Valve inner leakage signal identification method based on variational mode decomposition and search algorithm

Abstract

The invention discloses a valve inner leakage signal identification method based on variational mode decomposition and a search algorithm, and belongs to the field of valve inner leakage signal identification, and the method comprises the steps: collecting an original signal of an acoustic emission sensor; optimizing the decomposition number and penalty factor of the variational mode decomposition VMD algorithm by using a particle swarm algorithm to obtain an optimized variational mode decomposition VMD algorithm; performing signal noise reduction processing on the original signal by using the optimized variational mode decomposition (VMD) algorithm to obtain a denoised valve signal; extracting an energy ratio characteristic quantity from the denoised valve signal by using a small packet wave method; and according to the energy ratio characteristic quantity, utilizing a support vector machine SSA-SVM optimized by a sparrow algorithm to obtain a valve inner leakage signal identification result. The method solves the problem that the recognition efficiency of a classifier is low due to the fact that the signal noise reduction effect of an existing method is not ideal.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of valve internal leakage signal identification, and particularly relates to a valve internal leakage signal identification method based on variational mode decomposition and a search algorithm. BACKGROUND

[0002] Valve internal leakage is a common fault in pipeline systems of petroleum, chemical and other industries, which usually causes flow-induced noise. The acoustic emission detection method can determine the valve internal leakage state by collecting acoustic signals in the valve and pipeline, and has the advantages of non-invasiveness, strong adaptability, real-time monitoring and low cost. Although acoustic emission technology can effectively detect internal leakage noise, in complex environments, noise such as pipe wall vibration and electromagnetic interference is mixed with effective signals, which causes the traditional frequency domain filtering method to fail, thereby affecting the detection accuracy. To solve this problem, existing research generally uses acoustic emission parameter analysis (such as root mean square, energy proportion) or machine learning models (such as SVM, neural network) for leakage identification.

[0003] In the aspect of signal processing, one existing technical solution is the relationship between the RMS of the acoustic emission signal and the leakage rate, valve size and type; one existing technical solution is a monitoring system based on the relationship between the RMS of the acoustic emission signal and the leakage flow rate; one existing technical solution is to analyze the energy distribution under different leakage modes through wavelet packet analysis, and proposes an energy proportion method to distinguish different internal leakage modes; in addition, one existing technical solution is to combine artificial neural networks with acoustic emission technology to identify four fault modes of check valves; one existing technical solution is to apply artificial neural networks to analyze internal combustion engine valve faults; one existing technical solution is a valve fault diagnosis method based on optimized SVM with an accuracy of 95%; one existing technical solution is to propose a high-accuracy internal leakage identification algorithm combining kernel principal component analysis and SVM; one existing technical solution is to study a fault diagnosis method for compressor valves, and both SVM and artificial neural networks can effectively diagnose faults; one existing technical solution is to propose a back propagation neural network model with an accuracy of over 90%; one existing technical solution is to propose a back propagation neural network that improves prediction accuracy through dimensionality reduction; one existing technical solution is to predict the degree of valve leakage through SVM classification with an accuracy of 93%; one existing technical solution is a neural network model based on wavelet packet denoising and sparrow search optimization that can accurately predict the internal leakage rate; one existing technical solution is to propose an identification model based on a convolutional neural network with an accuracy of 99.6%. Although these methods have achieved good identification results, in a complex noise environment, especially when facing multiple valve types and working conditions, the adaptability and generalization ability of the model are still insufficient. Many methods rely on empirical parameters or traditional optimization algorithms, resulting in low efficiency of classifier parameter optimization and easy falling into local optimum. In addition, these methods lack adaptive optimization mechanisms, resulting in unsatisfactory signal denoising effect, which in turn affects the generalization ability and accuracy of the model under different working conditions. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a valve internal leakage signal recognition method based on variational mode decomposition and search algorithm, which solves the problem that the signal noise reduction effect of the existing method is not ideal, resulting in low classifier recognition efficiency.

[0005] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a valve internal leakage signal recognition method based on variational mode decomposition and search algorithm, comprising: Collect the original signal of the acoustic emission sensor; The particle swarm optimization algorithm is used to optimize the number of decompositions and penalty factors of the variational mode decomposition (VMD) algorithm, and the optimized variational mode decomposition (VMD) algorithm is obtained. The optimized variational mode decomposition (VMD) algorithm is used to perform signal noise reduction on the original signal to obtain the noise-reduced valve signal. The energy ratio feature is extracted from the valve signal after noise reduction using the small packet wave method; According to the energy proportion feature, the support vector machine SSA-SVM optimized by the sparrow algorithm is used to obtain the valve internal leakage signal recognition result.

