Full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method

Abstract

The present invention belongs to the field of non-destructive testing technology, and in particular relates to a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method. The leakage defect three-dimensional reconstruction method realizes the precise positioning and three-dimensional morphological reconstruction of tiny complex defects in austenitic thin-walled stainless steel materials, and provides technical support for the detection and three-dimensional characterization of leakage micropores in austenitic thin-walled stainless steel materials. The leakage micropore three-dimensional reconstruction method includes the following steps: recording the A-Scan original signal and position information of each path point to obtain full-matrix micro-ultrasonic data; taking the minimum envelope entropy as the optimization goal, performing variational modal decomposition and reconstruction on the A-Scan original signal of each path point, and calculating the optimal solution of the variational model; performing two-dimensional mask extraction on the processed full-matrix micro-ultrasonic data to generate C-scan images of different depths; extracting two-dimensional defect edge masks at different depths; and splicing and volume rendering the two-dimensional defect edge mask results.

Description

Technical Field

[0001] The present invention belongs to the technical field of nondestructive testing, and in particular relates to a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method. Background Art

[0002] Austenitic thin-walled stainless steel, a new metal material with excellent mechanical properties, good wear resistance, and heat insulation, is widely used in the chemical industry, architectural decoration, and marine engineering fields. However, in actual use, due to the influence of the conveying medium and the operating environment, austenitic thin-walled stainless steel can cause leakage defects due to corrosion. To prevent leakage of conveying media, technicians must promptly detect defects in austenitic thin-walled stainless steel before leakage (micropores) occurs or at the early stage of leakage.

[0003] Further research revealed that, due to the unique physical properties of austenitic thin-walled stainless steel (such as its large and unevenly distributed internal grains and its significant anisotropy), existing ultrasonic defect detection technologies face numerous challenges in defect identification, characterization, and location. For example, the high degree of internal grain heterogeneity in austenitic thin-walled stainless steel welds results in significant scattering noise during ultrasonic propagation, significantly reducing the signal-to-noise ratio (SNR) of the detection signal. Furthermore, because the scattering noise of austenitic thin-walled stainless steel has a complex frequency distribution and non-stationary characteristics, traditional defect detection techniques typically rely on preset parameters or specific filter designs, making it difficult to effectively address this dynamically changing noise. Finally, the small size and irregular shape of leaking micropores in austenitic thin-walled stainless steel further lead to image artifacts during C-scan imaging. This makes it difficult for existing ultrasonic defect detection techniques to accurately extract the target regional features, severely impacting the accuracy of 3D reconstruction of the leak defect. Summary of the Invention

[0004] The present invention provides a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method, which realizes the precise positioning and three-dimensional morphological reconstruction of tiny and complex defects in austenitic thin-walled stainless steel materials, and provides technical support for the detection and three-dimensional characterization of leakage micropores in austenitic thin-walled stainless steel materials.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] The present invention provides a method for three-dimensional reconstruction of leaking micropores based on full-matrix micro-ultrasound, comprising the following steps:

[0007] Step P1: Use a point-focused micro-ultrasonic probe to perform surface scanning on the test piece; record the A-Scan original signal and position information of each path point to obtain full-matrix micro-ultrasonic data;

[0008] Step P2: Perform variational mode decomposition on the A-Scan original signal of each path point and decompose it into a series of intrinsic mode function components;

[0009] Step P3: Perform a Hilbert transform on each intrinsic mode function component to obtain its single-sided spectrum; modulate the frequency of each modal spectrum analysis signal to the corresponding baseband based on the mixed-estimated center frequency of each modal spectrum analysis signal; and estimate the sum of the bandwidths of each modal spectrum analysis signal;

[0010] Step P4: Introduce the augmented Lagrangian function to solve the variational model reconstructed after variational mode decomposition; obtain the saddle point through the ADMM algorithm and calculate the optimal solution of the variational model;

[0011] Step P5: Using the adaptive sparrow search algorithm, with the minimum envelope entropy as the optimization goal, repeat steps P2 to P4 to perform variational modal decomposition and reconstruction on the A-Scan raw signal at each path point until the full matrix micro-ultrasound data processing is completed. The optimal parameters of the variational modal decomposition are obtained by iterative solution.

