A pipeline leakage signal denoising method based on improved WOA-VMD combined with SVD

Abstract

The application discloses a pipeline leakage signal denoising method based on improved WOA-VMD combined with SVD, relates to the technical field of signal processing, and acquires a pipeline leakage signal of a pipeline defect position; a penalty factor alpha and a mode number K in a VMD algorithm are used as optimization targets of a whale optimization algorithm (WOA) algorithm; the VMD algorithm is set according to the optimal penalty factor alpha and the optimal mode number K obtained by the WOA algorithm, and VDM decomposition is carried out on the pipeline leakage signal to obtain K IMF components; the K IMF components are identified by using a correlation coefficient method, and effective IMF components are identified; SVD decomposition and denoising processing are respectively carried out on each effective IMF component by using an SVD algorithm, and noise signals in each effective IMF component are removed; and each effective IMF component after the denoising processing is synthesized to obtain a pipeline leakage signal after denoising. The application combines the WOA, VMD and SVD algorithms, forms an integrated signal processing framework, and has a significant improvement in performance compared with a single method.

Description

A Denoising Method for Pipeline Leakage Signals Based on Improved WOA-VMD Combined with SVD Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a method for denoising pipeline leakage signals based on an improved WOA-VMD combined with SVD. Background Technology

[0002] Acoustic emission (AE) refers to the phenomenon of transient elastic waves emitted by a material due to the rapid release of energy in a localized area; it is sometimes also called stress wave emission. Deformation and crack propagation of materials under stress are important mechanisms of structural failure. Sources directly related to deformation and fracture mechanisms are called acoustic emission sources. In recent years, another type of elastic wave source, such as fluid leakage, friction, impact, and combustion, which are not directly related to deformation and fracture mechanisms, has been termed other or secondary acoustic emission sources.

[0003] Acoustic emission is a common physical phenomenon. The frequency range of acoustic emission signals varies widely across different materials, from infrasound (a few Hz), acoustic frequencies (20 Hz–20 kHz), to ultrasonic frequencies (several MHz). The amplitude of the acoustic emission signals also varies greatly, from microscopic dislocation movements at 10⁻¹³ m to seismic waves on the order of 1 m. The output of a sensor can range from several µV to hundreds of mV; however, in most cases, only very sensitive sensors can detect the faint vibrations. Therefore, the detected acoustic signal waveform is a superposition of sound waves arriving at the sensor via different paths (reverberation effect), which complicates the problem. Furthermore, the sensor itself exhibits a so-called "ringing" effect (sensor response), further complicating the output signal. In many cases, extracting useful information from such a complex signal becomes crucial.

[0004] In oil and gas pipeline systems, external noise and random interference from the acquisition system often result in noisy acoustic signals captured by sensors, severely impacting effective signal analysis. Therefore, denoising before analysis is crucial to improve accuracy. Noise issues are particularly severe in oil and gas pipeline leak signals, necessitating effective denoising. Currently, traditional denoising methods for pipeline leak signal processing often suffer from low processing efficiency and incomplete noise removal. Common signal denoising techniques, such as filters and wavelet transforms, often fail to effectively remove mixed noise and exhibit significant limitations when processing complex signals.

[0005] Variational Mode Decomposition (VMD) is an effective denoising technique. It not only has a solid theoretical foundation but also solves the mode aliasing problem in Empirical Mode Decomposition (EMD) and the challenge of wavelet basis function selection in wavelet decomposition. It can decompose complex signals into several band-limited mode functions (IMFs), each of which can essentially be considered a single-frequency waveform. VMD extracts modal components from the signal by solving a variational problem. The solution process involves minimizing the bandwidth of the solution, thus separating different frequency components in the signal. However, there is no clear-cut standard for determining the number of modes K and the penalty factor α in VMD decomposition. In practice, these parameters K and α often need to be manually set, introducing a high degree of subjectivity.

