet packet threshold method based on a high signal-to-noise ratio, including the optimal wavelet packet basis function, number of decomposition layers, threshold function, and threshold rule; performing multi-scale decomposition of the leakage acoustic wave acquisition signal based on various factors in the wavelet packet threshold method, and using the correlation coefficient method to filter and reconstruct the node coefficients of the last layer to obtain a preprocessed signal;

[0059] S3: Using the minimum envelope entropy as the fitness function, the enhanced whale algorithm is used to optimize the variational mode decomposition method to obtain the optimal parameter combination;

[0060] S4: Using the optimal parameter combination to construct an optimized variational modal decomposition, decompose the preprocessed signal to obtain intrinsic modal function components; using the correlation coefficient method to screen and reconstruct the intrinsic modal function components to achieve noise reduction of leakage information.

[0061] Next, combine Figure 2 , a leakage sound wave denoising method based on wavelet packet threshold and VMD disclosed in this embodiment is described in detail.

[0062] In S1, a leakage acoustic wave acquisition signal is obtained, which specifically includes: setting an acoustic wave sensor on the measured pipeline, collecting signals from the leakage point through the sensor, and obtaining the leakage acoustic wave acquisition signal.

[0063] In S2, preferably, the key factors of the wavelet packet threshold method are screened based on a high signal-to-noise ratio, the leakage acoustic wave acquisition signal is decomposed at multiple scales based on the various factors in the wavelet packet threshold method, and the last layer node coefficients are screened and reconstructed using the correlation coefficient method to obtain a preprocessed signal; the specific steps are:

[0064] S201: First, obtain the sample original signal;

[0065] S202: Optimizing wavelet packet basis functions: fixing the number of decomposition layers, threshold function, and threshold rule, performing wavelet packet decomposition on the original sample signal using N wavelet packet basis functions, and reconstructing the last layer of low-frequency signal obtained by decomposition to obtain a first reconstructed signal; calculating the signal-to-noise ratio of the first reconstructed signal and the original sample signal, and selecting the first M wavelet packet basis functions as initial wavelet packet basis functions based on the maximum signal-to-noise ratio; wherein 2≤M<N.

[0066] As an implementation method, the number of decomposition layers is fixed to 4, the threshold function is a hard threshold, the threshold rule is heursure, and multiple wavelet packet basis groups of sym, db, coif, fk, bior, and rbio clusters are used for wavelet packet decomposition. The low-frequency signal of the last decomposition layer is reconstructed, and the signal-to-noise ratio of the reconstructed signal to the original signal is calculated. The first five wavelet packet basis functions are selected according to the maximum signal-to-noise ratio rule. The signal-to-noise ratio calculation formula is as follows:

[0067] (1)

[0068] Where: To decompose the power of the low-frequency signal in the last layer, is the power of the original sample signal.

[0069] S203: Optimizing the number of decomposition layers: fixing the threshold function and the threshold rule, and setting the number of decomposition layers to 1 to a layers; using the M initial wavelet packet basis functions, performing wavelet packet decomposition on the original signal of the sample based on the decomposition layers of 1 to a layers, and reconstructing the last layer of low-frequency signal obtained by the decomposition to obtain a second reconstructed signal; calculating the signal-to-noise ratio of the second reconstructed signal and the original signal of the sample, and screening out the optimal wavelet packet basis function and the optimal number of decomposition layers based on the maximum signal-to-noise ratio; wherein a>1.

[0070] As an implementation method, the fixed threshold function is a hard threshold, the threshold rule is heursure, the wavelet basis function selects the 5 wavelet packet basis functions selected in S202, the decomposition layer number is set to 1 to 10 layers, and the 5 wavelet packet basis functions are processed by the wavelet packet threshold method with the decomposition layer number of 1 to 10 layers respectively. The low-frequency signal of the last decomposition layer is reconstructed, and the signal-to-noise ratio of the reconstructed signal to the original signal is calculated. The optimal wavelet packet basis function and decomposition layer number are selected according to the maximum signal-to-noise ratio rule.