[0012] Step P6: Perform two-dimensional mask extraction on the full-matrix micro-ultrasound data processed in step P5, and convert the full-matrix micro-ultrasound data into C-scan images of different depths layer by layer; smooth the C-scan images using a Gaussian filter, and use the first-order differential coefficient to obtain the gradient intensity and direction of the C-scan images;

[0013] Step P7: determining an evaluation function of a first double threshold based on the gradient magnitude histogram and the minimum intra-class variance; determining an evaluation function of a second double threshold based on the gradient magnitude histogram and the maximum intra-class variance;

[0014] Step P8: extracting a defect mask image within the threshold based on the first double-threshold evaluation function and the second double-threshold evaluation function;

[0015] Step P9: performing edge detection on the acquired defect mask image layer by layer to obtain two-dimensional defect edge masks at different depths;

[0016] Step P10: perform splicing and volume rendering on the two-dimensional defect edge mask results to achieve three-dimensional reconstruction of the defect structure of the test piece to be inspected.

[0017] Preferably, the process of estimating the sum of the bandwidths of the modal spectrum analysis signals in step P3 is specifically described as follows:

[0018] While ensuring that the decomposition sequence is a modal component with a limited bandwidth and a center frequency, the sum of the bandwidths of the spectral analysis signals of each modal is minimized; wherein, the variational problem subject to the constraints satisfies:

[0019] Formula (4);

[0020] In formula (4), K is the number of modes to be decomposed; 、 They correspond to the kth modal component and center frequency after decomposition respectively; is the unit pulse function; is the convolution operator; In order to seek partial guidance; To find the analytical signal of the original function; It represents the shift of the spectrum to the baseband; j is the imaginary unit.

[0021] Preferably, the process of solving the variational model reconstructed after variational mode decomposition in step P4 is specifically described as follows:

[0022] The mixed-estimated center frequency and baseband of each modal spectrum analysis signal are continuously updated until the iteration meets the stopping condition: Formula (7);

[0023] Solving equation (7) yields Perform inverse Fourier transform to obtain the variational model reconstructed after variational mode decomposition.

[0024] Preferably, the process of extracting the defect mask image within the threshold in step P8 is specifically described as follows:

[0025] When the mask extraction algorithm based on Otsu threshold segmentation cannot extract the main body of the target defect in the defect mask image, the mask segmentation algorithm based on SAM is used to train the defect mask image by labeling the target defect area and the non-target defect area, and the target defect mask image is accurately obtained layer by layer.

[0026] The present invention provides a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method. Specifically, the leakage defect three-dimensional reconstruction method includes the following steps: recording the A-Scan raw signal and position information of each path point to obtain full-matrix micro-ultrasonic data; performing variational modal decomposition and reconstruction on the A-Scan raw signal of each path point with minimum envelope entropy as the optimization goal, and calculating the optimal solution of the variational model; performing two-dimensional mask extraction on the processed full-matrix micro-ultrasonic data to generate C-scan images at different depths; extracting two-dimensional defect edge masks at different depths; and splicing and volume rendering the two-dimensional defect edge mask results.

[0027] The full-matrix micro-ultrasound-based three-dimensional reconstruction method for leaky micropores with the above-mentioned steps has at least the following technical advantages over the existing technology:

[0028] (1) The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention realizes the effective extraction of defect information of austenitic thin-walled stainless steel materials under high noise background, and realizes the transformation of ultrasonic data from one dimension to two dimensions to three dimensions;

[0029] (2) The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention solves the technical defects of the prior art in that the scattering noise of austenitic thin-walled stainless steel materials is large and the defects are difficult to accurately characterize. It is suitable for defect detection of various types of leakage in a high noise background;

[0030] (3) The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention realizes the three-dimensional reconstruction of the leakage defect structure of the austenitic thin-walled stainless steel material to be tested, which provides assistance for the accurate characterization and analysis of the leakage defect. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the following drawings:

[0032] Figure 1 A schematic diagram of the process of the full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention;

[0033] Figure 2 This is a schematic diagram of a point-focused micro-ultrasonic probe performing surface scanning on a test piece;

[0034] Figure 3 This is a physical picture of the stainless steel cladding layer defect test piece;

[0035] Figure 4 Design drawing for stainless steel cladding layer defect test piece;

[0036] Figure 5 This is a physical picture of the stainless steel base material defect specimen;

[0037] Figure 6 Design drawing for stainless steel base material defect specimen;

