Weld defect signal denoising method and storage medium

Abstract

The invention provides a weld defect signal denoising method and a storage medium, and the method comprises the steps: scanning a preset carbon steel weld through an ultrasonic phased array, and obtaining the weld scanning original signal data of the carbon steel weld; the weld joint scanning original signal data comprises noise signal data; taking the envelope entropy as a fitness function, and applying an SSA optimization algorithm to obtain a decomposition modal number K and a penalty factor alpha; performing VMD decomposition processing on the weld scanning signal data according to the optimized parameter decomposition mode number K and the penalty factor alpha to obtain a correlation coefficient and a variance contribution rate of each mode component, and screening noisy mode components to obtain noisy mode components and non-noisy mode components; determining an optimal wavelet basis function and a decomposition layer number by using the index T, and performing wavelet packet decomposition and reconstruction on the noisy modal component to obtain a denoised modal component; and carrying out reconstruction processing in combination with the noisy mode component to obtain a denoised signal.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing and analysis, and in particular to a weld defect signal denoising method and a storage medium. Background Art

[0002] The defect A-scan data collected by the ultrasonic phased array detector inevitably contains noise data, including test block material structure noise, equipment noise, environmental noise, etc., which undoubtedly affects the analysis of the A-scan signal. In addition, when manually interpreting defects, the characteristics of the A-scan waveform are one of the important reference indicators. The presence of noise will also interfere with this process, thereby affecting the accuracy of the interpretation. Therefore, after obtaining the defect A-scan data, it is necessary to perform necessary noise reduction processing.

[0003] The traditional signal denoising effect is generally judged by the monotonic relationship between a single indicator and the denoising effect, such as signal-to-noise ratio, root mean square error, smoothness, correlation coefficient, noise modulus, etc. However, in the actual denoising process, the pure signal is unknown, that is, the true value of the signal is unknown, and there are certain limitations at this time.

[0004] Therefore, the industry is in urgent need of a new technology for weld defect signal denoising. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned prior art, the present invention provides a weld defect signal denoising method, which can effectively retain the detailed characteristics of the ultrasonic phased array defect A-scan signal and achieve a better noise reduction effect.

[0006] To achieve the above object, the present invention provides a method for denoising a weld defect signal, comprising the following steps:

[0007] Scanning a preset carbon steel weld by an ultrasonic phased array to obtain weld scanning raw signal data of the carbon steel weld; the weld scanning raw signal data includes noise signal data;

[0008] Using envelope entropy as a fitness function, a sparrow optimization algorithm is used to obtain the decomposition mode number K and the penalty factor α of the weld scanning raw signal data;

[0009] Performing variational modal decomposition processing on the weld scanning signal data according to the decomposition modal number K and the penalty factor α to obtain the correlation coefficient and variance contribution rate of each modal component, and screening the noisy modal component to obtain the noisy modal component and the non-noisy modal component;

[0010] The composite evaluation index T is constructed by combining the two indicators of root mean square error and smoothness;

[0011] The optimal wavelet basis function and the number of decomposition layers are determined by using the composite evaluation index T, and the noisy modal components are decomposed and reconstructed using wavelet packets to obtain the modal components after noise reduction;

[0012] The denoised signal can be obtained by reconstructing the denoised modal components in combination with the non-noised modal components.

[0013] Furthermore, the noise signal data includes test block material structure noise, equipment noise and environmental noise.

[0014] Furthermore, the envelope entropy is used as the fitness function, and the sparrow optimization algorithm is used to obtain the decomposition mode number K and the penalty factor α; specifically:

[0015] Initialize the parameters of the sparrow optimization algorithm;

[0016] Set the minimum and maximum values ​​of the decomposition mode number K and the penalty factor α, the population size, and the maximum number of iterations;

[0017] Run the sparrow optimization algorithm to obtain the optimal parameter decomposition mode number K and penalty factor α, and the corresponding minimum envelope entropy value.

[0018] Furthermore, the criteria for screening the noisy modal components are:

[0019] When the correlation coefficient R<0.3 or the variance contribution rate VCR≤0.01, it is judged that the modal component contains noise modal component;

[0020] Otherwise, the modal component does not contain the noise modal component.

[0021] Furthermore, the two indicators of root mean square error and smoothness are integrated to construct a composite evaluation index T, including:

[0022] The two indicators of root mean square error and smoothness are normalized to obtain normalized indicators;

[0023] The coefficient of variation weighting method is used to assign weights to the two indicators of normalized root mean square error and smoothness to obtain the weighted values.