[0006] Wow Algorithm (WOA) is an emerging optimization algorithm that uses the simulation of whale predation behavior to find the optimal solution and has good global search capabilities.

[0007] Singular Value Decomposition (SVD) is a linear algebraic technique that decomposes any complex data matrix into the product of three simple matrices: an orthogonal matrix U, a diagonal matrix S, and the transpose V of an orthogonal matrix V. T In Singular Value Decomposition (SVD), the diagonal elements of a diagonal matrix S are singular values, arranged in descending order. Singular values ​​reflect the energy or information of the matrix along the corresponding singular vector. In signal processing, larger singular values ​​typically correspond to the principal components of the signal, while smaller singular values ​​are often associated with noise. Summary of the Invention

[0008] To overcome the shortcomings of the prior art, this invention provides a pipeline leakage signal denoising method based on an improved WOA-VMD combined with SVD. By combining the three algorithms WOA, VMD and SVD, an integrated signal processing framework is formed, which significantly improves performance compared to a single method.

[0009] To achieve the above objectives, the present invention adopts the following technical solution, including:

[0010] A method for denoising pipeline leakage signals based on an improved WOA-VMD combined with SVD includes the following steps:

[0011] S1, obtain the original acoustic emission signal of the pipeline defect location, i.e., the pipeline leakage signal;

[0012] S2, taking the penalty factor α and the number of modes K in the variational mode decomposition algorithm (VMD algorithm) as the optimization objectives of the whale algorithm (WOA algorithm), and finding the optimal penalty factor α and the optimal number of modes K in the VMD algorithm;

[0013] S3. Based on the optimal penalty factor α and the optimal number of modes K obtained in step S2, set the VMD algorithm, and use the VMD algorithm to perform VDM decomposition on the pipeline leakage signal in step S1 to obtain K intrinsic mode components, i.e., IMF components.

[0014] S4. Use the correlation coefficient method to identify the K IMF components obtained in step S3 and identify the valid IMF components.

[0015] S5. The Singular Value Decomposition (SVD) algorithm is used to decompose and denoise each effective IMF component, removing noise signals from each effective IMF component. The denoised effective IMF components are then combined to obtain the denoised pipeline leakage signal.

[0016] Preferably, in step S2, the optimization objective is selected based on the fitness value, and the fitness function is:

[0017]

[0018] Where a(j) is the envelope signal of the intrinsic mode components obtained after VMD decomposition of the pipeline leakage signal and Hilbert demodulation; j represents the j-th sampling point in the envelope signal a(j), and N is the number of sampling points; P j Let a(j) be the probability distribution sequence obtained after normalization; calculate the probability distribution sequence P. j The entropy value is used to obtain the envelope entropy E. p ; Calculate the envelope entropy E of each eigenmode component after VMD decomposition. p Select the minimum envelope entropy E p As a fitness value;

[0019] Find the penalty factor α and the number of modes K corresponding to the minimum fitness value as the optimal penalty factor α and the optimal number of modes K.

[0020] Preferably, in step S4, the correlation coefficient method is used to calculate the correlation coefficient between each IMF component and the pipeline leakage signal. IMF components with a correlation coefficient greater than a first threshold are considered valid IMF components; as detailed below:

[0021]

[0022]

[0023] Where x(t) represents the signal value at time t in the pipeline leakage signal; imf(t) represents the signal value at time t in the IMF component; t represents time, t=1,2,...,N, and N represents the signal length; This represents the average signal value of a pipeline leak signal; The average signal value of the IMF component is represented; Conf(x,IMF) represents the correlation coefficient between the IMF component and the pipeline leakage signal.

[0024] First threshold η r =ρ·R max , where R max ρ represents the maximum correlation coefficient obtained from the calculation, and ρ is a set proportion.