[0071] S204: Optimizing threshold functions and threshold rules: setting b threshold functions and c threshold rules; using the optimal wavelet packet basis function and the optimal number of decomposition layers, performing wavelet decomposition on the original sample signal based on different threshold functions and threshold rules, and reconstructing the last layer of low-frequency signal obtained by decomposition to obtain a third reconstructed signal; calculating the signal-to-noise ratio of the third reconstructed signal and the original sample signal, and screening out the optimal threshold function and optimal threshold rule based on the maximum signal-to-noise ratio; wherein b>1, c>1.

[0072] As an implementation method, the optimal wavelet packet basis function and the number of decomposition layers in S203 are fixed, the threshold function selects a soft threshold and a hard threshold respectively, and the threshold rules select four threshold rules: rigrsure, heursure, minimaxi, and sqtwolog respectively. The low-frequency signals of the last layer of decomposition under the two threshold functions and the four threshold rules are reconstructed respectively, and the signal-to-noise ratio of the reconstructed signal to the original signal is calculated, and the threshold function and threshold rule are selected according to the rule with the maximum signal-to-noise ratio.

[0073] Through the above three rounds of optimization process, this embodiment can construct a wavelet packet threshold method based on high signal-to-noise ratio screening. By accurately selecting the wavelet packet basis function, the number of decomposition layers, the threshold function and the threshold rule, the noise component in the signal can be more effectively removed while retaining the useful information in the signal, thereby improving the quality and availability of the signal.

[0074] S205: The parameters of the wavelet packet threshold method are composed of the optimal wavelet packet basis function and decomposition level in S203 and the optimal threshold function and threshold rule in step S204, and the collected signal is decomposed into multiple scales.

[0075] S206: Calculate the Pearson correlation coefficient between the coefficients of each node in the last layer and the leakage acoustic wave acquisition signal. The calculation formula is:

[0076] (2)

[0077] Where, is the node coefficient, Leakage acoustic wave acquisition signal, 、 is the corresponding mathematical expectation, and the value range of the Pearson correlation coefficient is between 0 and 1;

[0078] S207: Reconstruct the nodes whose correlation is greater than or equal to a preset value to obtain a preprocessed signal. The preset value is preferably 0.7.

[0079] In S3, the minimum envelope entropy is used as the fitness function, and the enhanced whale algorithm is used to optimize the variational mode decomposition method to obtain the optimal parameter combination. The specific steps are as follows:

[0080] S301: Set the number of whale populations and optimize the target parameters Perform range restriction and generate initial whale individuals, initialize the variational mode decomposition method parameters. After calculating the variational mode decomposition of each whale individual, sum the envelope entropy of the signal modal components and record the whale individual with the minimum envelope entropy. The envelope entropy calculation formula is:

[0081] (3)

[0082] Where, Represented as preprocessed signal The envelope signal sequence obtained after Hilbert demodulation; yes The normalized form of is the corresponding envelope entropy value.

[0083] S302: Generate a random number P in the range of 0 to 1 and update the individual whale position;

[0084] S3021: If P < 0.5, introduce a weight factor to the original whale algorithm, enter global exploration, and calculate parameters , when | When |>1, the whale enters the random search phase. The random search formula is:

[0085] (4)

[0086] Where: t is the current iteration number; is the currently randomly selected position vector; is the position vector of the individual whale; is the distance between the individual whale and its prey; | | is the absolute value; · is the element-wise multiplication; and The size is controlled by the parameter To decide, w is the weight factor that changes with the number of iterations. The specific calculation is as follows:

[0087] (5)

[0088] (6)

[0089] (7)

[0090] (8)

[0091] Where: and is a random real number in (0, 1); is a factor that decreases gradually from 2 to 0; is the maximum number of group iterations;

[0092] S3022: If P < 0.5, | When |≤1, the original whale algorithm enters the contraction and encirclement phase with the introduction of differential mutation perturbation factors. The formula is as follows:

[0093] (9)

[0094] (10)

[0095] Where: is the currently randomly selected position vector, is the differential variation perturbation factor, F is the variation scale factor;

[0096] S3023: If P ≥ 0.5, the optimization algorithm enters the local optimization stage and changes the position update method of the original whale algorithm from spiral update to Archimedean spiral update. The individual whale moves towards the current best individual whale position. The formula is as follows:

[0097] (11)

[0098] Where: b is a constant coefficient, l It is a random number between [-1, 1].