[0024] According to the weighted values ​​and the normalized indicators, a composite evaluation indicator T is constructed using a linear combination method.

[0025] Furthermore, the composite evaluation index T is expressed as:

[0026] T=W PRMSE ×PRMSE+W Pr ×Pr

[0027]

[0028]

[0029] Among them, RMSE is the root mean square error, r is the smoothness, PRMSE and Pr are the normalized values ​​of the root mean square error and smoothness, respectively; σ is the standard deviation, μ is the mean, and W is the weight of the root mean square error and smoothness, respectively.

[0030] Furthermore, the composite evaluation index T is used to determine the optimal wavelet basis function and the number of decomposition layers, and the noisy modal components are decomposed and reconstructed by wavelet packets to obtain the denoised modal components; specifically:

[0031] Set the decomposition level of wavelet packet analysis to 2-5 levels, and change the wavelet basis function and the decomposition level;

[0032] When the composite evaluation index T takes the minimum value, the optimal wavelet basis function and the specific number of decomposition layers are obtained;

[0033] According to the optimal wavelet basis function and wavelet decomposition layer number, the noisy modal component is taken as the processing object and input into the wavelet basis function for wavelet packet decomposition and reconstruction, so as to obtain the modal component after noise reduction.

[0034] The present invention also provides a computer storage medium storing an executable program code, wherein the executable program code is used to execute a weld defect signal denoising method as described in any one of claims 1 to 7.

[0035] The present invention has the following beneficial effects:

[0036] (1) The present invention provides a weld defect signal denoising method based on Sparrow search algorithm (SSA)-Variational mode decomposition (VMD)-Wavelet packet analysis (WPD) (SSA-VMD-WPD), which introduces a composite evaluation index T to optimize the decomposition layer number and wavelet basis function of wavelet packet analysis, sets the decomposition layer number of wavelet packet analysis to 2-5 layers, and the wavelet basis functions include 54 types such as haar, db series, sym series, etc., changes the wavelet basis function and the decomposition layer number, calculates the value of the index T based on the signal to be processed, and when the T value is the minimum, determines the wavelet packet decomposition layer number and wavelet basis function corresponding to this value condition as the optimal parameters. Different from traditional indicators, the composite evaluation index T can measure the denoising effect from the detail information and approximation information. Among the traditional evaluation indicators, the signal-to-noise ratio, root mean square error, and cross-correlation coefficient can be used to judge the denoising effect from the detail information of the signal, while the smoothness can be measured from the approximation information of the signal. This paper adopts the composite evaluation index method, which can integrate the detail information and approximation information of the signal. At the same time, by adding Gaussian noise to the simulation signal to simulate the noisy signal, and comparing with the denoising methods such as SSA-VMD, WPD, and wavelet decomposition, the weld defect signal denoising method based on sparrow optimization-variational mode decomposition-wavelet packet analysis (SSA-VMD-WPD) proposed in this paper has better effect.

[0037] (2) By performing noise reduction analysis on the actual signals collected in the experiment, the method proposed in the present invention can effectively retain the detailed features of the ultrasonic phased array defect A-scan signal and has a good noise reduction effect. The noise reduction method proposed in the present invention is used to process all the signals collected in the experiment, providing a good data basis for subsequent research.

[0038] In addition to the above-described purposes, features and advantages, the present invention has other purposes, features and advantages. The present invention will be further described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0040] Figure 1 It is a flow chart of a weld defect signal denoising method of the present invention;

[0041] Figure 2 It is a simulation signal waveform diagram of the present invention;

[0042] Figure 3 It is a diagram of iterative optimization process of simulation signal of the present invention;

[0043] Figure 4 is the time-frequency diagram and spectrum diagram of each modal component of the VMD decomposition of the simulation signal of the present invention; wherein, Figure 4 (a) time-frequency diagram of modal components, (b) time-frequency diagram of modal components;

[0044] Figure 5 It is the actual signal waveform diagram of the present invention;

[0045] Figure 6 is the time-frequency diagram and spectrum diagram of each modal component of the VMD decomposition of the actual signal of the present invention; wherein, Figure 6 (a) time-frequency diagram of modal components, (b) time-frequency diagram of modal components;

[0046] Figure 7 It is a waveform diagram of the actual signal after noise reduction processing of the present invention. DETAILED DESCRIPTION

[0047] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0048] like Figure 1 As shown, the present invention provides a method for denoising a weld defect signal, and the specific steps are as follows:

[0049] Step 1: Scan a preset carbon steel weld by ultrasonic phased array to obtain weld scanning raw signal data of the carbon steel weld; the weld scanning raw signal data includes noise signal data. Specifically, the noise signal data includes test block material structure noise, equipment noise and environmental noise.

