Nondestructive testing method and system based on bimodal cooperation

Abstract

The invention relates to the related technical field of ultrasonic nondestructive testing, and provides a nondestructive testing method and system based on bimodal collaboration.The method comprises the steps that a probe is used for emitting scanning signals, the scanning signals are made to obliquely enter a functional gradient laser cladding layer in a pulse mode, and the scanning signals comprise surface waves and transverse waves; receiving a scattered signal from a defect or an interface; performing continuous wavelet transform on the scattering signal to extract a time domain envelope and a dominant frequency component of the scattering signal; aligning the arrival moments of the defect main lobes of the surface wave mode and the transverse wave mode in a time domain; performing weighted stacking on complementary frequency bands of the surface wave mode and the transverse wave mode in a frequency domain to obtain a complex domain fusion signal; and performing degradation correction on the complex domain fusion signal based on a point spread function generated by analysis of the bimodal scattering model and a Richardson-Luxi iterative deconvolution algorithm to output a curve three-dimensional image.

Description

Technical Field

[0001] This application relates to the field of ultrasonic nondestructive testing technology, and in particular to a nondestructive testing method and system based on dual-modal collaboration. Background Technology

[0002] With the continuous development of materials surface engineering technology, laser cladding technology has become one of the key means to improve the surface performance of metal parts. By melting and rapidly solidifying powder or wire under the action of a high-energy laser beam, laser cladding can form a high-hardness, wear-resistant, and corrosion-resistant coating on the substrate surface, achieving a significant enhancement of the surface performance of parts. In particular, the introduction of functionally graded materials (FGMs) allows the cladding layer to exhibit continuous changes in chemical composition and mechanical properties, not only meeting the dual requirements of surface protection and high-strength support in extreme environments, but also greatly extending the service life of critical components. However, the complexity of the multilayer heterogeneous structure and microcrystalline structure of functionally graded laser cladding layers also brings greater challenges to subsequent quality assessment and defect detection.

[0003] Currently, non-destructive testing of functionally graded laser cladding layers mainly relies on methods such as thermal imaging, X-ray imaging, defect imaging, and single-mode ultrasonic testing. While thermal and X-ray imaging can achieve rapid scanning of the overall structure, they struggle to provide high-resolution imaging of localized defects such as microcracks and pores. Traditional ultrasonic testing in multilayer metal structures is susceptible to interference from interface acoustic discontinuities and grain noise, resulting in weak response to deep defects and difficulty in distinguishing shallow microcracks. Especially when surface and deep defects coexist, a single testing mode often cannot simultaneously achieve both detection sensitivity and imaging accuracy, failing to meet the stringent reliability assessment requirements of high-end equipment and critical components. Summary of the Invention

[0004] In view of this, it is necessary to provide a nondestructive testing method and system based on dual-modal collaboration, which can at least overcome one of the above-mentioned defects.

[0005] In a first aspect, embodiments of this application provide a nondestructive testing method based on dual-modal collaboration, applied to the defect detection of functionally graded laser cladding layers, the method comprising:

[0006] A scanning probe is used to emit a scanning signal, which is then incident obliquely onto the functionally graded laser cladding layer in the form of pulses. The scanning signal includes surface waves and transverse waves.

[0007] Receive scattered signals from defects or interfaces;

[0008] The scattered signal is subjected to continuous wavelet transform to extract the time-domain envelope and dominant frequency component of the scattered signal;

[0009] Align the arrival times of the defect main lobes of the surface wave mode and the transverse wave mode in the time domain;

[0010] The complementary frequency bands of the surface wave mode and the transverse wave mode are weighted and superimposed in the frequency domain to obtain a complex domain fused signal;

[0011] The point spread function generated analytically based on the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm are used to perform degradation correction on the complex domain fused signal to output a three-dimensional image of the curve.

[0012] In one embodiment of this application, the scattering signal includes surface wave-surface wave, surface wave-transverse wave, transverse wave-transverse wave, and transverse wave-surface wave.

[0013] In one embodiment of this application, the method further includes:

[0014] Based on electron backscatter diffraction experimental data, X-ray diffraction experimental data, and phase field simulation results, the grain size, orientation, and reinforcing phase distribution of the functionally graded laser cladding layer along the thickness direction are reconstructed.

[0015] Based on the grain size change rate, texture intensity, and phase change point, the functionally graded laser cladding layer is divided into multiple sub-layers, and the microstructure model is calibrated within each sub-layer using adaptive grid discretization technology, wherein the sub-layer boundary is determined by the phase change point.

[0016] A spatial point-pair model is constructed by Monte Carlo random sampling, and the multi-point spatial correlation function and texture-corrected elastic modulus covariance matrix are derived by combining higher-order statistical tensors and non-stationary random field theory.

[0017] Using the elastic modulus covariance matrix and multi-point spatial correlation function as input, the tensor form Green function of surface waves and transverse waves in multilayer heterogeneous media is analytically solved by the Dyson equation and the quasi-crystal approximation method.