[0038] Figure 7 Schematic diagram of the convergence curve of minimum envelope entropy;

[0039] Figure 8a Schematic diagram of the variational modal decomposition results of the stainless steel weld overlay defect specimen;

[0040] Figure 8b Schematic diagram of the variational modal decomposition results of the stainless steel base material defect specimen;

[0041] Figure 9a for Figure 8a Schematic diagram of the effect after denoising;

[0042] Figure 9b for Figure 8b Schematic diagram of the effect after denoising;

[0043] Figure 10a This is a schematic diagram of the C-scan image of the near-surface layer of a stainless steel cladding defect specimen;

[0044] Figure 10b This is a schematic diagram of the C-scan image of the crack layer of the stainless steel cladding defect specimen;

[0045] Figure 10c This is a schematic diagram of the C-scan image of the rectangular machining groove layer of the stainless steel cladding layer defect specimen;

[0046] Figure 10d This is a schematic diagram of the C-scan image of the bottom surface layer of the stainless steel cladding layer defect specimen;

[0047] Figure 11a This is a schematic diagram of a C-scan image of the surface layer of a stainless steel base material defect specimen;

[0048] Figure 11b This is a schematic diagram of a C-scan image of the near-surface layer of a stainless steel base material defect specimen;

[0049] Figure 11c This is a schematic diagram of a C-scan image of the leaking micropores inside a stainless steel base material defect specimen;

[0050] Figure 11d This is a schematic diagram of a C-scan image of the bottom surface layer of a stainless steel base material defect specimen;

[0051] Figure 12 This is the defect mask effect image obtained by the mask extraction algorithm based on Otsu threshold segmentation;

[0052] Figure 13 This is the defect mask effect image obtained by the SAM-based mask segmentation algorithm;

[0053] Figure 14 Schematic diagram of the reconstruction results obtained by reconstructing the stainless steel cladding layer defect specimen using the existing technology;

[0054] Figure 15 Schematic diagram of the reconstruction result obtained by reconstructing a stainless steel cladding layer defect specimen using the reconstruction method of the present invention;

[0055] Figure 16Schematic diagram of the reconstruction results obtained by reconstructing a stainless steel base material defect specimen using existing technology;

[0056] Figure 17 Schematic diagram of the reconstruction results obtained by reconstructing a stainless steel base material defect specimen using the reconstruction method of the present invention. DETAILED DESCRIPTION

[0057] The present invention provides a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method, which realizes the precise positioning and three-dimensional morphological reconstruction of tiny and complex defects in austenitic thin-walled stainless steel materials, and provides technical support for the detection and three-dimensional characterization of leakage micropores in austenitic thin-walled stainless steel materials.

[0058] like Figure 1 As shown, the present invention provides a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method, which includes the following steps:

[0059] Step P1: Use a point-focused micro-ultrasonic probe to perform surface scanning on the test piece. Record the A-Scan raw signal and position information of each path point to obtain full-matrix micro-ultrasonic data.

[0060] For details, please refer to Figure 2 As shown, Figure 2 The schematic diagram of the detection of the surface scanning detection of the test piece by the point-focused micro-ultrasonic probe is shown in FIG. Through step P1, the full matrix micro-ultrasonic data recording the A-Scan original signal and position information of each path point is obtained.

[0061] To verify the reconstruction effect of the full-matrix micro-ultrasonic leakage defect 3D reconstruction method provided by the present invention, a set of defective specimens are provided to simulate the potential leakage crack defects of the cladding layer (using a 5Mhz probe). The actual image and design drawing of the defective specimens are as follows: Figure 3 、 Figure 4 As shown in the figure, a rectangular groove is machined on the back of the specimen; then an artificial crack defect is machined in the groove, with a length of 3mm, a width of 0.3mm, and a defect burial depth of 4mm. In addition, a set of stainless steel base material defect specimens are provided to simulate base material micropore leakage (using a 15Mhz probe). The actual picture and design drawing of the defect specimen are shown in the figure. Figure 5 、 Figure 6 As shown in FIG, the specimen is completed by metal 3D printing, the thickness of the stainless steel sheet base material is 4 mm, and the aperture of the spiral leakage micropore is 0.8 mm.

[0062] Step P2: Perform variational mode decomposition on the A-Scan original signal of each path point and decompose it into a series of intrinsic mode function components.