[0099] In this embodiment, the adaptive selection of the three modes—global exploration, the shrinking and encircling phase (introducing differential mutation), and local optimization—is determined based on the algorithm's current state and parameters (such as the distance between the whale and its prey, the value of the random number P, etc.). This adaptive mechanism enables the algorithm to flexibly adjust its search strategy in different search phases and scenarios, thereby more effectively solving the optimization problem. Therefore, it can be said that the three modes constitute an adaptive search mechanism, forming an enhanced whale algorithm (EWOA) that maintains global search capabilities while also possessing strong local search capabilities.

[0100] Dynamic selection is performed based on the algorithm's operating state and the relationship between the whale and its prey. Global exploration and local optimization are responsible for the algorithm's global and local search capabilities, respectively, while the shrinking and encircling phase enhances population diversity by introducing differential mutation. This design enables the EWOA algorithm to maintain global search capabilities while also possessing strong local search capabilities, effectively solving optimization problems.

[0101] S303: Update the latest whale individual, perform variational modal decomposition on each whale individual, calculate the sum of the envelope entropy of the modal components after decomposition of each whale individual, and record the whale individual with the minimum envelope entropy. When the maximum number of iterations is reached, stop the iteration and output the whale individual with the minimum sum of envelope entropy. The individual contains the optimal number of decomposition layers. K and penalty factor α The variational mode decomposition steps are:

[0102] S3031: Convert the decomposition of the original signal into a variational problem, perform Hilbert transform on the mode, and calculate its analytical signal to obtain a single-sided spectrum. The signal is then shifted to baseband to align with its spectrum by multiplying by a set of exponential factors to adjust to the estimated center frequency:

[0103] (12)

[0104] The square of the L2 norm of the demodulated signal gradient is calculated as an estimate of the bandwidth of each mode, and a Gaussian smoothing exponent is applied. The constrained variational model of VMD is obtained as:

[0105] (13)

[0106] (14)

[0107] Where, represents the original input signal, 、 Represents the decomposition K（The number of decomposition levels in the current whale individual) modal functions and the center frequency corresponding to each mode.

[0108] S3032: By introducing a quadratic penalty factor α and augmented Lagrangian function λ , transforming the constrained variational problem into an unconstrained variational problem, thereby solving the above constrained optimization problem.

[0109] (15)

[0110] S3033: Use the alternating direction multiplier method to find the "saddle point" minimum of the extended Lagrangian expression. The optimal solution is the minimum of each mode. 、 The iteration ends when the following constraints are met:

[0111] (16)

[0112] S3034: Based on the signal frequency domain characteristics, divide the frequency band to achieve adaptive decomposition of the signal;

[0113] (17)

[0114] In S4, the optimal parameter combination is used to construct an optimized variational mode decomposition, and the preprocessed signal is decomposed to obtain the intrinsic mode function components. The intrinsic mode function components are screened and reconstructed using the correlation coefficient method to achieve noise reduction of leakage information. The specific steps are:

[0115] S401: Inputting the optimal parameter combination into a variational mode decomposition algorithm, decomposing the multi-scale acoustic wave signal, and calculating the Pearson correlation coefficient between each intrinsic mode function component and the leakage acoustic wave acquisition signal;

[0116] S402: Reconstruct the modal function components whose correlation is greater than or equal to a preset value to obtain a noise-reduced leakage acoustic wave signal. The preset value is preferably 0.7.

[0117] This embodiment creatively integrates the wavelet packet threshold method and the variational mode decomposition method. First, through the high signal-to-noise ratio screening mechanism of the wavelet packet threshold method, the key parameters of the wavelet packet decomposition are accurately determined, and the multi-scale fine decomposition of the collected signal is realized. The correlation coefficient method is used to screen and reconstruct the nodes with high correlation in the last layer to obtain the preprocessed signal, effectively remove the noise component, and retain the key characteristics of the signal. Subsequently, the enhanced whale algorithm is used to optimize the parameters of the variational mode decomposition, further improving the accuracy and efficiency of the decomposition, making the intrinsic mode function components obtained by decomposition more accurate. Finally, the correlation coefficient method is used to screen and reconstruct the modal components with high correlation, achieving deep noise reduction of the leakage information and significantly improving the quality and analyzability of the signal.