[0018] The differential scattering cross section is derived by combining the displacement potential function and multi-point correlation quantities, and the integral of the scattering cross section of the two-mode wave is numerically solved by the Monte Carlo method, thereby quantifying the sound velocity dispersion and scattering attenuation characteristics caused by grain noise.

[0019] In one embodiment of this application, the method further includes:

[0020] A gradient factor is introduced for each of the sub-layers, and the gradient factor is the rate of change of the elastic modulus and density along the thickness direction in the functionally graded cladding layer.

[0021] The local stiffness matrix of each sublayer is recursively corrected using the gradient factor, and the corrected global stiffness matrix is ​​assembled.

[0022] Solve the characteristic equation of the global stiffness matrix to obtain the sound velocity dispersion curves and energy attenuation characteristics of the surface wave and the transverse wave in the multilayer structure.

[0023] The attenuation error caused by grain noise is compensated according to the characteristic equation;

[0024] Four mode conversion operators are constructed to describe the reflection, transmission, and energy conversion ratios of each mode at the interface.

[0025] In one embodiment of this application, the method further includes:

[0026] Based on the sound velocity dispersion and scattering attenuation characteristics caused by the grain noise, a grain noise database for each sublayer of the functionally graded cladding layer is constructed.

[0027] The received scattered signal is noise suppressed by an adaptive filtering algorithm, and the weight coefficients of the adaptive filtering algorithm are updated in real time by the elastic modulus covariance matrix.

[0028] In one embodiment of this application, the method further includes:

[0029] The scattering problem is transformed into a micro-perturbation problem, and the scattered field is solved iteratively based on the defect location coordinates, size, orientation angle and burial depth.

[0030] The reflection coefficient, transmission coefficient, and mode conversion ratio at the multilayer interface are analyzed by using Green's function in tensor form and Sommerfeld integral.

[0031] The formula for generating defect echo prediction is as follows:

[0032]

[0033] Among them, A inc,i (ω) represents the incident spectrum, ω is the frequency, θ is the incident angle, and S ij (ω,θ) represents the scattering amplitude of the incident mode i transformed into mode j at frequency ω and incident angle θ, recorded by the two-dimensional mode transformation operator in the frequency domain. j Let d be the wavenumber of mode j, and d be the acoustic path distance from the defect to the center of the probe.

[0034] In one embodiment of this application, the method further includes:

[0035] The amplitude of each echo signal acquired by the scanning probe is normalized.

[0036] The instantaneous phase of the signal is extracted using Hilbert transform, and the wedge angle deviation is eliminated by least squares fitting.

[0037] The propagation time difference of each mode wave is calculated based on the sound velocity dispersion curve, and the time delay of surface wave and shear wave signals is calibrated.

[0038] Short-time Fourier transform combined with time-frequency masking is applied in the frequency domain to separate interface clutter and improve signal contrast.

[0039] In one embodiment of this application, the method further includes:

[0040] Based on the sound velocity dispersion and scattering attenuation characteristics caused by the grain noise, an adaptive signal-to-noise ratio (SNR) evaluation function is constructed. The optimal weighting coefficients for weighted superposition of complementary frequency bands of surface wave mode and transverse wave mode in the frequency domain are calculated in real time according to the evaluation function to maximize the overall SNR of the fused signal. The SNR evaluation function is constructed by the grain noise statistical model.

[0041] In one embodiment of this application, the method further includes:

[0042] In the Richardson-Lucy iterative deconvolution algorithm, the complex domain fused signal is used as the initial estimate, and the defect space grayscale distribution is updated in each iteration by combining the observation data with the convolution residual.

[0043] Secondly, embodiments of this application provide a nondestructive testing system based on dual-modal collaboration, applied to the detection of defects in functionally graded laser cladding layers, the system comprising:

[0044] A scanning probe is used to emit a scanning signal, which is incident obliquely onto the functionally graded laser cladding layer in the form of a pulse. The scanning signal includes surface waves and transverse waves.

[0045] A signal receiving module is used to receive scattered signals from defects or interfaces;

[0046] The signal processing module is used to perform continuous wavelet transform on the scattered signal to extract the time-domain envelope and dominant frequency component of the scattered signal; align the arrival times of the defective main lobes of the surface wave mode and the transverse wave mode in the time domain; perform weighted superposition of the complementary frequency bands of the surface wave mode and the transverse wave mode in the frequency domain to obtain a complex-domain fused signal; and perform degradation correction on the complex-domain fused signal based on the point spread function analytically generated by the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm to output a three-dimensional image of the curve.

[0047] The dual-modal collaborative nondestructive testing method and system provided in this application can fully utilize the high sensitivity of surface waves to shallow microcracks and the strong penetrating power of transverse waves to deep structural defects. Through time-frequency domain fusion and deconvolution imaging, the blind zones of each mode are mutually compensated, achieving efficient detection of defects in functionally graded laser cladding layers from the surface to the depth. At the same time, based on the grain noise statistical model and the dual-modal scattering theory of interface effect compensation, anisotropic grain noise and interface discontinuity interference in heterogeneous structures can be significantly suppressed, improving the detection signal-to-noise ratio and achieving sub-millimeter-level imaging resolution. The scheme of dynamic adaptive filtering and multi-scale numerical simulation optimization can also maintain high accuracy and reliability in online detection of complex working conditions and large-area components, effectively meeting the stringent requirements for quality assessment of key components in aerospace, marine engineering equipment, and nuclear power industries. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating a nondestructive testing method based on dual-modal collaboration provided in an embodiment of this application.