[0050] Step 2: Using envelope entropy as the fitness function, the sparrow optimization algorithm (SSA for short) is used to obtain the decomposition mode number K and the penalty factor α of the weld scanning raw signal data.

[0051] Specifically, before searching for the best solution, the parameters of the sparrow algorithm are initialized, and the minimum and maximum values ​​of the decomposition mode number K and the penalty factor α, the population size, and the maximum number of iterations are set. Secondly, the optimization algorithm is run to obtain the optimal parameters [K, α] and the corresponding minimum envelope entropy value.

[0052] Step 3: Perform variational mode decomposition (VMD) on the weld scanning signal data according to the decomposition mode number K and the penalty factor α to obtain the correlation coefficient and variance contribution rate of each modal component, and screen the noisy modal component to obtain the noisy modal component and the non-noisy modal component.

[0053] Among them, the criteria for screening noisy modal components are:

[0054] Correlation coefficient R<0.3; variance contribution rate VCR≤0.01. If any of the above conditions is met, it is judged that the modal component contains noise modal component;

[0055] Otherwise, the modal component does not contain the noise modal component.

[0056] Step 4: Combine the two indicators of root mean square error and smoothness to obtain the composite evaluation index T.

[0057] Specifically, the two indicators of root mean square error and smoothness are normalized to obtain normalized indicators;

[0058] The coefficient of variation weighting method is used to assign weights to the two indicators of normalized root mean square error and smoothness to obtain the weighted values.

[0059] According to the weighted values ​​and the normalized indicators, a composite evaluation indicator T is constructed using a linear combination method.

[0060] The denoising effect evaluation indicators are as follows: Traditional denoising effect evaluation indicators mainly include signal-to-noise ratio, root mean square error, correlation coefficient, variance contribution rate, smoothness, etc.

[0061] From the denoising principles of variational mode decomposition method, wavelet analysis, and wavelet packet analysis, it can be seen that the factors that affect the denoising effect are not unique, so it is necessary to select appropriate indicators to judge the denoising effect, so as to achieve the purpose of optimizing the denoising parameter selection and denoising method. The traditional signal denoising effect is generally judged by the monotonic relationship between a single indicator and the denoising effect, such as signal-to-noise ratio, root mean square error, smoothness, correlation coefficient, noise modulus, etc. However, in the actual denoising process, the pure signal is unknown, that is, the true value of the signal is unknown, and there are certain limitations at this time.

[0062] Therefore, it is considered to combine multiple indicators to construct a composite evaluation index, and to judge it by combining the detailed information and approximation information of the signal. In traditional evaluation indicators, signal-to-noise ratio, root mean square error, and cross-correlation coefficient can be used to judge the denoising effect from the detailed information of the signal, while smoothness can be measured from the approximation information of the signal. The present invention adopts the method of composite evaluation index T, and constructs it by combining the two indicators of root mean square error and smoothness.

[0063] The root mean square error is expressed as:

[0064]

[0065] Where f(t) is the original signal, f'(t) is the denoised signal, and n is the signal length. The root mean square error can reflect the difference between signals. For the denoised signal, the smaller the value, the better the denoising effect.

[0066] Smoothness is expressed as:

[0067]

[0068] Smoothness is an important indicator for judging the denoising effect, which can reflect the smoothness of the signal. After the signal is denoised, the smaller its smoothness is, the better the denoising effect is.

[0069] First, the two indicators are normalized:

[0070]

[0071] Where PRMSE and Pr are the normalized values ​​of RMSE and r respectively, max() is the maximum value operation, and min() is the minimum value operation.

[0072] Secondly, the coefficient of variation weighting method was used to assign weights to the indicators. The specific process is as follows:

[0073]

[0074] Among them, CV is the coefficient of variation, W is the weighted value, σ is the standard deviation, and μ is the mean. Secondly, the composite evaluation index T is constructed using the linear combination method, which shows that:

[0075] T=W PRMSE ×PRMSE+W Pr ×Pr

[0076] According to the relationship between each single indicator and the denoising effect, it can be seen that the smaller the value of the indicator T, the better the denoising effect of the signal.

[0077] Step 5: Use the composite evaluation index T to determine the optimal wavelet basis function and the number of decomposition layers, perform wavelet packet decomposition and reconstruction on the noisy modal component, and obtain the denoised modal component.