[0025] Preferably, the specific process of step S5 is as follows:

[0026] S51, the IMF component is represented as {imf(t)|t=1,2,...,N}; imf(t) is the signal value at time t in the IMF component, where t represents time, t=1,2,...,N, and N is the signal length;

[0027] Transform the one-dimensional effective IMF components into a two-dimensional matrix Y:

[0028]

[0029] Where L is the number of rows in the two-dimensional matrix Y, H is the number of columns in the two-dimensional matrix Y, and H = N - L + 1;

[0030] S52, perform singular value decomposition (SVD) on the two-dimensional matrix Y, decomposing it into an orthogonal matrix U, a diagonal matrix S, and the transpose V of the orthogonal matrix V. T ;

[0031] Y = USV T

[0032] S=[diag(σ1,σ2,…,σ p ),0]

[0033] Wherein, the diagonal matrix S is an L×H diagonal matrix, with diagonal elements σ1,σ2,…,σ p The values ​​are non-negative, the off-diagonal elements are 0, and the diagonal elements of the diagonal matrix S are σ1, σ2, ..., σ p That is, the singular values ​​of the effective IMF components; p = min{L,H};

[0034] S53, by calculating the energy differential spectrum D of each singular value i The singular values ​​are grouped into signal-correlated components and noise components, and a second threshold is set.

[0035]

[0036] Where, σi is the i-th singular value, and p is the number of singular values; D i is the energy differential spectrum of the i-th singular value σ i ; η σ is the second threshold;

[0037] Denote the singular values in D i greater than the second threshold η σ as the signal-related components, and denote the singular values in D i less than or equal to the second threshold η σ as the noise components;

[0038] S54. Using the anti-diagonal averaging method, reconstruct the signal-related components from the two-dimensional matrix into a time series to obtain the denoised effective IMF component imf'(t);

[0039]

[0040] where L * = min(L, H), H * = max(L, H); when L < H, Otherwise, y i,j , y j,i are the elements in the two-dimensional matrix Y;

[0041] S55. Synthesize the denoised effective IMF components to obtain the denoised pipeline leakage signal.

[0042] ]>Preferably, an improved WOA algorithm is used to find the optimal solution, that is, the optimal penalty factor α and the optimal number of modes K, as follows:

[0043] In the algorithm initialization, add a pheromone matrix P, which represents the pheromone concentration of each individual in the solution space;

[0044] For each individual X i , if its fitness value is better than the current global optimal solution, then increase the pheromone:

[0045] P ij = P ij + ΔP·(f(X best ) - f(X i ))

[0046]

[0047] where f(X i ) is the fitness value of the individual X i , and each individual represents a set of solutions of the number of modes K and the penalty factor α; f(Xbest X is the fitness value of the current optimal solution; best This is the current optimal solution; P ij These are elements in the pheromone matrix; ΔP is the pheromone increment; C is a constant used to control the pheromone increment.

[0048] During location updates, the search direction is guided by pheromone concentration to obtain the new individual X. new :

[0049] d i =|X best -X i |

[0050] X new =X i +α1·P ij -β1·d i

[0051] Where α1 and β1 are the weights of pheromone and distance, respectively, and d i For each individual X i With the current optimal solution X best The distance;

[0052] Evaluating the new individual X new fitness f(X) new ), and update the current optimal solution X. best :

[0053]

[0054] A computer program product comprising a computer program / instructions that, when executed by a processor, implement the aforementioned improved WOA-VMD combined with SVD method for denoising pipeline leakage signals.

[0055] A readable storage medium having a computer program stored thereon, which, when executed, implements the above-described improved WOA-VMD combined with SVD method for denoising pipeline leakage signals.

[0056] An electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described improved WOA-VMD combined with SVD method for denoising pipeline leakage signals.

[0057] The advantages of this invention are:

[0058] (1) This invention combines three algorithms, WOA, VMD and SVD, to form an integrated signal processing framework, which significantly improves performance compared to a single method.