[0063] After completing step P1, step P2 is further implemented to perform variational modal decomposition on the original A-Scan signal at each path point. Specifically, the purpose of performing variational modal decomposition on the original A-Scan signal at each path point is to decompose the original A-Scan signal (input signal) at each path point into multiple intrinsic mode function components (output sub-signals).

[0064] In this process, the intrinsic mode function components ,satisfy: Formula (1);

[0065] In formula (1), is the instantaneous amplitude, is the phase angle.

[0066] Step P3: Perform a Hilbert transform on each intrinsic mode function component to obtain its single-sided spectrum. Using the mixed-estimated center frequency of each modal spectrum analysis signal as a reference, modulate the frequency of each modal spectrum analysis signal to the corresponding baseband, and estimate the sum of the bandwidths of each modal spectrum analysis signal.

[0067] After completing step P2, step P3 is further implemented to estimate the sum of the bandwidths of the spectral analysis signals of each modal. Performing Hilbert transform, the obtained single-sided spectrum can be expressed as:

[0068] Formula (2);

[0069] In formula (2), 、 .

[0070] Then, the mixed-estimated center frequency of each modal spectrum analysis signal is obtained. As a benchmark, the frequency of each modal spectrum analysis signal is modulated to the corresponding baseband. The modulated baseband can be expressed as:

[0071] Formula (3);

[0072] In formula (3), is the vector description of the center frequency on the complex plane, is the center frequency.

[0073] Then, the sum of the bandwidths of the spectral resolution signals of each modality is estimated. As an alternative embodiment, the L² norm of the squared gradient of the modulated signal is calculated to estimate the sum of the bandwidths of the spectral resolution signals of each modality.

[0074] It is worth noting that, as a more preferred embodiment, it is preferred to minimize the sum of the bandwidths of the spectrum analysis signals of each modal while ensuring that the decomposition sequence is a modal component with a limited bandwidth and a center frequency. In this case, the variational problem under the constraint condition satisfies:

[0075] Formula (4);

[0076] In formula (4), K is the number of modes to be decomposed; They correspond to the kth modal component and center frequency after decomposition respectively; is the unit pulse function;* is the convolution operator; In order to seek partial guidance; To find the analytical signal of the original function; It represents the shift of the spectrum to the baseband; j is the imaginary unit.

[0077] Step P4: Introduce the augmented Lagrangian function to solve the variational model reconstructed after variational mode decomposition; obtain the saddle point through the ADMM algorithm, and calculate the optimal solution of the variational model.

[0078] After completing step P3, step P4 is further implemented to calculate the optimal solution of the variational model. Specifically, the purpose of introducing the augmented Lagrangian function to solve the variational model reconstructed after variational mode decomposition is to transform the constrained variational problem into an unconstrained variational problem.

[0079] Among them, as a more preferred embodiment of the present invention, the process of solving the variational model reconstructed after variational mode decomposition is specifically described as follows:

[0080] The mixed-estimated center frequency and baseband of each modal spectrum analysis signal are continuously updated until the iteration meets the stopping condition: Formula (7);

[0081] Solving equation (7) yields ( Specific satisfaction, ) performs inverse Fourier transform to obtain the variational model reconstructed after variational mode decomposition.

[0082] It is worth noting that in the process of continuous updating and iteration, the variational model update equation and center frequency can be expressed as:

[0083] Formula (5); Formula (6); In formulas (5) and (6), is the Fourier transform, n is the number of iterations, is the fidelity coefficient, is the augmented Lagrange multiplication operator, is the penalty coefficient.

[0084] Step P5: Using the adaptive sparrow search algorithm and taking the minimum envelope entropy as the optimization goal, repeat steps P2 to P4 to perform variational modal decomposition and reconstruction on the A-Scan original signal at each path point until the full-matrix micro-ultrasound data processing is completed. The optimal parameters of the variational modal decomposition are obtained by iterative solution.