[0006] Furthermore, the particle swarm optimization is used to optimize the number of decompositions and the penalty factor of the variational mode decomposition (VMD) algorithm to obtain the optimized variational mode decomposition (VMD) algorithm, specifically: A1. Initialize particle position and particle velocity; is the number of decompositions of the variational mode decomposition VMD algorithm; is the penalty factor of the variational mode decomposition VMD algorithm; A2. Update the variational mode decomposition (VMD) algorithm based on the particle position and calculate the envelope entropy value of each mode; A3. Update the local optimal value and the global optimal value according to the envelope entropy value of each mode; A4. Update particle positions and particle velocities based on local and global optimal values. A5. Determine whether the maximum number of iterations has been reached. If so, update the variational mode decomposition (VMD) algorithm based on the updated particle positions to obtain an optimized variational mode decomposition (VMD) algorithm. Otherwise, return to step A2.

[0007] Furthermore, the expression of the updated particle velocity is:

[0008] in, For the After the iterative update The particle in Speed ​​in dimension; is the inertia weight; For the In the iteration The particle in Speed ​​in dimension; and All are learning factors; is a random number in [0,1]; For the In the iteration The particle in Local optimum in dimension; For the In the iteration The particle in Position on the dimension; For the In the first iteration, the particle swarm The global optimum in dimension.

[0009] Furthermore, the expression of the updated particle position is:

[0010] in, For the After the iterative update The particle in Position on the dimension.

[0011] Furthermore, the optimized variational mode decomposition (VMD) algorithm is used to perform signal noise reduction processing on the original signal to obtain a valve signal after noise reduction, specifically: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal, and the signal is reconstructed through the correlation coefficient to obtain the valve signal after noise reduction.

[0012] Furthermore, the optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal, and the signal is reconstructed through the correlation coefficient to obtain the valve signal after noise reduction, specifically: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal to obtain the time domain signal of each mode. Calculate the correlation coefficient between the time domain signal of each mode and the original signal respectively; The time domain signals of the modes with correlation coefficients greater than the correlation threshold are retained to obtain several valid time domain signals of the modes; The time domain signals of each effective mode are linearly added to complete the signal reconstruction and obtain the valve signal after noise reduction.

[0013] Further, the original signal is decomposed by using the optimized VMD algorithm to obtain time-domain signals of each mode, specifically: The original signal is decomposed into time-domain signals of M modes by using the optimized VMD algorithm, with the minimum bandwidth sum as the target. The target expression is as follows:

[0014] Wherein, The modal component set to be minimized is as follows: And the center frequency set is as follows: ; The derivative operation with respect to time t is as follows: The Dirac function is as follows: The time is as follows: The center frequency of the Mth mode is as follows: The imaginary unit is as follows: The Mth modal component is as follows: The original signal is as follows: The L2 norm is as follows: Further, the decomposition scale of the wavelet packet method satisfies the following condition:

[0015] Further, the decomposition scale of the wavelet packet method satisfies the following condition:

[0016] Wherein, The number of sampling points is as follows: The sampling rate is as follows:

[0017] The present application has the following advantages: the VMD parameters are optimized by the PSO, the noise is suppressed, and the robustness is improved; the multi-scale features are extracted as classification basis by using the wavelet packet energy proportion combined with the frequency domain characteristics of the valve leakage signal; meanwhile, the kernel function parameters of the SVM are optimized by the SSA, and the classification efficiency and accuracy are improved. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 The method flowchart of the present application is as follows.

[0019] Figure 2 The structure diagram of the experimental system in the embodiment of the present application is as follows.

[0020] Figure 3 The signal schematic diagram of the gate valve under different leakage rates in the embodiment of the present application is as follows. ​​​​​

[0021] Figure 4 This is a flow chart of signal decomposition and reconstruction in an embodiment of the present invention.

[0022] Figure 5 Flowchart of the PSO-VMD algorithm in an embodiment of the present invention.

[0023] Figure 6 In the embodiment of the present invention =6. Time domain diagram of each component.