[0059] (2) This invention uses the VMD algorithm to remove high oscillations and then uses the SVD algorithm to filter out residual low-frequency random wavenumbers. It can handle various types of noise (such as white noise, impulse noise, etc.) and improves the adaptability to different signal environments.

[0060] (3) The present invention optimizes the parameters of the VMD algorithm through the WOA algorithm, making the signal decomposition more accurate, effectively suppressing noise, and improving the signal-to-noise ratio (SNR) of the signal.

[0061] (4) The present invention adopts an improved WOA algorithm, which shows better performance in search efficiency and convergence speed than the traditional WOA algorithm. This means that in practical applications, the algorithm can find suitable parameters faster, thereby speeding up the entire denoising process.

[0062] (5) The method of the present invention can achieve good results in denoising pipeline leakage signals, while filtering out noise and retaining the effective features of pipeline leakage signals. Attached Figure Description

[0063] Figure 1 is a flowchart of a pipeline leakage signal denoising method based on an improved WOA-VMD combined with SVD according to the present invention.

[0064] Figure 2 is an iterative graph of fitness values ​​before and after the WOA algorithm improvement in the embodiment of the present invention.

[0065] Figure 3 shows the time-domain plot and spectrum of each IMF component obtained by VMD decomposition in an embodiment of the present invention.

[0066] Figure 4 shows the time-domain diagram and spectrum of the denoised pipeline leakage signal obtained in the embodiment of the present invention. Detailed Implementation

[0067] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] As shown in Figure 1, a denoising method for pipeline leakage signals based on WOA (Whale Algorithm), VMD (Variational Mode Decomposition), and SVD (Singular Value Decomposition) is implemented as follows:

[0069] S1 uses an acoustic emission acquisition device to obtain the original acoustic emission signal of the pipeline defect location, i.e., the pipeline leakage signal.

[0070] S2. Parameter settings are configured for the optimized WOA algorithm. The penalty factor α and the number of modes K from the VMD algorithm are used as the optimization objectives of the WOA algorithm, and the envelope entropy of the signal is used as the fitness function of the WOA algorithm. Envelope entropy represents the sparsity characteristics of the pipeline leakage signal. When the intrinsic mode components (IMF components) have more noise and less feature information, the envelope entropy value is larger; conversely, the envelope entropy value is smaller. The pipeline leakage signal from step 1 is input into the WOA algorithm, and the optimization objective is selected according to the fitness function.

[0071] The fitness function is shown below:

[0072]

[0073] In the formula, a(j) is the envelope signal of the intrinsic mode components obtained after VMD decomposition of the pipeline leakage signal and demodulated by Hilbert; j represents the j-th sampling point in the envelope signal a(j), and N is the number of sampling points; P j Let a(j) be the probability distribution sequence obtained after normalization; calculate the probability distribution sequence P. j The entropy value is used to obtain the envelope entropy E. p .

[0074] Calculate the envelope entropy of all intrinsic mode components obtained by VMD decomposition under this location condition (penalty factor α and number of modes K), and take the smallest envelope entropy value as the fitness value.

[0075] Find the penalty factor α and the number of modes K corresponding to the minimum fitness value as the optimal penalty factor α and the optimal number of modes K.

[0076] S3. Based on the optimization objective (penalty factor α and number of modes K) output in step 2, set the parameters of the VMD algorithm, and import the pipeline leakage signal in step 1 into the VMD algorithm for VMD decomposition to obtain K intrinsic mode components, i.e., IMF components.

[0077] S4. The K IMF components obtained in step 3 are then distinguished into effective IMF components and noisy IMF components using the correlation coefficient method. The details are as follows:

[0078] The correlation coefficient method is used to evaluate the correlation between each IMF component and the pipeline leakage signal. This is achieved by calculating the correlation coefficient between each IMF component and the pipeline leakage signal. IMF components with higher correlation coefficients are considered to contain more information from the original signal and are therefore classified as valid IMF components; conversely, components with lower correlation coefficients are considered noisy IMF components. A first threshold is defined as η. r =ρ·R max , where R max To calculate the maximum correlation coefficient, ρ is a set ratio. In this embodiment, based on the characteristics of the acoustic emission crack signal from the pipeline, ρ is set to 0.025. The correlation coefficient is less than or equal to the first threshold η. r The IMF components will be treated as noisy IMF components and removed, provided the correlation coefficient is greater than the first threshold η. r The IMF component will be considered a valid IMF component.