[0085] After completing step P4, further implement step P5 to iteratively solve the optimal parameters of variational mode decomposition. The convergence curve of minimum envelope entropy can be referred to as Figure 7 As shown in the figure, the results of variational modal decomposition of stainless steel weld overlay defect specimens and stainless steel base material defect specimens can be referred to as Figure 8a Figure 8b As shown. Figure 8a 、 Figure 8b For adaptive denoising, the denoising effect can be found in the following examples: Figure 9a Figure 9b As shown in the figure, it is worth noting that by performing variational modal decomposition and reconstruction on the A-Scan raw signal at each path point, noise signals unrelated to the useful signal in the full-matrix micro-ultrasound data can be effectively removed (noise components are accurately separated from useful signal components in the frequency domain), avoiding the signal distortion or residual noise problems caused by fixed filter bandwidth in existing technologies. In addition, the optimal parameters are calculated through iterative adaptive calculation without manual parameter setting and error, achieving adaptive denoising of echo signals from highly scattering noise materials and preventing interference with subsequent calculations.

[0086] Step P6: Perform two-dimensional mask extraction on the full-matrix micro-ultrasound data processed in step P5, and convert the full-matrix micro-ultrasound data into C-scan images of different depths layer by layer; use a Gaussian filter to smooth the C-scan image, and use the first-order differential coefficient to obtain the gradient intensity and direction of the C-scan image.

[0087] On the basis of completing step P5, step P6 is further implemented to perform two-dimensional mask extraction on the full-matrix micro-ultrasound data. In this process, one point that needs to be additionally explained is that in the process of using the Gaussian filter to smooth the C-scan image, the Gaussian filter can effectively remove the two-dimensional noise therein, thereby performing non-maximum suppression on the gradient direction of the gradient amplitude matrix of the C-scan image and refining the edge. The processing result can roughly divide the gradient amplitude values ​​of the C-scan image into the following three categories: C0 (low gradient amplitude, belonging to non-edge points), C1 (medium gradient amplitude, may be an edge point) and C2 (high gradient amplitude, belonging to an edge point). Among them, the C-scan images of stainless steel weld overlay defect specimens at different depths (near surface layer, crack layer, rectangular processing groove layer, bottom layer) can be referred to as follows. Figure 10a-Figure 10d As shown in the figure, the C-scan images of stainless steel base material defect specimens at different depths (surface layer, near surface layer, inside of leakage micropores, near bottom layer) can be referred to as follows: Figure 11a-Figure 11d shown.

[0088] Step P7: Based on the gradient magnitude histogram and the minimum intra-class variance, determine the evaluation function of the first double threshold; based on the gradient magnitude histogram and the maximum intra-class variance, determine the evaluation function of the second double threshold.

[0089] As a preferred embodiment of the present invention, the following first dual-threshold evaluation function and second dual-threshold evaluation function are provided as examples. The first dual-threshold evaluation function is determined based on the gradient magnitude histogram and the minimum intra-class variance, and satisfies: Formula (8);

[0090] In formula (8), , , , is the ratio of non-edge points to total pixels, is the ratio of suspected edge points to the total pixels, is the ratio of edge points to total pixels, is the variance within each class.

[0091] Based on the gradient magnitude histogram and the maximum intra-class variance, the evaluation function of the second double threshold is determined to satisfy: Formula (9);

[0092] In formula (9), is the grayscale mean of each class j, is the grayscale mean of the entire image.

[0093] One thing to point out is that the first double threshold evaluation function and a second double threshold evaluation function The parameters k and m therein respectively refer to the optimal thresholds of statistical significance. That is to say, by traversing all possible k and m (where k < m), the k and m that optimize the variational mode decomposition are found, and then the low threshold thrLow and the high threshold thrHigh can be obtained respectively.

[0094] Step P8: Based on the evaluation function of the first double threshold and the evaluation function of the second double threshold, extract the defect mask image within the threshold.

[0095] On the basis of completing Step P7, Step P8 is further implemented to extract the defect mask image within the threshold. As a relatively preferred implementation manner of the present invention, the process of extracting the defect mask image within the threshold in this Step P8 is specifically described as follows:

[0096] When the mask extraction algorithm based on Otsu threshold segmentation cannot extract the main body of the target defect in the defect mask image (that is, the C-scan image has a complex effect and there is interference), the mask segmentation algorithm based on SAM is used. The defect mask image is trained by labeling the target defect area and the non-target defect area, and the target defect mask image is accurately obtained layer by layer.