[0049] Figure 2 This is a schematic diagram of a wedge scanning provided in an embodiment of this application.

[0050] Figure 3 This is a schematic diagram of a nondestructive testing method based on dual-modal collaboration, provided as another embodiment of this application.

[0051] Figure 4 This is a schematic diagram of a wedge probe provided in one embodiment of this application.

[0052] Figure 5 This is a schematic diagram of a nondestructive testing system module provided in an embodiment of this application.

[0053] Figure 6 A schematic diagram of an electronic device provided in an embodiment of this application.

[0054] Explanation of main component symbols:

[0055] Non-destructive testing system 10

[0056] Scanning probe 100

[0057] Signal receiving module 200

[0058] Signal processing module 300

[0059] Electronic devices 20

[0060] Processor 21

[0061] Memory 22

[0062] Steps S100-S600 Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of the invention described below can be combined with each other as long as they do not conflict with each other.

[0064] It should be noted that, in the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0065] It should be noted that in the embodiments of this application, the terms "first," "second," etc., are used only for descriptive purposes and should not be construed as indicating or implying relative importance, nor as indicating or implying order. Features specified as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0066] Based on the embodiments described in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] With the continuous development of materials surface engineering technology, laser cladding technology has become one of the key means to improve the surface performance of metal parts. By melting and rapidly solidifying powder or wire under the action of a high-energy laser beam, laser cladding can form a high-hardness, wear-resistant, and corrosion-resistant coating on the substrate surface, significantly enhancing the surface performance of parts. In particular, the introduction of functionally graded materials (FGMs) allows the cladding layer to exhibit continuous changes in chemical composition and mechanical properties, not only meeting the dual requirements of surface protection and high-strength support in extreme environments, but also greatly extending the service life of critical components. However, the complexity of the multilayer heterogeneous structure and microcrystalline structure of the laser cladding layer also brings greater challenges to subsequent quality assessment and defect detection.

[0068] Currently, non-destructive testing of functionally graded laser cladding layers mainly relies on methods such as thermal imaging, X-ray imaging, defect imaging, and single-mode ultrasonic testing. While thermal and X-ray imaging can achieve rapid scanning of the overall structure, they struggle to provide high-resolution imaging of localized defects such as microcracks and pores. Traditional ultrasonic testing in multilayer metal structures is susceptible to interference from interface acoustic discontinuities and grain noise, resulting in weak response to deep defects and difficulty in distinguishing shallow microcracks. Especially when surface and deep defects coexist, a single testing mode often cannot simultaneously achieve both detection sensitivity and imaging accuracy, failing to meet the stringent reliability assessment requirements of high-end equipment and critical components.

[0069] The dual-modal collaborative nondestructive testing method and system provided in this application can fully utilize the high sensitivity of surface waves to shallow microcracks and the strong penetrating power of transverse waves to deep structural defects. Through time-frequency domain fusion and deconvolution imaging, the blind zones of each mode are mutually compensated, achieving efficient detection of defects in functionally graded laser cladding layers from the surface to the depth. At the same time, based on the grain noise statistical model and the dual-modal scattering theory of interface effect compensation, anisotropic grain noise and interface discontinuity interference in heterogeneous structures can be significantly suppressed, improving the detection signal-to-noise ratio and achieving sub-millimeter-level imaging resolution. The scheme of dynamic adaptive filtering and multi-scale numerical simulation optimization can also maintain high accuracy and reliability in online detection of complex working conditions and large-area components, effectively meeting the stringent requirements for quality assessment of key components in aerospace, marine engineering equipment, and nuclear power industries.

[0070] Figure 1 This is a schematic flowchart illustrating a nondestructive testing method based on dual-modal collaboration provided in an embodiment of this application. Figure 1 The method shown includes at least the following steps: S100: emitting a scanning signal using a scanning probe, causing the scanning signal to be incident obliquely onto the functionally graded laser cladding layer in pulse form; S200: receiving scattered signals from defects or interfaces; S300: performing continuous wavelet transform on the scattered signals to extract the time-domain envelope and dominant frequency components of the scattered signals; S400: aligning the arrival times of the defect main lobes of the surface wave mode and the transverse wave mode in the time domain; S500: weighting and superimposing the complementary frequency bands of the surface wave mode and the transverse wave mode in the frequency domain to obtain a complex-domain fused signal; S600: performing degradation correction on the complex-domain fused signal based on the point spread function analytically generated by the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm to output a three-dimensional image of the curve.

[0071] S100: The scanning probe emits a scanning signal, which is then incident on the functionally graded laser cladding layer in a pulsed manner.

[0072] Specifically, in the embodiments of this application, the scanning probe excites surface (Rayleigh) waves and SH transverse waves in the frequency band of 1MHz to 10MHz, respectively. The pulse width of each transmission is adjustable, and the transmission power and coupling angle can be pre-calibrated according to the characteristics of the cladding layer material.