[0078] Specifically, the wavelet basis function and the number of decomposition layers are changed, the number of decomposition layers is 2-5 layers, the wavelet basis functions are haar, db series, sym series, etc., a total of 54 types, the wavelet basis function and the number of decomposition layers are changed, the composite evaluation index T is used to determine the optimal wavelet basis function and the number of decomposition layers, the value of the index T based on the signal to be processed is calculated, when the T value is the minimum, the number of wavelet packet decomposition layers and wavelet basis functions corresponding to this value condition are determined as the optimized parameters, and wavelet packet decomposition and reconstruction are performed for the noisy modal components, and wavelet packet analysis is performed with the wavelet packet decomposition layers and wavelet basis functions obtained after optimization as parameters, thereby obtaining the modal components after denoising.

[0079] Step 6: Reconstruct the noise-reduced modal components in combination with the non-noise-containing modal components, that is, reconstruct the noisy modal components and the non-noise-containing modal components after wavelet packet analysis, and finally obtain the noise-reduced signal.

[0080] The method of the present invention is used to carry out noise reduction research using simulated signals.

[0081] First, the Gaussian model is used to establish the ultrasonic echo signal model:

[0082]

[0083] A defect echo with 512 sampling points is set, where the amplitude coefficient β = 0.8, the bandwidth factor α = 50 (MHz)2, the center frequency fc = 5.0MHz, the echo arrival time τ = 2.5μs, and the phase φ = 8.89. Gaussian white noise with signal-to-noise ratios of 5dB, 10dB, and 15dB is added respectively, and noisy signals with signal-to-noise ratios of 5.2540dB, 10.0703dB, and 15.1890dB are obtained. The waveforms of the original signal and the noisy signal can be seen Figure 2 .

[0084] The envelope entropy is used as the fitness function, and the SSA algorithm is used to optimize the parameters. First, the parameters are initialized, and the minimum value of the decomposition mode number K and the penalty factor α is set to [200, 2], the maximum value is [3000, 10], the number of iterations is 20, and the population size is 30. Taking the noisy signal with a signal-to-noise ratio of 15.1890dB as an example, the iterative optimization process diagram can be seen Figure 3 .

[0085] Depend on Figure 3 It can be seen that during the iteration process, the fitness value drops sharply and the optimization effect is good. The parameters obtained by optimization are K = 4, α = 498. Based on this parameter, the signal is decomposed by VMD, and the variance contribution rate VCR of each modal component and its correlation coefficient R with the noisy simulation signal are solved. The results are shown in Table 1. The time-frequency diagram and spectrum diagram of each modal component decomposed by VMD are shown in Figure 4.

[0086] Table 1 Calculation results of VMD decomposition indexes for each modal component

[0087]

[0088] According to the screening criteria: the modal components with correlation coefficient R<0.3 or variance contribution rate VCR≤0.01 are judged as noisy components, and IMF3 and IMF4 are noisy modal components. Figure 4 From the analysis, it can be seen that the time-frequency diagrams of IMF1 and IMF2 are similar to the noisy simulation signal, and the energy of the two is concentrated near 5MHz, which is consistent with the center frequency set by the simulation signal. Therefore, IMF1 and IMF2 can be considered as non-noisy modal components, and IMF3 and IMF4 are noisy modal components. Here, IMF3 and IMF4 are further subjected to wavelet packet denoising. Taking T as the index, the calculation results of the index value T of IMF3 can be obtained through the denoising layer number of wavelet packets and the optimization of wavelet basis functions, as shown in Table 2.

[0089] Table 2 Processing results of different denoising methods

[0090]

[0091]

[0092]

[0093] According to Table 2, it is finally determined to use bior2.4 to perform three-layer wavelet packet decomposition on IMF3. Similarly, sym7 is selected through calculation to perform three-layer wavelet packet decomposition on IMF4, and finally the denoised signal is reconstructed.

[0094] In order to compare and verify the denoising performance of the method of the present invention, the denoising method of sparrow optimization algorithm-variational mode decomposition (abbreviated as SSA-VMD), the denoising method of wavelet packet decomposition (abbreviated as WPD), the wavelet analysis denoising method (Waveletdecomposition, abbreviated as WD), the denoising method based on sparrow optimization-variational mode decomposition-wavelet analysis (SSA-VMD-WD) and the denoising method based on sparrow optimization-variational mode decomposition-wavelet packet analysis (abbreviated as SSA-VMD-WPD) are adopted to process the signal and calculate its signal-to-noise ratio and root mean square error. Among them, when the signal is subjected to wavelet packet analysis and wavelet decomposition denoising, the composite evaluation index T is used to optimize the number of decomposition layers and wavelet basis functions. The results are shown in Table 3.