[0079] The specific formula for calculating the correlation coefficient is as follows:

[0080]

[0081] Where x(t) represents the signal value at time t in the pipeline leakage signal; imf(t) represents the signal value at time t in the IMF component; t represents time, t=1,2,...,N, and N represents the signal length; This represents the average signal value of a pipeline leak signal; represents the average signal value of the IMF component; Conf(x,IMF) represents the correlation coefficient between the IMF component and the pipeline leakage signal.

[0082] S5. The Singular Value Decomposition (SVD) algorithm is used to perform SVD decomposition and denoising on each effective IMF component to remove noise signals from each effective IMF component. The denoised effective IMF components are then synthesized to obtain the denoised pipeline leakage signal. The details are as follows:

[0083] S51, the IMF component is represented as {imf(t)|t=1,2,...,N}; imf(t) is the signal value at time t in the IMF component, where t represents the time, t=1,2,...,N, and N is the signal length. The one-dimensional effective IMF component is transformed into a two-dimensional matrix Y, i.e., the Hankel matrix:

[0084]

[0085] Where L is the number of rows in the two-dimensional matrix Y, H is the number of columns in the two-dimensional matrix Y, H = N - L + 1; L is the number of samples that are close to or greater than the lowest frequency component within one period, to ensure that at least one complete period of the lowest frequency component can be captured in each time window, and to avoid the frequency component being distorted or aliased.

[0086] S52, Singular Value Decomposition: Perform SVD decomposition on a two-dimensional matrix Y, decomposing it into an orthogonal matrix U, a diagonal matrix S, and the transpose V of the orthogonal matrix V. T ;

[0087] Y = USV T

[0088] S=[diag(σ1,σ2,…,σ p ),0]

[0089] Wherein, the diagonal matrix S is an L×H diagonal matrix, with diagonal elements σ1,σ2,…,σ p The values ​​are non-negative, the off-diagonal elements are 0, and the diagonal elements of the diagonal matrix S are σ1, σ2, ..., σ p That is, the singular values ​​of the effective IMF components; p = min{L,H}.

[0090] S53, Signal Reconstruction: By calculating the energy differential spectrum D of each singular value. i The singular values ​​are grouped into signal-correlated components and noise components, and a second threshold is set.

[0091]

[0092] Where, σ i Let p be the i-th singular value, and p be the number of singular values; D i For the i-th singular value σ i Energy differential spectrum; η σ The second threshold;

[0093] D i Greater than the second threshold η σ The singular values ​​are denoted as the signal correlation components, and D is... i Less than or equal to the second threshold η σ The singular values ​​are denoted as noise components.

[0094] In this invention, the second threshold is set as the average value of the singular values. For signals and noise that satisfy certain statistical distribution characteristics, the average value can be used as a standard for judging the noise boundary.

[0095] S54: Signal reconstruction: The signal correlation components are reconstructed from a two-dimensional matrix into a time series using the anti-angle averaging method, and the denoised effective IMF component imf'(t) is obtained.

[0096] The total length of the signal is N, and the signal is divided into three segments for reconstruction using the anti-diagonal averaging method:

[0097]

[0098] In the formula, L * = min(L, H), H * = max(L, H); when L < H, Otherwise, y i,j 、y j,i are the elements in the two-dimensional matrix Y. After processing, the low-frequency random wavenumber signal is effectively filtered out.

[0099] S55. Synthesize each effective IMF component after denoising processing to obtain the denoised pipeline leakage signal.