[0024] Figure 7 In the embodiment of the present invention =6 when the frequency domain diagram of each component.

[0025] Figure 8 Graph showing correlation coefficients in an embodiment of the present invention.

[0026] Figure 9 This is a diagram of a reconstructed signal in an embodiment of the present invention.

[0027] Figure 10 This is a wavelet packet energy ratio diagram of a DN100 gate valve at different leakage rates in an embodiment of the present invention.

[0028] Figure 11 This is a wavelet packet energy ratio diagram when the gate valve is normally closed in an embodiment of the present invention.

[0029] Figure 12 Schematic diagram of the classification accuracy of SVMs composed of different kernel functions in an embodiment of the present invention.

[0030] Figure 13 Schematic diagram of the classification results of SSA-SVM and grid search method optimized SVM in an embodiment of the present invention.

[0031] in, Figure 3 (a) is a schematic diagram of the signal when the gate valve leakage rate is 0 L / min; Figure 3 (b) is the signal diagram of the gate valve leakage rate of 26.6L / min; Figure 3 (c) is a schematic diagram of the signal when the gate valve leakage rate is 42 L / min; Figure 3 (d) is a schematic diagram of the signal when the gate valve leakage rate is 63.7 L / min. DETAILED DESCRIPTION

[0032] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0033] like Figure 1 As shown, in one embodiment of the present invention, a valve internal leakage signal identification method based on variational mode decomposition and search algorithm includes: Collect the original signal of the acoustic emission sensor; The particle swarm optimization algorithm is used to optimize the number of decompositions and penalty factors of the variational mode decomposition (VMD) algorithm, and the optimized variational mode decomposition (VMD) algorithm is obtained. The optimized variational mode decomposition (VMD) algorithm is used to perform signal noise reduction on the original signal to obtain the noise-reduced valve signal. The energy ratio feature is extracted from the valve signal after noise reduction using the small packet wave method; According to the energy proportion feature, the support vector machine SSA-SVM optimized by the sparrow algorithm is used to obtain the valve internal leakage signal recognition result.

[0034] In this embodiment, the device for simulating valve internal leakage is constructed experimentally, such as Figure 2 As shown, the device includes an air compressor, storage tank, pressure gauge, flow meter, valve, and muffler. Gate valves and ball valves were used as experimental objects. The valve opening, upstream and downstream pressure differential, and pipe diameter were controlled to collect internal leakage signals and normal closing signals under different operating conditions.

[0035] First, the time and frequency domain characteristics of the valve internal leakage acoustic emission signal were studied. Time domain analysis showed that the RMS value was closely related to the leakage rate, pipe diameter, and pressure difference. The RMS value increased significantly when the leakage rate increased. When the pipe diameter and leakage rate remained constant, the RMS value increased with the pressure difference. The signal information entropy also increased with the increase of leakage rate, indicating that the leakage signal contains more information. Frequency domain analysis showed that, as Figure 3 As shown, the spectrum of the internal leakage signal is concentrated between 0 and 150 kHz. This spectral aggregation becomes more pronounced as the leakage rate increases, and the amplitude of the internal leakage signal is much higher than that of the normally closed signal. The spectrum of different valve types also exhibits similar variations, making it possible to identify valve internal leakage without distinguishing between valve types.

[0036] The particle swarm optimization is used to optimize the number of decompositions and penalty factor of the variational mode decomposition (VMD) algorithm to obtain the optimized variational mode decomposition (VMD) algorithm, specifically: A1. Initialize particle position and particle velocity; is the number of decompositions of the variational mode decomposition VMD algorithm; is the penalty factor of the variational mode decomposition VMD algorithm; A2. Update the variational mode decomposition (VMD) algorithm based on the particle position and calculate the envelope entropy value of each mode; A3. Update the local optimal value and the global optimal value according to the envelope entropy value of each mode; A4. Update particle positions and particle velocities based on local and global optimal values. A5. Determine whether the maximum number of iterations has been reached. If so, update the variational mode decomposition (VMD) algorithm based on the updated particle positions to obtain an optimized variational mode decomposition (VMD) algorithm. Otherwise, return to step A2.