[0073] Understandably, this step, by simultaneously exciting surface waves and transverse waves, ensures both a high-sensitivity response to shallow microcracks in the cladding layer and sufficient penetration capability for deep defects.

[0074] S200: Receives scattered signals from defects or interfaces.

[0075] In one embodiment of this application, the scattered signals include surface wave-surface wave, surface wave-transverse wave, transverse wave-transverse wave, and transverse wave-surface wave.

[0076] Specifically, in the embodiments of this application, after the probe switches to the receiving mode, it sequentially captures four types of scattered echoes—surface wave-surface wave, surface wave-transverse wave, transverse wave-transverse wave, and transverse wave-surface wave—using A-scan, and performs amplitude normalization and time delay correction on each echo signal.

[0077] Understandably, this step fully acquires the defect scattering characteristics of the four complementary modes, providing multi-angle and multi-depth defect information for subsequent fusion and deconvolution imaging.

[0078] S300: Performs continuous wavelet transform on the scattered signal to extract the time-domain envelope and dominant frequency component of the scattered signal.

[0079] Specifically, in the embodiments of this application, the normalized A-scan signal is input into the continuous wavelet transform (CWT), and the Morlet wavelet basis is selected to perform multi-scale decomposition on the signal, and the time-domain envelope curve of the echo signal and the corresponding main frequency component envelope are extracted respectively.

[0080] Understandably, wavelet transform can simultaneously provide local features of a signal in both the time and frequency domains, making it easier to accurately locate the arrival instant and spectral characteristics of defect echoes on the time axis.

[0081] S400: Arrival time of the defective main lobe of the surface wave mode and the transverse wave mode aligned in the time domain.

[0082] Specifically, in the embodiments of this application, the time delay difference between the echo envelope curves of each mode is calculated by cross-correlation algorithm, and the arrival time of the main lobe of the surface wave and the shear wave is interpolated and compensated to make the defective main lobes of different modes aligned at the same time reference point.

[0083] Understandably, this step eliminates the time difference between the two modes of wave propagation in terms of propagation speed and path, laying a strict time synchronization foundation for subsequent frequency domain fusion.

[0084] S500: In the frequency domain, the complementary frequency bands of surface wave modes and shear wave modes are weighted and superimposed to obtain a complex domain fused signal.

[0085] Specifically, in the embodiments of this application, the aligned time-domain signal is transformed to the frequency domain again by FFT, the weight coefficients are dynamically calculated according to the pre-constructed signal-to-noise ratio (SNR) evaluation function, and then the complex spectra of the two modes in the complementary frequency band are weighted and superimposed to obtain the fused complex domain signal.

[0086] Understandably, frequency domain weighted fusion utilizes the high SNR characteristics of surface waves in the low-frequency part and the penetration advantage of transverse waves in the high-frequency part to achieve optimal information integration of defect signals.

[0087] S600: Based on the point spread function generated analytically by the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm, the complex domain fused signal is degraded to output a three-dimensional image of the curve.

[0088] Specifically, in the embodiments of this application, the fused complex domain signal obtained in steps S200-S500 is used as the observed image. The pre-analyzed dual-modal point spread function (PSF) is substituted into the Richardson-Lucy algorithm to iteratively update the defect image estimation. After each iteration, dynamic adaptive filtering is applied to suppress background noise and speckle effect until the three-dimensional contour map of the defect is output.

[0089] Understandably, this step recovers the true spatial distribution of defects through complex domain deconvolution, which not only improves the imaging resolution (sub-millimeter level) but also directly obtains the depth and morphology information of the defects.

[0090] In this embodiment of the application, the method further includes: reconstructing the grain size, orientation, and reinforcing phase distribution of the functionally graded laser cladding layer along the thickness direction based on electron backscatter diffraction experimental data, X-ray diffraction experimental data, and phase field simulation results.

[0091] Specifically, using the grain orientation distribution map obtained by EBSD and the phase composition diffraction peak intensity measured by XRD, and combining the phase interface evolution information output by phase field simulation, the grain size and orientation at each location are extracted by image segmentation and statistical analysis algorithms, and a three-dimensional grain structure reconstruction model is generated in the thickness direction according to a predetermined step size.

[0092] Understandably, this step provides high-precision microstructure input for subsequent acoustic scattering modeling, ensuring that the physical basis of Green's function solution and noise compensation is consistent with the actual cladding layer structure.

[0093] In this embodiment, the functionally graded laser cladding layer is divided into multiple sub-layers based on the grain size change rate, texture intensity, and phase transition point. An adaptive grid discretization technique is applied to calibrate the microstructure model within each sub-layer, wherein the sub-layer boundary is determined by the phase transition point.

[0094] Specifically, the phase transition depth is first determined based on the gradient rate of grain size change with depth and the location of texture extrema, and the entire cladding layer is divided into several sub-layers with variable thickness. Then, a refined grid is automatically generated within each sub-layer, and the grid density is dynamically adjusted according to the grain size distribution and texture change rate to ensure that the statistical characteristics of the grains in each grid unit are uniform and describable.