[0100] Using the method of the present invention to denoise the pipeline leakage signal can obtain a good effect, and while filtering out the noise, the effective features in the pipeline leakage signal are also retained.

[0101] In the present invention, when using the WOA algorithm for optimization, the traditional WOA algorithm is also improved. Pheromone is an important concept in the ant colony algorithm. Ants communicate and find paths by releasing and perceiving pheromones. Through the concentration of pheromones, ants can find a better path and avoid suboptimal solutions. The present invention draws on the pheromone mechanism in the ant colony algorithm, introduces pheromones in the WOA algorithm to guide the search process, enhances the utilization of excellent solutions, and can effectively select the appropriate modal number K and penalty factor α. The specific improvements are as follows:

[0102] In the algorithm initialization, add a pheromone matrix P, which represents the pheromone concentration of each individual in the solution space. The pheromone increment ΔP is usually related to the fitness function value. The learning factors α1 and β1 are used to balance pheromone guidance and distance guidance. During the iteration process, update the pheromone according to the fitness function. For each individual X i , if its fitness is better than the current global optimal solution, increase the pheromone:

[0103] P ij = P ij +ΔP·(f(X best ) - f(X i ))

[0104] Among them, f(X i ) is the fitness value of the individual X i ; each individual represents a set of solutions of the modal number K and the penalty factor α; f(X bestX is the fitness value of the current optimal solution; best This is the current optimal solution; P ij These are elements in the pheromone matrix; ΔP is the pheromone increment, which can be set as:

[0105]

[0106] Here, C is a constant used to control the increment of pheromones. After each iteration, the pheromone concentration decreases. This ensures that poorer solutions are gradually eliminated, preventing pheromone concentration.

[0107] During location updates, the search direction is guided by pheromone concentration to obtain the new individual X. new :

[0108] d i =|X best -X i |

[0109] X new =X i +α1·P ij -β1·d i

[0110] Where α1 and β1 are the weights of pheromone and distance, respectively, and d i For each individual X i With the current optimal solution X best The distance. By adjusting these two parameters, the influence of pheromones and distance in the update can be controlled. Evaluating the new individual X new fitness f(X) new ), and update the current optimal solution X. best :

[0111]

[0112] Example 1

[0113] S101 employs an acoustic emission acquisition device to obtain the raw acoustic emission signal, i.e., the pipeline leakage signal, at the location of the pipeline defect. The signal excitation section includes a calibration device and a steel pipe (L = 1000 mm, d = 30 mm), while the signal receiving section includes a sensor, a preamplifier, an acoustic emission host, and host computer software. The standard configuration of the acoustic emission host features a 10 Mbps sampling rate per channel, 16-bit sampling accuracy, low system noise, and a high dynamic range acquisition card. Each acquisition card includes a 1 Gb waveform buffer, perfectly achieving full waveform acquisition without data loss, with a signal length of 1000 mm. The acoustic emission signal from the pipeline crack is exported to the host computer for noise reduction processing.

[0114] Specifically, the signal excitation part mainly uses a calibration device to simulate crack signals on the surface of the steel pipe. The mechanical vibration of the material is collected by the sensor in the signal receiving part and converted into an electrical signal. The signal is amplified by the preamplifier, then processed and recorded by the host computer, and transmitted to the host computer for further processing to obtain the original acoustic emission signal, i.e., the pipeline leakage signal.

[0115] S102, set the parameters for the optimized WOA algorithm, take the penalty factor α and the number of modes K in the VMD algorithm as the optimization objectives, and take the minimum envelope entropy of the signal as the fitness function of the WOA algorithm. Input the pipeline leakage signal in step S101 into the WOA algorithm, and select the optimization objective parameter values ​​according to the fitness function.