[0097] It should be noted that the reason for using the mask segmentation algorithm based on SAM is that in the case where the defect main body is not clear enough, the traditional mask extraction algorithm based on Otsu threshold segmentation will be greatly affected. As Figure 12 shown, Figure 12 is the defect mask effect diagram obtained by the mask extraction algorithm based on Otsu threshold segmentation. It can be found that for the stainless steel base material defect specimen, due to the complex morphology of its spiral micropores and the continuous influence of the surface inlet holes on the data of each layer in the tomography, the mask extraction algorithm based on Otsu threshold segmentation is no longer used, but the mask segmentation algorithm based on SAM is used instead.

[0098] Among them, the internal spiral holes of the stainless steel base material can be simply divided into four regions: background, upper surface micropore inlet, effective range of a certain layer slice, and artifacts. Therefore, the main target of detection is to extract the data within the effective range of the slice. Specifically, the SAM model is used to mark this region, and the non-interested regions are also removed by labeling, thus completing the mask extraction of the stainless steel base material defect specimen by the mask segmentation algorithm based on SAM, as shown in Figure 13 shown.

[0099] Step P9: Perform edge detection on the obtained defect mask image layer by layer to obtain two-dimensional defect edge masks at different depths.

[0100] After completing step P8, step P9 is further performed to obtain a two-dimensional defect edge mask at different depths. Specifically, the two-dimensional defect edge mask is used to construct the defect volume. Edge detection can be performed using the Canny operator.

[0101] Step P10: perform splicing and volume rendering on the two-dimensional defect edge mask results to achieve three-dimensional reconstruction of the defect structure of the test piece to be inspected.

[0102] On the basis of completing step P9, step P10 is further implemented to perform three-dimensional reconstruction of the defect structure of the test piece to be inspected. As an optional implementation method, the following three-dimensional reconstruction method is provided here for illustration. First, the smart pointer "vtkPNGReader" is used to read the slice (sequence image) data. Among them, the slice data is an image sequence of a two-dimensional target defect mask, and each slice corresponds to a cross-sectional view at a specific position. Assuming that each slice is a two-dimensional matrix, the slice sequence can be represented as a three-dimensional array. Among them, z is the index of the slice and N is the number of slice sequences. Then, the pixel interval is set, and the smart pointer filter "vtkContourFilter" is used to take the image data set as input and output isosurfaces or isolines. Next, the extracted isosurface is mapped to a triangular mesh, which can be further processed (such as smoothing, coloring), and then the final visualization result is generated through the lighting model and camera settings.

[0103] For details, please refer to Figure 14-17 As shown, Figure 14 Schematic diagram of the reconstruction results obtained by reconstructing the stainless steel cladding layer defect specimen using the existing technology, as shown in Figure 15 Schematic diagram of the reconstruction result obtained by reconstructing a stainless steel cladding layer defect specimen using the reconstruction method of the present invention; Figure 16 Schematic diagram of the reconstruction results obtained by reconstructing the stainless steel base material defect specimen using the existing technology, as shown in Figure 17 A schematic diagram of the reconstruction results obtained by reconstructing a stainless steel base material defect specimen using the reconstruction method of the present invention. A comparison reveals that, through the above process, the full-matrix micro-ultrasound-based 3D reconstruction method for leaky micropores provided by the present invention effectively extracts defect information in high-noise environments, transforming signals from one dimension to two dimensions and then to three dimensions. This addresses the issue of high scattering noise and difficulty in accurately characterizing defects in thin-walled austenitic stainless steel. The 3D reconstruction results are significantly improved compared to existing technologies, making it applicable to the detection and 3D characterization of various defects in high-noise environments.

[0104] The present invention provides a full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method. Specifically, the leakage defect three-dimensional reconstruction method includes the following steps: recording the A-Scan raw signal and position information of each path point to obtain full-matrix micro-ultrasonic data; performing variational modal decomposition and reconstruction on the A-Scan raw signal of each path point with minimum envelope entropy as the optimization goal, and calculating the optimal solution of the variational model; performing two-dimensional mask extraction on the processed full-matrix micro-ultrasonic data to generate C-scan images at different depths; extracting two-dimensional defect edge masks at different depths; and splicing and volume rendering the two-dimensional defect edge mask results.