[0095] Understandably, by using hierarchical calibration and adaptive discretization, the tissue differences within the heterogeneous layer are refined to each grid cell, laying a structurally uniform numerical foundation for the accurate calculation of spatial correlation functions and scattering characteristics.

[0096] In this embodiment, a spatial point-pair model is constructed using Monte Carlo random sampling. The multi-point spatial correlation function and the texture-corrected elastic modulus covariance matrix are derived using higher-order statistical tensors and non-stationary random field theory. Using the elastic modulus covariance matrix and the multi-point spatial correlation function as input, the tensor form of the Green's function for surface waves and transverse waves in multilayer heterogeneous media is analytically solved using the Dyson equation and the quasi-crystalline approximation method.

[0097] Specifically, tens of thousands of random point pairs are uniformly selected within each sublayer, and the second and third statistical moments of their elastic modulus perturbation are calculated to express the micro-inhomogeneity in the form of higher-order statistical tensors. Then, the obtained statistics are input into the Dyson equation, and combined with the quasicrystal approximation and displacement potential function, the Rayleigh and SH mode Green function tensor expressions containing interlayer interface effects are analytically solved.

[0098] Understandably, this step takes into account the non-stationary elastic disturbances caused by grains and texture, and obtains a mathematical kernel (Green function) that can accurately describe the propagation and scattering of dual-mode waves in multilayer structures, providing a core operator for subsequent calculation of scattering cross-sections.

[0099] In this embodiment, the differential scattering cross section is derived by combining the displacement potential function and multi-point correlation quantities, and the integral of the dual-mode wave scattering cross section is numerically solved by the Monte Carlo method to quantify the sound velocity dispersion and scattering attenuation characteristics caused by grain noise.

[0100] Specifically, based on the tensor forms of the aforementioned Green function and displacement potential function, the elastic modulus perturbation caused by the grains is mapped to scattering potential energy using a small perturbation approximation. Then, the differential scattering cross section formula for the two-mode wave is constructed, and the scattering contribution of all spatial point pairs is accumulated using Monte Carlo numerical integration. Finally, the sound velocity dispersion curve and the frequency-dependent attenuation coefficient are obtained.

[0101] Understandably, this numerical solution tightly couples the microscopic statistical model with the macroscopic acoustic response, accurately characterizing the influence of grain noise on the propagation attenuation and phase distortion of dual-mode ultrasound in functionally graded cladding layers.

[0102] In this embodiment, the method further includes: introducing a gradient factor for each sublayer, where the gradient factor is the continuous rate of change of the elastic modulus and density along the thickness direction in the functionally graded cladding layer; recursively correcting the local stiffness matrix of each sublayer using the gradient factor and assembling the corrected global stiffness matrix; solving the characteristic equation of the global stiffness matrix to obtain the sound velocity dispersion curves and energy attenuation characteristics of surface waves and transverse waves in the multilayer structure; compensating for attenuation errors caused by grain noise based on the characteristic equation; and constructing four mode conversion operators to describe the reflection, transmission, and energy conversion ratios of each mode at the interface.

[0103] Specifically, the elastic modulus gradient and density gradient of each sublayer are calculated using material composition and phase field simulation data, and these are assigned to the material properties of the finite element elements. Following a shallow-to-deep order, the stiffness matrices of each sublayer are multiplied by their corresponding gradient factors and weighted, and a global stiffness matrix for the overall structure is generated using stiffness matrix assembly rules. Then, eigenvalue decomposition is performed on this global matrix to obtain the intrinsic mode dispersion curves and corresponding energy attenuation coefficients of Rayleigh and SH waves at various frequencies, which are used to perform band-level correction on the attenuation curves based on grain statistics. Finally, for the intrinsic modes at each interface, a 4×4 mode transformation matrix is ​​constructed according to four transformation relationships: Rayleigh→Rayleigh, Rayleigh→SH, SH→Rayleigh, and SH→SH, for use in subsequent signal model reflection and transmission coefficient multiplication calculations.

[0104] Understandably, this step directly integrates the continuous variation characteristics of the functionally graded layer into the wave equation and scattering model, and achieves synergistic compensation of interface effects and grain noise through feature analysis, thereby ensuring that the propagation and conversion laws of the dual-mode signal in the multilayer structure are highly consistent with the actual material.

[0105] In one embodiment of this application, the method further includes: constructing a grain noise database for each sublayer of the functionally graded cladding layer based on the sound velocity dispersion and scattering attenuation characteristics caused by grain noise; and suppressing noise in the received scattering signal using an adaptive filtering algorithm, wherein the weight coefficients of the adaptive filtering algorithm are updated in real time by the elastic modulus covariance matrix.

[0106] Specifically, according to different excitation frequencies and incident angles, the attenuation curves and phase distortion curves of each sublayer in the 1MHz–10MHz frequency band are pre-calculated and stored to form a multidimensional lookup table. During the detection process, the difference between the database curve and the current signal under the corresponding frequency bandwidth is read in real time, and the difference is used as the noise power estimation input in the adaptive filter (such as Wiener filter or Kalman filter). The filter gain is dynamically adjusted by the covariance matrix to remove random speckle noise caused by grain inhomogeneity.