[0116] In this embodiment, after optimization using the WOA algorithm, the fitness function value reaches a minimum of 2.919 when the number of iterations is 3, as shown in Figure 2. The iteration speed and fitness value of the improved WOA algorithm (P-WOA in Figure 2) are significantly improved compared to the original (traditional) WOA algorithm (WOA in Figure 2). The optimal output parameters are penalty factor α = 994 and optimal number of modes K = 3.

[0117] S103. Based on the optimal target penalty factor α and the optimal number of modes K output in step S102, the parameters of the VMD algorithm are set, and the pipeline leakage signal in step 1 is imported into the VMD algorithm for decomposition to obtain three intrinsic mode components, namely IMF components. The time domain diagram and spectrum diagram of the components are drawn, as shown in Figure 3.

[0118] S104, the three IMF components obtained in step S3 are distinguished into effective IMF components and noisy IMF components using the correlation coefficient method, with a first threshold η. r The value is set to 0.25. The correlation coefficient of the decomposed IMF1 component is 0.9796, the correlation coefficient of the IMF2 component is 0.2369, and the correlation coefficient of the IMF3 component is 0.1939. Specifically, IMF1 is the effective IMF component, and IMF2 and IMF3 are the noisy IMF components.

[0119] S105, the SVD algorithm is used to decompose and denoise the effective IMF component, i.e., the IMF1 component, in step S104, to remove noise signals from the effective IMF component. The denoised effective IMF component is then reconstructed to obtain the final denoised pipeline leakage signal. The time domain diagram and spectrum diagram of the denoised pipeline leakage signal are shown in Figure 4. In Figure 4, |P1(f)| represents the amplitude or intensity of the frequency.

[0120] The signal-to-noise ratio (SNR) of the denoised pipeline leakage signal obtained using only the VMD algorithm is 13.4495 dB, and the root mean square error (RMSE) is 0.0219. The SNR of the denoised pipeline leakage signal obtained using only the SVD algorithm is 11.0395 dB, and the RMS error is 0.0366. The SNR of the denoised pipeline leakage signal obtained using the WOA-VMD combined SVD method of this invention is 14.1311 dB, and the RMS error is 0.0183. The results and Figure 4 showing the denoised signal from this invention demonstrate that the method proposed in this invention has a higher SNR and a lower RMS error, resulting in better denoising performance.