[0095] Table 3 Processing results of different denoising methods

[0096]

[0097] According to the results in Table 3, it can be seen that the denoising method based on sparrow optimization-variational mode decomposition-wavelet packet analysis SSA-VMD-WPD processing obtains the maximum signal-to-noise ratio and the minimum root mean square error. Therefore, the denoising effect of the method of the present invention is the best.

[0098] In order to further verify the effect of the SSA-VMD-WPD noise reduction method, noise reduction processing is performed on the actual defect echo signal obtained by experimental detection. Taking the unfused defect echo signal as an example, the noise reduction process and effect are introduced in detail. The actual signal waveform is shown in Figure 5 As shown. The SSA algorithm is used to optimize the parameters K = 10, α = 3000, and VMD is used for processing. The decomposition results are as follows Figure 6 The variance contribution rate of each modal component and the correlation coefficient with the noisy simulation signal are calculated, and the results are shown in Table 4.

[0099] Table 4 Calculation results of VMD decomposition indexes for each modal component

[0100]

[0101] It can be obtained from Table 4 that IMF1, IMF2, IMF3, and IMF4 are non-noisy modal components, and the remaining modal components are determined to be noisy modal components. The noisy modal components are subjected to the next step of calculation. The decomposition layer number and wavelet basis function of the wavelet packet are optimized based on the composite evaluation index T. The optimization result is used as the wavelet packet analysis parameter input to complete the wavelet packet analysis of the noisy modal component, and finally reconstructed with the non-noisy modal component to obtain the denoised signal. SSA-VMD, WPD, SSA-VMD-WAVE, and WAVE are used to process the signal and calculate its signal-to-noise ratio and root mean square error. Among them, when the signal is subjected to WPD and WAVE denoising, the composite evaluation index T is used to optimize the decomposition layer number and wavelet basis function. Through the processing and analysis based on the noisy simulation signal, the processing result of the method of the present invention obtains the maximum signal-to-noise ratio and the minimum root mean square error, so the denoising effect of the method of the present invention is the best. In addition, VMD and wavelet packet analysis are common signal denoising methods. This method combines the two denoising methods, but the parameters K and α of VMD decomposition, the number of decomposition layers of wavelet packet analysis, and the selection of wavelet basis functions are all problems that need to be solved. This method uses the sparrow optimization algorithm and introduces the indicator T to achieve the optimization of the above parameters, providing a method for the optimization of the above parameters.

[0102] The waveform results after noise reduction are shown in Figure 7 .Depend on Figure 7It can be seen that the signal after denoising based on SSA-VMD-WPD retains the detailed characteristics of the original signal of the defect better. Combined with the previous study on the denoising effect of the simulation signal, the maximum signal-to-noise ratio (21.3816) and the minimum root mean square error (0.0105) are obtained after SSA-VMD-WPD processing. Therefore, SSA-VMD-WPD has a good effect on the denoising of the A-scan signal of the ultrasonic phased array.

[0103] In summary, the present invention provides a weld defect signal denoising method based on sparrow optimization-variational modal decomposition-wavelet packet analysis, which introduces a composite evaluation index T to optimize the decomposition layer number and wavelet basis function of wavelet packet analysis, sets the decomposition layer number of wavelet packet analysis to 2-5 layers, and the wavelet basis function includes 54 types such as haar, db series, sym series, etc., changes the wavelet basis function and the decomposition layer number, calculates the value of the index T based on the signal to be processed, and when the T value is the minimum, determines the wavelet packet decomposition layer number and wavelet basis function corresponding to this value condition as the optimal post-parameter. Different from traditional indicators, the composite evaluation index T can measure the denoising effect from the detail information and approximation information. In traditional evaluation indicators, the signal-to-noise ratio, root mean square error, and mutual correlation coefficient can judge the denoising effect from the detail information of the signal, and the smoothness can be measured from the approximation information of the signal. This paper adopts the method of composite evaluation index, which can comprehensively consider the detail information and approximation information of the signal. At the same time, a noisy signal is simulated by adding Gaussian noise to the simulation signal, and a comparative study is conducted with denoising methods such as SSA-VMD, WPD, and wavelet decomposition. The SSA-VMD-WAVE denoising method proposed in the present invention has a better effect.