[0037] The expression for the updated particle velocity is:

[0038] in, For the After the iterative update The particle in Speed ​​in dimension; is the inertia weight; For the In the iteration The particle in Speed ​​in dimension; and All are learning factors; is a random number in [0,1]; For the In the iteration The particle in Local optimum in dimension; For the In the iteration The particle in Position on the dimension; For the In the first iteration, the particle swarm The global optimum in dimension.

[0039] The expression for the updated particle position is:

[0040] in, For the After the iterative update The particle in Position on the dimension.

[0041] In this embodiment, noise in experimental and industrial environments interferes with signal acquisition and, in turn, affects subsequent analysis. Therefore, for operations such as feature extraction and classification, the original signal must be preprocessed to reduce noise. Observing the acoustic emission signals of the valve when it is normally closed and when it is leaking internally reveals that the ambient noise has a wide frequency band, with a uniform distribution of frequency components and no specific range. Therefore, it cannot be removed using a filter with a defined frequency band. Furthermore, the signal amplitudes differ significantly between closed and leaking valves. The flow-induced noise signal energy of a leaking valve is greater than that of the ambient noise, and the signal energy increases with increasing leakage.

[0042] Therefore, this valve uses the VMD method to decompose the original signal of the acoustic emission sensor based on the energy difference between the effective signal and the ambient noise, and proposes a modal signal with high similarity through correlation operation for reconstruction, thereby achieving noise reduction processing. Before using VMD to process the signal, its parameters must be determined first. Traditional methods usually rely on experience, but this valve uses the PSO algorithm to optimize the VMD parameters, and then uses the optimized VMD to decompose and reconstruct the signal, and completes the signal reconstruction through the correlation coefficient. The process of PSO-VMD decomposition and reconstruction is as follows: Figure 4 shown.

[0043] The key parameter in the VMD method is the number of decompositions and penalty factor , parameters have a significant impact on the signal decomposition effect. Too large will result in over-decomposition, too small will result in under-decomposition; If the value is too large, the signal will be decomposed into multiple modes, and if it is too small, noise will be introduced. This valve uses the PSO algorithm to optimize the VMD parameters, using the envelope entropy of the collected signal as the fitness function, and minimizing the envelope entropy to determine the optimal parameters. The process diagram of the combination of PSO and VMD is as follows Figure 5 As shown. By calculating the envelope entropy, the optimized VMD parameters are is 6, is 300.

[0044] The optimized variational mode decomposition (VMD) algorithm is used to perform signal noise reduction processing on the original signal to obtain the valve signal after noise reduction, specifically: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal, and the signal is reconstructed through the correlation coefficient to obtain the valve signal after noise reduction.

[0045] The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal, and the signal is reconstructed through the correlation coefficient to obtain the valve signal after noise reduction, specifically: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal to obtain the time domain signal of each mode. Calculate the correlation coefficient of the time-domain signal of each mode and the original signal respectively; Reserve the time-domain signal of the mode with the correlation coefficient greater than the correlation threshold to obtain several effective time-domain signals of the mode; Linearly add the effective time-domain signals of the mode to complete signal reconstruction to obtain the valve signal after noise reduction.

[0046] The original signal is decomposed by using the optimized VMD algorithm to obtain the time-domain signal of each mode, specifically: The original signal is decomposed into time-domain signals of the mode by using the optimized VMD algorithm with the minimum bandwidth sum as the target; The number of decompositions of the VMD algorithm is The expression of the target is:

[0047] Wherein, The modal component set and the center frequency set are minimized; The derivative operation is performed on time ; The Dirac function is ; The time is ; The center frequency of the th mode is ; The imaginary unit is ; The th modal component is ; The original signal is ; The L2 norm is

[0048] In this embodiment, the original signal of the DN100 gate valve under the condition of 0.2 MPa pressure difference between upstream and downstream and 108 L / min leakage rate is decomposed by VMD, the VMD parameters optimized by PSO are adopted, =6, =300. When =6, =300, the decomposed modal signals are as shown in Figure 6 . Through time-domain to frequency-domain conversion, the frequency spectrum of each component is obtained, as shown in Figure 7 , the VMD components IMF1 to IMF6 are arranged from low to high in frequency.