[0118] Example 2

[0119] This embodiment provides a leakage sound wave noise reduction system based on wavelet packet thresholding and VMD, including:

[0120] The original signal acquisition module is used to obtain the leakage sound wave acquisition signal;

[0121] The signal preprocessing module is used to screen the key factors of the wavelet packet threshold method based on a high signal-to-noise ratio. The key factors include the optimal wavelet packet basis function, the number of decomposition layers, the threshold function, and the threshold rule. The module performs multi-scale decomposition of the leakage acoustic wave acquisition signal based on the various factors in the wavelet packet threshold method, and uses the correlation coefficient method to screen and reconstruct the node coefficients of the last layer to obtain a preprocessed signal.

[0122] Algorithm optimization module, which uses the minimum envelope entropy as the fitness function and adopts the enhanced whale algorithm to optimize the variational mode decomposition method to obtain the optimal parameter combination;

[0123] The denoising module is used to construct an optimized variational mode decomposition using the optimal parameter combination, decompose the preprocessed signal to obtain intrinsic mode function components; and use the correlation coefficient method to screen and reconstruct the intrinsic mode function components to achieve denoising of leakage information.

[0124] Example 3

[0125] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps of the leakage acoustic wave denoising method based on wavelet packet thresholding and VMD as described in the first embodiment above are implemented.

[0126] Example 4

[0127] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the leakage acoustic wave denoising method based on wavelet packet threshold and VMD as described in the first embodiment are implemented.

[0128] The steps or modules involved in Examples 2 to 4 above correspond to those in Example 1. For detailed implementations, please refer to the relevant description of Example 1. The term "computer-readable storage medium" should be understood to mean a single medium or multiple media that includes one or more instruction sets; it should also be understood to include any medium that can store, encode, or carry an instruction set for execution by a processor and cause the processor to perform any method of the present invention.

[0129] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for noise reduction of leakage sound waves based on wavelet packet threshold and VMD, characterized in that: include: Acquire leakage acoustic wave acquisition signals; The key factors of the wavelet packet threshold method are screened based on a high signal-to-noise ratio, and the key factors include the optimal wavelet packet basis function, the number of decomposition layers, the threshold function, and the threshold rule; the leakage acoustic wave acquisition signal is decomposed at multiple scales based on the various factors in the wavelet packet threshold method, and the node coefficients of the last layer are screened and reconstructed using the correlation coefficient method to obtain a preprocessed signal; the component parameters of the wavelet packet threshold method are screened based on a high signal-to-noise ratio, specifically as follows: the original sample signal is obtained, and according to the control variable idea, three rounds of wavelet decomposition are performed on the original sample signal based on different wavelet packet basis functions, the number of decomposition layers, the threshold function, and the threshold rule, and the optimal parameters are screened based on the maximum signal-to-noise ratio between the reconstructed signal and the original sample signal; In the first round, the number of decomposition layers, threshold function, and threshold rule are fixed, and N wavelet packet basis functions are used to perform wavelet packet decomposition on the sample original signal. The low-frequency signal of the last layer obtained by decomposition is reconstructed to obtain the first reconstructed signal. The signal-to-noise ratio of the first reconstructed signal and the sample original signal is calculated, and the first M wavelet packet basis functions are selected as the initial wavelet packet basis functions based on the maximum signal-to-noise ratio; where 2≤M<N; In the second round, the threshold function and threshold rule are fixed, and the number of decomposition layers is set to 1 to a layers; the original sample signal is subjected to wavelet packet decomposition based on the decomposition layers 1 to a layers using the M initial wavelet packet basis functions, and the last layer of low-frequency signal obtained by the decomposition is reconstructed to obtain a second reconstructed signal; the signal-to-noise ratio of the second reconstructed signal to the original sample signal is calculated, and the optimal wavelet packet basis function and the optimal number of decomposition layers are screened based on the maximum signal-to-noise ratio; wherein a>1; In the third round, the optimal wavelet packet basis function and the optimal number of decomposition layers are fixed, b threshold functions and c threshold rules are set; the original sample signal is subjected to wavelet decomposition using the optimal wavelet packet basis function and the optimal number of decomposition layers based on different threshold functions and threshold rules, and the last layer of low-frequency signal obtained by decomposition is reconstructed to obtain a third reconstructed signal; the signal-to-noise ratio of the third reconstructed signal to the original sample signal is calculated, and the optimal threshold function and optimal threshold rule are selected based on the maximum signal-to-noise ratio; where b>1 and c>1; Taking the minimum envelope entropy as the fitness function, the enhanced whale algorithm is used to optimize the variational mode decomposition method to obtain the optimal parameter combination; wherein, the enhanced whale algorithm is used to update the initial individual whale positions, specifically including: generating a random number p, and judging whether the random number p is greater than or equal to 0.5; if so, introducing a weight factor, and selecting the spiral update mechanism to update the individual whale positions; if not, continue to judge the parameters Is it greater than 1? If so, the random search mechanism is selected to update the position of each individual whale. If not, the differential mutation perturbation factor is introduced and the shrinkage and encirclement mechanism is selected to update the position of each individual whale: ； ； in, is the currently randomly selected position vector, is the differential variation perturbation factor, F is the variation scale factor; is the distance between the individual whale and its prey; is a parameter; The optimal parameter combination is used to construct an optimized variational modal decomposition, and the preprocessed signal is decomposed to obtain the intrinsic modal function components; the intrinsic modal function components are screened and reconstructed using the correlation coefficient method to achieve denoising of leakage information.