[0107] Understandably, this step, through the combination of "prior database + real-time covariance update," allows the filter to be supported by a statistical model and adapt to batch differences in workpieces, effectively improving the clarity and reliability of the signal.

[0108] In this embodiment, the method further includes: transforming the scattering problem into a micro-perturbation problem, and iteratively solving the scattering field based on the defect location coordinates, size, orientation angle, and burial depth. The reflection coefficient, transmission coefficient, and mode conversion ratio at the multilayer interface are analyzed using a tensor form Green's function and a Sommerfeld integral.

[0109] Specifically, a perturbation potential model is first constructed using the initial parameters of the defect (x, y, z positions, length, width, thickness, and orientation angle). The wave equation is then linearized within the framework of the tensor Green's function, and the scattered field is expressed as a perturbation integral. Subsequently, the Sommerfeld integral path integration method is used to calculate the component coefficients from the incident mode to the scattered mode for each interface layer. The obtained reflection / transmission coefficients and mode conversion ratios are then substituted back into the perturbation integral expression to update the defect parameters. This iteration is repeated until the error between the scattered field and the experimentally acquired signal converges to a preset threshold.

[0110] Understandably, this step achieves an integrated solution of "model-driven + data correction" through perturbation iteration, so that the inversion of defect parameters is based on physical model derivation and closely matches actual wave field observations.

[0111] In this embodiment of the application, the defect echo prediction is generated using the following formula:

[0112]

[0113] Among them, A inc,i (ω) represents the incident spectrum, ω is the frequency, θ is the incident angle, and Sij (ω,θ) represents the scattering amplitude of the incident mode i transformed into mode j at frequency ω and incident angle θ, recorded by the two-dimensional mode transformation operator in the frequency domain. j Let d be the wavenumber of mode j, and d be the acoustic path distance from the defect to the center of the probe.

[0114] Specifically, by calling the previously solved mode conversion operator data and scattering path length, the contribution of each pair of (i→j) modes is calculated and accumulated, and then the phase delay term is superimposed to obtain the theoretically predicted echo.

[0115] Understandably, this prediction model concisely couples the incident spectrum, interface transformation, and propagation attenuation, making it suitable for both simulation comparison and real-time matching and parameter inversion in actual detection.

[0116] In this embodiment, the method further includes: performing amplitude normalization processing on each echo signal acquired by the scanning probe; extracting the instantaneous phase of the signal using Hilbert transform and eliminating wedge angle deviation through least-squares fitting; calculating the propagation time difference of each mode wave based on the sound velocity dispersion curve and calibrating the time delay of the surface wave and shear wave signals; and applying short-time Fourier transform combined with time-frequency masking in the frequency domain to separate interface clutter and improve signal contrast.

[0117] Specifically, firstly, each A-scan echo signal is normalized by dividing it by its maximum amplitude value. Then, the Hilbert transform is applied to the normalized signal to calculate the instantaneous phase curve, and the functional relationship between phase and incident angle is fitted using a wedge theory model. The actual incident angle deviation is then calculated using the least squares algorithm to correct subsequent data. Next, based on the pre-calculated Rayleigh and SH wave velocity dispersion curves, the time-frequency correspondence in each frequency band is derived, and the time delay of the surface and shear wave signals is aligned accordingly. Finally, a short-time Fourier transform is performed on the corrected signal to generate a time-frequency spectrum. The designed binary time-frequency mask is then applied to suppress multiple reflections and stray energy at the interface, retaining only the main lobe energy concentrated within the time-frequency window of the defect echo.

[0118] Understandably, this step, through precise phase correction and time delay matching, combined with time-frequency domain clutter suppression, not only eliminates system errors caused by hardware installation and material delamination, but also effectively isolates interface reflection interference, significantly improving the identifiability of defect signals.

[0119] In this embodiment of the application, the method further includes: constructing an adaptive signal-to-noise ratio (SNR) evaluation function based on the sound velocity dispersion and scattering attenuation characteristics caused by grain noise, and calculating the optimal weighting coefficients for weighted superposition of complementary frequency bands of surface wave mode and transverse wave mode in the frequency domain in real time based on the evaluation function, so as to maximize the overall SNR of the fused signal. The SNR evaluation function is constructed by a grain noise statistical model.

[0120] Specifically, the signal-to-noise ratio (SNR) is defined based on the noise attenuation spectrum and dispersion data of each sublayer grain obtained from Monte Carlo simulation. A sliding window method is used to estimate the defect signal and noise power spectral density in real time within the selected frequency band. Normalized weighting coefficients w(ω) are generated according to the SNR value, and w(ω) is applied to the complex spectrum superposition of surface waves and transverse waves to obtain the optimal fusion result.

[0121] Understandably, this step minimizes the impact of low SNR segments of the chip noise through frequency band adaptive weighting, while retaining effective information in high-penetration regions, thus achieving dynamic optimal fusion of multi-mode signals.