[0105] The full-matrix micro-ultrasound-based 3D reconstruction method for leaky micropores with the above-mentioned steps has at least the following technical advantages over the existing technology:

[0106] (1) The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention realizes the effective extraction of defect information of austenitic thin-walled stainless steel materials under high noise background, and realizes the transformation of ultrasonic data from one dimension to two dimensions to three dimensions;

[0107] (2) The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention solves the technical defects of the prior art in that the scattering noise of austenitic thin-walled stainless steel materials is large and the defects are difficult to accurately characterize. It is suitable for defect detection of various types of leakage in a high noise background;

[0108] (3) The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method provided by the present invention realizes the three-dimensional reconstruction of the leakage defect structure of the austenitic thin-walled stainless steel material to be tested, which provides assistance for the accurate characterization and analysis of the leakage defect.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A full-matrix micro-ultrasonic leakage defect 3D reconstruction method, characterized by: The following steps are included: Step P1: Use a point-focused micro-ultrasonic probe to perform surface scanning on the test piece; record the A-Scan original signal and position information of each path point to obtain full-matrix micro-ultrasonic data; Step P2: Perform variational mode decomposition on the A-Scan original signal of each path point and decompose it into a series of intrinsic mode function components; Step P3: Perform a Hilbert transform on each intrinsic mode function component to obtain its single-sided spectrum; modulate the frequency of each modal spectrum analysis signal to the corresponding baseband based on the mixed-estimated center frequency of each modal spectrum analysis signal; and estimate the sum of the bandwidths of each modal spectrum analysis signal; Step P4: Introduce the augmented Lagrangian function to solve the variational model reconstructed after variational mode decomposition; obtain the saddle point through the ADMM algorithm and calculate the optimal solution of the variational model; Step P5: Using the adaptive sparrow search algorithm, with the minimum envelope entropy as the optimization goal, repeat steps P2 to P4 to perform variational modal decomposition and reconstruction on the A-Scan raw signal at each path point until the full matrix micro-ultrasound data processing is completed. The optimal parameters of the variational modal decomposition are obtained by iterative solution. Step P6: Perform two-dimensional mask extraction on the full-matrix micro-ultrasound data processed in step P5, and convert the full-matrix micro-ultrasound data into C-scan images of different depths layer by layer; smooth the C-scan images using a Gaussian filter, and use the first-order differential coefficient to obtain the gradient intensity and direction of the C-scan images; Step P7: determining an evaluation function of a first double threshold based on the gradient magnitude histogram and the minimum intra-class variance; Determine the evaluation function of the second double threshold based on the gradient magnitude histogram and the maximum intra-class variance; Step P8: extracting a defect mask image within the threshold based on the first double-threshold evaluation function and the second double-threshold evaluation function; Step P9: performing edge detection on the acquired defect mask image layer by layer to obtain two-dimensional defect edge masks at different depths; Step P10: perform splicing and volume rendering on the two-dimensional defect edge mask results to achieve three-dimensional reconstruction of the defect structure of the test piece to be inspected.

2. The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method according to claim 1 is characterized in that: The process of estimating the sum of the bandwidths of the modal spectrum analysis signals in step P3 is specifically described as follows: While ensuring that the decomposition sequence is a modal component with a limited bandwidth and a center frequency, the sum of the bandwidths of the spectrum analysis signals of each modal is minimized; Among them, the variational problem subject to constraints satisfies: Formula (4); In formula (4), K is the number of modes to be decomposed; Corresponding to the decomposition modal components and center frequencies; is the unit pulse function; is the convolution operator; In order to seek partial guidance; To find the analytical signal of the original function; It represents the shift of the spectrum to the baseband; j is the imaginary unit.

3. The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method according to claim 1, characterized in that: The process of solving the variational model reconstructed after variational mode decomposition in step P4 is specifically described as follows: The mixed-estimated center frequency and baseband of each modal spectrum analysis signal are continuously updated until the iteration meets the stopping condition: Formula (7); Solving equation (7) yields Perform inverse Fourier transform to obtain the variational model reconstructed after variational mode decomposition.

4. The full-matrix micro-ultrasonic leakage defect three-dimensional reconstruction method according to claim 1, characterized in that: The process of extracting the defect mask image within the threshold in step P8 is specifically described as follows: When the mask extraction algorithm based on Otsu threshold segmentation cannot extract the main body of the target defect in the defect mask image, the mask segmentation algorithm based on SAM is used to train the defect mask image by labeling the target defect area and the non-target defect area, and the target defect mask image is accurately obtained layer by layer.

Images