2. The method for reducing leakage sound wave noise based on wavelet packet threshold and VMD according to claim 1, characterized in that: The multi-scale decomposition of the leakage acoustic wave acquisition signal based on the various parameters in the wavelet packet threshold method is performed, and the correlation coefficient method is used to screen and reconstruct the last layer of node coefficients to obtain the preprocessed signal, specifically including: The optimal wavelet packet basis function, decomposition layer number, threshold function and threshold rule are used to perform multi-scale decomposition on the leakage acoustic wave acquisition signal, and the Pearson correlation coefficient between the coefficient of each node in the last layer and the leakage acoustic wave acquisition signal is calculated. Nodes with correlation greater than or equal to a preset value are reconstructed to obtain preprocessed signals.

3. The method for reducing leakage sound wave noise based on wavelet packet threshold and VMD according to claim 1, characterized in that: The method uses the minimum envelope entropy as the fitness function and adopts the enhanced whale algorithm to optimize the variational mode decomposition method to obtain the optimal parameter combination, specifically including: Initialize the enhanced whale algorithm and determine the initial whale individual based on the minimum envelope entropy; each initial whale individual represents a set of parameter combinations ; The enhanced whale algorithm is used to update the initial individual whale positions; Variational modal decomposition is performed on each whale individual at each updated position, the sum of the envelope entropy of the modal components after decomposition of each whale individual is calculated, and the whale individual with the smallest envelope entropy is recorded; when the maximum number of iterations is reached, the whale individual with the smallest sum of envelope entropy is output as the optimal whale individual; the optimal whale individual represents the optimal parameter combination, including the optimal number of decomposition layers and penalty factor.

4. The method for reducing leakage sound wave noise based on wavelet packet threshold and VMD according to claim 3, characterized in that: Initializing the enhanced whale algorithm and determining the initial whale individual based on the minimum envelope entropy includes: Set the number of whale populations, generate whale individuals, and optimize the target parameters Range restriction; the individual whales represent a group Parameter combination, where is the number of decomposition layers, is the penalty factor; Initialize the parameters of the variational mode decomposition method, perform variational mode decomposition on each whale individual, calculate the sum of the envelope entropies of the modal components of the decomposed signal, and record and save the whale individual with the minimum envelope entropy as the initial whale individual.

5. The method for reducing noise of leakage sound waves based on wavelet packet threshold and VMD according to claim 1, wherein: The optimal parameter combination is used to construct an optimized variational mode decomposition, and the preprocessed signal is decomposed to obtain the intrinsic mode function component; The correlation coefficient method is used to screen and reconstruct the intrinsic mode function components to achieve noise reduction of leakage information, including: Inputting the optimal parameter combination into a variational mode decomposition algorithm, decomposing the multi-scale acoustic wave signal, and calculating the Pearson correlation coefficient between each intrinsic mode function component and the leakage acoustic wave acquisition signal; The modal function components with correlation greater than or equal to a preset value are reconstructed to obtain the noise-reduced leakage sound wave signal.