[0104] In addition, the weld defect signal denoising method provided by the present invention can effectively retain the detailed features of the ultrasonic phased array defect A-scan signal and has a good denoising effect. The denoising method proposed by the present invention is used to process all the signals collected in the experiment, providing a good data basis for subsequent research.

[0105] On the other hand, the present invention also provides a computer storage medium storing an executable program code; the executable program code is used to execute any of the above-mentioned weld defect signal denoising methods.

[0106] Exemplarily, the program code may be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to accomplish the present invention.

[0107] The memory may be an internal storage unit of the terminal device, such as a hard disk or a memory. The memory may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal device. Further, the memory may also include both an internal storage unit of the terminal device and an external storage device. The memory is used to store the program code and other programs and data required by the terminal device. The memory may also be used to temporarily store data that has been output or is to be output.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A weld defect signal denoising method, characterized in that: The following steps are involved: Scanning a preset carbon steel weld by an ultrasonic phased array to obtain weld scanning raw signal data of the carbon steel weld; the weld scanning raw signal data includes noise signal data; Using envelope entropy as a fitness function, a sparrow optimization algorithm is used to obtain the decomposition mode number K and the penalty factor α of the weld scanning raw signal data; Performing variational modal decomposition processing on the weld scanning signal data according to the decomposition modal number K and the penalty factor α to obtain the correlation coefficient and variance contribution rate of each modal component, and screening the noisy modal component to obtain the noisy modal component and the non-noisy modal component; The composite evaluation index T is constructed by combining the two indicators of root mean square error and smoothness; The optimal wavelet basis function and the number of decomposition layers are determined by using the composite evaluation index T, and the noisy modal components are decomposed and reconstructed using wavelet packets to obtain the modal components after noise reduction; The denoised signal can be obtained by reconstructing the denoised modal components in combination with the non-noised modal components.

2. A weld defect signal denoising method according to claim 1, characterized in that: The noise signal data includes test block material structure noise, equipment noise and environmental noise.

3. A weld defect signal denoising method according to claim 1, characterized in that: Taking envelope entropy as the fitness function, the sparrow optimization algorithm is used to obtain the decomposition mode number K and the penalty factor α; specifically: Initialize the parameters of the sparrow optimization algorithm; Set the minimum and maximum values ​​of the decomposition mode number K and the penalty factor α, the population size, and the maximum number of iterations; Run the sparrow optimization algorithm to obtain the optimal parameter decomposition mode number K and penalty factor α, and the corresponding minimum envelope entropy value.

4. A weld defect signal denoising method according to claim 1, characterized in that: The criteria for screening noisy modal components are: When the correlation coefficient R<0.3 or the variance contribution rate VCR≤0.01, it is judged that the modal component contains noise modal component; Otherwise, the modal component does not contain the noise modal component.

5. A weld defect signal denoising method according to claim 1, characterized in that: The composite evaluation index T is constructed by combining the two indicators of root mean square error and smoothness, including: The two indicators of root mean square error and smoothness are normalized to obtain normalized indicators; The coefficient of variation weighting method is used to assign weights to the two indicators of normalized root mean square error and smoothness to obtain the weighted values. According to the weighted values ​​and the normalized indicators, a composite evaluation indicator T is constructed using a linear combination method.

6. A weld defect signal denoising method according to claim 1, characterized in that: The composite evaluation index T is expressed as: T=W PRMSE ×PRMSE+W Pr ×Pr Among them, RMSE is the root mean square error, r is the smoothness, PRMSE and Pr are the normalized values ​​of the root mean square error and smoothness, respectively; σ is the standard deviation, μ is the mean, and W is the weight of the root mean square error and smoothness, respectively.

7. A weld defect signal denoising method according to claim 1, characterized in that: The composite evaluation index T is used to determine the optimal wavelet basis function and the number of decomposition layers, and the noisy modal components are decomposed and reconstructed by wavelet packets to obtain the denoised modal components; specifically: Set the decomposition level of wavelet packet analysis to 2-5 levels, and change the wavelet basis function and the decomposition level; When the composite evaluation index T takes the minimum value, the optimal wavelet basis function and the specific number of decomposition layers are obtained; According to the optimal wavelet basis function and wavelet decomposition layer number, the noisy modal component is taken as the processing object and input into the wavelet basis function for wavelet packet decomposition and reconstruction, so as to obtain the modal component after noise reduction.

8. A computer storage medium storing executable program code, characterized in that: The executable program code is used to execute a weld defect signal denoising method as described in any one of claims 1-7.

Images