[0122] In one embodiment of this application, the method further includes: in the Richardson-Lucy iterative deconvolution algorithm, using the complex domain fused signal as the initial estimate, and in each iteration, combining the observation data with the convolution residual to update the gray-scale distribution of the defect space.

[0123] Specifically, the complex domain fusion spectrum obtained in step S500 is restored to the spatial domain observation image by inverse FFT. The image and the pre-analyzed point spread function are input, and the image estimation is updated sequentially by iterative formula. During the iteration process, the local brightness is adaptively adjusted by combining the observation-estimation error spectrum until the overall residual converges to the preset threshold and the final three-dimensional distribution of defects is output.

[0124] Understandably, this iterative deconvolution not only restores the blur caused by scattering and system response, but also guides updates through convolution residuals in each iteration, enabling the defect morphology and location to be gradually and precisely restored, achieving sub-millimeter imaging accuracy.

[0125] Please refer to the following: Figure 2 and Figure 3 , Figure 2 This is a schematic diagram of wedge detection provided in an embodiment of this application. Figure 3 This is a schematic diagram of a nondestructive testing method based on dual-modal collaboration, provided as another embodiment of this application.

[0126] In the embodiments of this application, Figure 2 The process of dual-mode wave excitation-scattering-reception is illustrated. Figure 3 The dual-mode wave scattering of the scanned defect is shown.

[0127] Please refer to the following: Figure 4 , Figure 4 This is a schematic diagram of a wedge probe provided in one embodiment of this application.

[0128] In this embodiment of the application, the dual critical angle wedge probe is composed of two preset wedges, each wedge as follows: Figure 4As shown, these correspond to the critical coupling angles for surface waves (Rayleigh waves) and transverse waves (SH waves), respectively, enabling dual-mode excitation and reception at the same probe location. The probe body is made of highly damped material and equipped with an adjustable coupling device and a miniature transducer unit, achieving a broadband response of 1MHz–10MHz.

[0129] Specifically, the tilt angle θ_R of the upper wedge of the probe is set as the critical angle of Rayleigh wave incidence according to the acoustic impedance ratio of the cladding substrate and the coating, while the tilt angle θ_SH of the lower wedge is determined according to the shear velocity of the SH wave. The two are fixed on the same base by a precision mechanical structure. In actual testing, the upper wedge first excites and receives surface waves, and then switches to the lower wedge to excite transverse waves, or simultaneously acquires four modes of echo signals in dual-channel mode. High-viscosity coupling adhesive is used between the wedge and the workpiece surface to ensure stable signal transmission and high repeatability.

[0130] Understandably, this integrated dual-critical-angle design not only simplifies the probe replacement process, but also maximizes the coupling efficiency of the two ultrasonic modes through precise angle matching, thereby improving the signal-to-noise ratio and ensuring comprehensive coverage of both shallow and deep defects.

[0131] The following is an exemplary embodiment to describe a nondestructive testing method based on dual-modal collaboration provided in this application.

[0132] In this embodiment of the application, the object to be tested is the laser cladding layer on the surface of the outer ring (material: 42CrMo4 steel) of a certain type of wind turbine bearing. The cladding material is NiCrBSi-WC composite powder, and the thickness direction is gradient-distributed (80% WC content on the surface and 10% on the substrate side).

[0133] In this embodiment of the application, the experimental equipment is the integrated dual-critical-angle wedge probe provided by the present invention: operating frequency 1–10MHz, surface wave critical angle α1=62°, transverse wave critical angle α2=45°, adapted to the curved surface of wind turbine bearing (radius R=250mm).

[0134] In this embodiment, the ultrasonic testing system is equipped with a pulse transmitter and receiver (PRF 1kHz), a multi-channel digital signal acquisition module (sampling rate 100MS / s), and a dynamic adaptive filter bank. The simulation platform is finite element software with multi-level mesh generation (micro-grain scale: 0.5μm; macro-defect scale: 50μm).

[0135] In this embodiment, EBSD / XRD experimental data of the functionally graded cladding layer were collected, and combined with phase-field simulation results, the grain size (1–10 μm), orientation (texture intensity f(g) = 0.3–0.7), and WC reinforcement phase distribution (volume fraction φ = 10%–80%) in the thickness direction (0–2 mm) were reconstructed. This was based on the grain size variation rate (Δd / dz = 0.2–0.5 / μm). -1 The cladding layer is divided into 5 sub-layers (each 0.4 mm) based on the phase transition point (φ = 50%).

[0136] Each sub-layer is adaptively discretized using a grid, and Monte Carlo random sampling is used to generate 10... 6 For each pair of spatial points, the multi-point spatial correlation function C(r) = exp(-r / ξ)cos(2πr / λ0) is derived by combining the higher-order statistical tensor (fourth-order covariance matrix) and the theory of non-stationary random fields, where ξ is the correlation length (0.8–2.5μm) and λ0 is the characteristic wavelength (3–8μm).

[0137]

[0138] Calculate the sound velocity dispersion curves of Rayleigh waves (c_R = 2980 m / s) and SH waves (c_SH = 3230 m / s) in each sublayer, and solve for the scattering cross-section integral using Monte Carlo integration:

[0139]

[0140] The energy attenuation coefficient α(ω) caused by grain noise is obtained as 0.2–1.5 dB / mm (frequency 2–10 MHz).