[0049] In this embodiment, the correlation coefficient reflects the similarity between the time-domain signals. The correlation coefficient of the signal and each component is as shown in Figure 8As shown. The correlation coefficients of IMF1 and IMF2 are large, 0.724 and 0.682 respectively, indicating that they are highly correlated with the original signal; while the coefficients of the last four components are all less than 0.4, indicating that the correlation is low. Therefore, only IMF1 and IMF2 are retained for reconstruction. According to the VMD principle, IMF1 and IMF2 are linearly added to obtain the reconstructed signal. The reconstructed time domain waveform and spectrum are shown in Figure 9 As shown, the amplitude of the signal is similar to the original signal, but it is smoother in the time domain and the high-frequency part of the spectrum is removed.

[0050] Decomposition scale of the small packet wave method satisfy:

[0051] in, is the number of sampling points; is the sampling rate.

[0052] In this example, the SVM classification algorithm was used to classify and identify valves that were properly closed and those that had internal leakage. A wavelet packet method was used to extract energy proportion features from the de-noised signal, which served as input for the SVM classification algorithm. The classification results of SVMs using different optimization algorithms and kernel functions were compared to determine the optimal kernel function type, penalty factor parameters, and kernel function parameters.

[0053] In this example, the frequency of the valve internal leakage signal is concentrated between 0 and 150 kHz, while the frequency spectrum of the normally closed signal is relatively uniform. Therefore, the frequency distribution can be used as a basis for internal leakage determination. Therefore, internal leakage identification is performed by extracting acoustic emission signal features based on wavelet packet energy ratio. Using the db10 wavelet to decompose the valve internal leakage acoustic emission signal yields excellent decomposition characteristics.

[0054] In order to capture better details while not making the decomposition too redundant, a 5-layer decomposition is selected to capture details. The node energy after decomposition is set to , the total energy is assumed to be P , the energy proportion of each component is assumed to be Then the total energy P The energy ratio of each node is as follows:

[0055] Taking a DN100 gate valve as an example, we selected data at different leakage rates for wavelet packet decomposition and calculated the energy contribution. A five-layer wavelet packet decomposition yielded 32 nodes, each corresponding to a frequency band. The sampling rate was 1 MS / s, and each frequency band covered a 15.625 kHz range. Figure 10The energy ratio of the valve internal leakage signal after decomposition is shown in the figure. The energy ratio decreases as the frequency band increases. The energy of the first three frequency bands is concentrated, and the peak is located in the second or third frequency band. The signal spectrum trends under different leakage rate conditions are similar, but the values ​​are different. For a normally closed valve, the signal spectrum is uniform. Figure 11 This is the wavelet packet energy ratio diagram of a normal valve, and its energy ratio is relatively uniform.

[0056] In summary, the internal leakage signals of gate valves and ball valves show energy concentration, while the energy of normal closing signals is evenly distributed.

[0057] The classification results for valve closed and internal leakage data at different leakage rates show that internal leakage data has distinct energy distribution characteristics. The penalty factor c was set to 80, and the kernel function parameter g was set to 0.15. The wavelet packet energy ratio feature sample set was input into an SVM composed of four kernel functions for classification. Figure 12 The classification results are shown. Overlapping marks indicate correct classification, while mismatched marks indicate incorrect classification. Table 1 summarizes the classification results using different kernel functions. By comparing different kernel functions (sigmoid, linear, RBF, and polynomial), we found that the RBF kernel function performed best in SVM classification, achieving a classification accuracy of 94.444%.

[0058] Table 1 Characteristic data of valve normal closing and internal leakage

[0059] In order to further improve the classification accuracy, the classification accuracy of SVM is used as the fitness function of SSA, and the parameters c and g of SVM are optimized by SSA. Compared with the traditional grid search method, SSA-SVM can more effectively find the optimal parameters during the optimization process and improve the classification accuracy. The classification results of SVM using RBF kernel function are shown in the figure below. Figure 13 As shown in Table 2, the SVM model optimized using SSA achieved an accuracy of 98.148%, significantly outperforming the SVM optimized using the grid search method (95.370%). The classification accuracy for different kernel functions is shown in Table 2. The RBF kernel performed best among all kernel function types, achieving an accuracy of 98.148%.