[0121] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for denoising pipeline leakage signals based on an improved WOA-VMD combined with SVD, characterized in that, Includes the following steps: S1. Obtain the original acoustic emission signal of the pipeline defect location, i.e., the pipeline leakage signal; S2. Use the penalty factor α and the number of modes K in the Variational Mode Decomposition (VMD) algorithm as the optimization objectives of the Whale Algorithm (WOA) algorithm to find the optimal penalty factor α and the optimal number of modes K in the VMD algorithm; S3. Set the VMD algorithm according to the optimal penalty factor α and the optimal number of modes K obtained in step S2, and use the VMD algorithm to perform VDM decomposition on the pipeline leakage signal in step S1 to obtain K intrinsic mode components (IMF components); S4. Use the correlation coefficient method to identify the K IMF components obtained in step S3 and identify the effective IMF components; S5. Use the Singular Value Decomposition (SVD) algorithm to perform SVD decomposition and denoising on each effective IMF component to remove the noise signal in each effective IMF component; synthesize the denoised effective IMF components to obtain the denoised pipeline leakage signal; The specific process of step S5 is as follows: S51, the IMF component is represented as ； Y represents the signal value at time t in the IMF component, where t represents time (t=1,2,...,N) and N is the signal length. The one-dimensional effective IMF component is transformed into a two-dimensional matrix Y. Where L is the number of rows in the two-dimensional matrix Y, and H is the number of columns in the two-dimensional matrix Y, H = N - L + 1; S52, perform singular value decomposition (SVD) on the two-dimensional matrix Y, decomposing it into an orthogonal matrix U, a diagonal matrix S, and the transpose of an orthogonal matrix V. ； Wherein, the diagonal matrix S is an L×H diagonal matrix, with diagonal elements Non-negative values, off-diagonal elements are 0, and diagonal elements of the diagonal matrix S. These are the singular values ​​of the effective IMF components; S53, by calculating the energy differential spectrum of each singular value. The singular values ​​are grouped into signal-correlated components and noise components, and a second threshold is set. in, Let be the i-th singular value, and p be the number of singular values; For the i-th singular value The energy differential spectrum; The second threshold; Greater than the second threshold The singular values ​​are denoted as the signal correlation components, and the... Less than or equal to the second threshold The singular values ​​are denoted as noise components; S54, the signal correlation components are reconstructed from a two-dimensional matrix into a time series using the anti-angle averaging method, yielding the denoised effective IMF components. ； In the formula, ， ; when L < H, ;otherwise, ； 、 Let X be an element in a two-dimensional matrix Y; S55, synthesize the effective IMF components after denoising to obtain the denoised pipeline leakage signal; the improved WOA algorithm is used to find the optimal solution, namely the optimal penalty factor α and the optimal number of modes K, as shown below: In the algorithm initialization, a pheromone matrix P is added to represent the pheromone concentration of each individual in the solution space; for each individual X i If its fitness value is better than the current global optimum, then increase the pheromone level. in, For individual X i The fitness value of each individual represents a solution with a set of modalities K and a penalty factor α; It is the fitness value of the current optimal solution; This is the current optimal solution; These are elements in the pheromone matrix; C is the pheromone increment; C is a constant used to control the pheromone increment; during position updates, the search direction is guided by the pheromone concentration to obtain new individuals. ： in, 1 and 1 represents the weights of pheromones and distance, respectively. For each individual X i Compared with the current optimal solution Distance; assessing new individuals fitness f( ), and update the current optimal solution. ： 。 2. The pipeline leakage signal denoising method based on improved WOA-VMD combined with SVD according to claim 1, characterized in that, In step S2, the optimization objective is selected based on the fitness value, and the fitness function is: in, The envelope signal obtained by Hilbert demodulation of the intrinsic mode components after VMD decomposition of the pipeline leakage signal; j represents the envelope signal. The j-th sampling point in the sample, where N is the number of sampling points; P j for The probability distribution sequence obtained after normalization; calculate the probability distribution sequence P. j The entropy value is used to obtain the envelope entropy. ; Calculate the envelope entropy of each intrinsic mode component after VMD decomposition. Select the minimum envelope entropy As the fitness value, find the penalty factor α and the number of modes K corresponding to the minimum fitness value as the optimal penalty factor α and the optimal number of modes K.

3. The method for denoising pipeline leakage signals based on an improved WOA-VMD combined with SVD according to claim 1, characterized in that, In step S4, the correlation coefficient method is used to calculate the correlation coefficient between each IMF component and the pipeline leakage signal. IMF components with a correlation coefficient greater than a first threshold are considered valid IMF components; the details are as follows: in, This represents the signal value at time t in the pipeline leakage signal; This represents the signal value at time t in the IMF component; t represents time, t=1,2,...,N, and N represents the signal length. This represents the average signal value of a pipeline leak signal; This represents the average signal value of the IMF components; The first threshold represents the correlation coefficient between the IMF component and the pipeline leakage signal. ,in, ρ represents the maximum correlation coefficient obtained from the calculation, and ρ is a set proportion.

4. A computer program product, characterized in that, It includes a computer program / instruction that, when executed by a processor, implements the pipe leakage signal denoising method based on an improved WOA-VMD combined with SVD as described in any one of claims 1 to 3.

5. A readable storage medium, characterized in that, It stores a computer program, which, when executed, implements the pipeline leakage signal denoising method based on an improved WOA-VMD combined with SVD as described in any one of claims 1 to 3.

6. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the pipeline leakage signal denoising method based on any one of claims 1 to 3.

Images