6. A leakage sound wave noise reduction system based on wavelet packet threshold and VMD, characterized in that: include: The original signal acquisition module is used to obtain the leakage sound wave acquisition signal; A signal preprocessing module is used to screen key factors of the wavelet packet threshold method based on a high signal-to-noise ratio, wherein the key factors include the optimal wavelet packet basis function, the number of decomposition layers, the threshold function, and the threshold rule; a multi-scale decomposition of the leakage acoustic wave acquisition signal is performed based on the various factors in the wavelet packet threshold method, and the correlation coefficient method is used to screen and reconstruct the node coefficients of the last layer to obtain a preprocessed signal; the component parameters of the wavelet packet threshold method are screened based on a high signal-to-noise ratio, specifically as follows: obtaining the original sample signal, performing three rounds of wavelet decomposition on the original sample signal based on different wavelet packet basis functions, the number of decomposition layers, the threshold function, and the threshold rule according to the control variable concept, and screening the optimal parameters based on the maximum signal-to-noise ratio between the reconstructed signal and the original sample signal; In the first round, the number of decomposition layers, threshold function, and threshold rule are fixed, and N wavelet packet basis functions are used to perform wavelet packet decomposition on the sample original signal. The low-frequency signal of the last layer obtained by decomposition is reconstructed to obtain the first reconstructed signal. The signal-to-noise ratio of the first reconstructed signal and the sample original signal is calculated, and the first M wavelet packet basis functions are selected as the initial wavelet packet basis functions based on the maximum signal-to-noise ratio; where 2≤M<N; In the second round, the threshold function and threshold rule are fixed, and the number of decomposition layers is set to 1 to a layers; the original sample signal is subjected to wavelet packet decomposition based on the decomposition layers 1 to a layers using the M initial wavelet packet basis functions, and the last layer of low-frequency signal obtained by the decomposition is reconstructed to obtain a second reconstructed signal; the signal-to-noise ratio of the second reconstructed signal to the original sample signal is calculated, and the optimal wavelet packet basis function and the optimal number of decomposition layers are screened based on the maximum signal-to-noise ratio; wherein a>1; In the third round, the optimal wavelet packet basis function and the optimal number of decomposition layers are fixed, b threshold functions and c threshold rules are set; the original sample signal is subjected to wavelet decomposition using the optimal wavelet packet basis function and the optimal number of decomposition layers based on different threshold functions and threshold rules, and the last layer of low-frequency signal obtained by decomposition is reconstructed to obtain a third reconstructed signal; the signal-to-noise ratio of the third reconstructed signal to the original sample signal is calculated, and the optimal threshold function and optimal threshold rule are selected based on the maximum signal-to-noise ratio; where b>1 and c>1; The algorithm optimization module is used to use the minimum envelope entropy as the fitness function and adopt the enhanced whale algorithm to optimize the variational mode decomposition method to obtain the optimal parameter combination; wherein, the enhanced whale algorithm is used to update the initial individual whale positions, specifically including: generating a random number p, and judging whether the random number p is greater than or equal to 0.5; if so, introducing a weight factor and selecting the spiral update mechanism to update the individual whale positions; if not, continuing to judge the parameters Is it greater than 1? If so, the random search mechanism is selected to update the position of each individual whale. If not, the differential mutation perturbation factor is introduced and the shrinkage and encirclement mechanism is selected to update the position of each individual whale: ； ； in, is the currently randomly selected position vector, is the differential variation perturbation factor, F is the variation scale factor; is the distance between the individual whale and its prey; is a parameter; The denoising module is used to construct an optimized variational mode decomposition using the optimal parameter combination, decompose the preprocessed signal to obtain intrinsic mode function components; and use the correlation coefficient method to screen and reconstruct the intrinsic mode function components to achieve denoising of leakage information.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the leakage sound wave denoising method based on wavelet packet threshold and VMD according to any one of claims 1 to 5 are implemented.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the leakage sound wave denoising method based on wavelet packet threshold and VMD according to any one of claims 1 to 5 are implemented.

Images