[0060] Table 2 SSA-SVM recognition rates with different kernel function types

[0061] This study proposes a combined noise reduction and identification method based on variational mode decomposition (VMD) and a sparrow search algorithm-optimized support vector machine (SSA-SVM) to improve the detection accuracy of valve internal leakage signals. PSO-VMD noise reduction significantly suppresses high-frequency noise and improves the time-domain smoothness of the signal. Combining wavelet packet energy ratio feature extraction with the SSA-SVM classification model, the method successfully achieves high-accuracy identification of internal leakage in multiple valve types. Experimental results show that the classification accuracy of the SSA-SVM model based on the RBF kernel function reaches 98.148%, a 2.778% improvement over the SVM optimized using the traditional grid search method. This method not only effectively addresses noise interference in complex industrial environments but also improves the accuracy and robustness of valve health monitoring, demonstrating its strong application potential.

Claims

1. A valve internal leakage signal recognition method based on variational mode decomposition and search algorithm, characterized in that: include: Collect the original signal of the acoustic emission sensor; The particle swarm optimization algorithm is used to optimize the number of decompositions and penalty factors of the variational mode decomposition (VMD) algorithm, and the optimized variational mode decomposition (VMD) algorithm is obtained. The optimized variational mode decomposition (VMD) algorithm is used to perform signal noise reduction on the original signal to obtain the valve signal after noise reduction. The energy ratio feature is extracted from the valve signal after noise reduction using the small packet wave method; According to the energy proportion feature, the support vector machine SSA-SVM optimized by the sparrow algorithm is used to obtain the valve internal leakage signal recognition result.

2. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 1 is characterized in that: The particle swarm optimization is used to optimize the number of decompositions and the penalty factor of the variational mode decomposition (VMD) algorithm to obtain the optimized variational mode decomposition (VMD) algorithm, specifically: A1. Initialize particle position and particle velocity; is the number of decompositions of the variational mode decomposition VMD algorithm; is the penalty factor of the variational mode decomposition VMD algorithm; A2. Update the variational mode decomposition (VMD) algorithm based on the particle position and calculate the envelope entropy value of each mode; A3. Update the local optimal value and the global optimal value according to the envelope entropy value of each mode; A4. Update particle positions and particle velocities based on local and global optimal values. A5. Determine whether the maximum number of iterations has been reached. If so, update the variational mode decomposition (VMD) algorithm based on the updated particle positions to obtain an optimized variational mode decomposition (VMD) algorithm. Otherwise, return to step A2.

3. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 2 is characterized in that: The expression for the updated particle velocity is: in, For the After the iterative update The particle in Speed ​​in dimension; is the inertia weight; For the In the iteration The particle in Speed ​​in dimension; and All are learning factors; is a random number in [0,1]; For the In the iteration The particle in Local optimum in dimension; For the In the iteration The particle in Position on the dimension; For the In the first iteration, the particle swarm The global optimum in dimension.

4. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 3 is characterized in that: The expression for the updated particle position is: in, For the After the iterative update The particle in Position on the dimension.

5. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 1 is characterized in that: The optimized variational mode decomposition (VMD) algorithm is used to perform signal noise reduction processing on the original signal to obtain the valve signal after noise reduction, specifically: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal, and the signal is reconstructed through the correlation coefficient to obtain the valve signal after noise reduction.

6. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 5 is characterized in that: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal, and the signal is reconstructed through the correlation coefficient to obtain the valve signal after noise reduction, specifically: The original signal is decomposed using the optimized variational mode decomposition (VMD) algorithm to obtain the time domain signals of each mode. Calculate the correlation coefficient between the time domain signal of each mode and the original signal respectively; The time domain signals of the modes with correlation coefficients greater than the correlation threshold are retained to obtain several valid time domain signals of the modes; The time domain signals of each effective mode are linearly added to complete the signal reconstruction and obtain the valve signal after noise reduction.

7. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 6 is characterized in that: The optimized variational mode decomposition (VMD) algorithm is used to decompose the original signal to obtain the time domain signal of each mode, specifically: Using the optimized variational mode decomposition (VMD) algorithm, the original signal is decomposed into The time domain signal of each mode; is the number of decompositions of the variational mode decomposition VMD algorithm; The expression of the target is: in, To minimize the set of modal components and the center frequency set ; For time The derivative operation of is the Dirac function; For time; For the The center frequency of each mode; is an imaginary unit; For the modal components; is the original signal; is the L2 norm.

8. The valve internal leakage signal identification method based on variational mode decomposition and search algorithm according to claim 1 is characterized in that: Decomposition scale of the small packet wave method satisfy: in, is the number of sampling points; is the sampling rate.

Images