[0141] A gradient factor γ = ΔE / Δρ (where E is the elastic modulus and ρ is the density) is introduced for each interface layer, and the local stiffness matrix [K] is recursively corrected. n Assemble the corrected global stiffness matrix [K]:

[0142]

[0143] Solve the characteristic equation |det([K]-ω 2 [M])|=0, and the sound velocity dispersion curves of Rayleigh wave and SH wave are obtained, which compensates for the error caused by the acoustic discontinuity of the interface (the error after correction is <2%).

[0144] For typical defects in wind turbine bearings (microcracks L=50μm, pores D=100μm, inclusions φ=50μm), a small perturbation method is used to develop the scattering differential section:

[0145]

[0146] Where a is the defect size, Δμ is the modulus perturbation, and f(θ) is the angle-dependent function.

[0147] By combining the Green's function and the Sommerfeld integral, the defect echo signal is solved iteratively:

[0148]

[0149] For the curved surface of wind turbine bearings, a curvature compensation algorithm is used to correct the probe contact surface (curvature radius R = 250mm ± 1%), reducing the acoustic field distortion rate to below 3%. The probe center frequency is set to 5MHz, the pulse width to 500ns, and the excitation voltage to 200Vpp.

[0150] Four modal scattering signals were acquired using pulse reflection scanning: surface wave-surface wave (RR): defect main lobe arrival time t1 = 1.2 μs, dominant frequency f1 = 6.8 MHz; surface wave-transverse wave (R-SH): t2 = 1.5 μs, f2 = 4.2 MHz; transverse wave-transverse wave (SH-SH): t3 = 1.8 μs, f3 = 5.5 MHz; transverse wave-surface wave (SH-R): t4 = 2.1 μs, f4 = 3.8 MHz. The acquired A-scan signals were normalized (amplitude error <5%), time delay calibrated (time difference <0.1 μs), and phase compensated (Hilbert transform correction).

[0151] Continuous wavelet transform (CWT) was performed on the four modal signals to extract the time-domain envelope (main lobe width Δt = 0.3 μs) and the dominant frequency component (RR: 6.8 MHz ± 0.5 MHz, SH-SH: 5.5 MHz ± 0.3 MHz). The signal-to-noise ratio of each modality was calculated (RR: 12 dB, SH-SH: 9 dB, R-SH: 7 dB).

[0152] At the arrival time of the main lobe of the time-domain alignment defect, the frequency domain selects the RR high-frequency band (5–10MHz) and the SH-SH low-frequency band (1–5MHz) for weighted superposition, and the overall SNR is improved to 15dB after fusion.

[0153] The point spread function (PSF) was analytically generated based on the two-mode scattering theory, with a main lobe width of 0.3 mm (λ / 2at5MHz) and side lobe suppression >20 dB. The Richardson-Lucy algorithm was applied iteratively 50 times, with the initial estimated image being a zero matrix, and the defect image was updated in each iteration.

[0154] During the iteration process, Wiener filtering (time domain) and matched pursuit (frequency domain) were used to jointly suppress low-frequency background noise (<1MHz) and high-frequency speckle noise (>8MHz). The output is a 3D image of the defect with a resolution of 0.2mm (λ / 10at5MHz). Defect types identified include: transverse microcracks (length L = 80μm, depth d = 300μm); internal pores (diameter D = 150μm, depth d = 1.2mm); and inclusions (size 100×80μm, distribution density ρ = 5 inclusions / mm). 2 .

[0155] A cladding layer model of the wind turbine bearing, incorporating grain noise (based on EBSD data), was established in finite element software. Surface waves and SH waves were excited, and the aforementioned three types of defects were introduced. Micro-CT inspection results showed that the defect size error was <8% and the burial depth error was <50μm, meeting the requirements for wind turbine bearing service life assessment.

[0156] The method provided in this application improves the signal-to-noise ratio of microcracks (50μm) to 15dB by suppressing grain noise through dual-modal collaborative methods. It achieves clear defect contours at the sub-millimeter (0.2mm) level with a burial depth positioning error of <50μm. It simultaneously covers surface (RR) and deep (SH-SH) defects, improving detection efficiency by 40%. It is adaptable to the irregular curved surfaces of wind turbine bearings, with probe coupling stability >95%. The defect classification report supports lifetime prediction (crack propagation rate da / dN calculation error <10%).

[0157] Figure 5 This is a schematic diagram of the modules of a nondestructive testing system provided in an embodiment of this application, as shown below. Figure 5 The non-destructive testing system 10 shown includes at least the following components: a scanning probe 100, a signal receiving module 200, and a signal processing module 300.

[0158] In this embodiment, the scanning probe 100 is used to emit a scanning signal, which is then incident obliquely onto the functionally graded laser cladding layer in the form of pulses. The scanning signal includes surface waves and transverse waves. Please refer to the following for details. Figure 1 The details and their corresponding descriptions are not repeated here.

[0159] In this embodiment, the signal receiving module 200 is used to receive scattered signals from defects or interfaces. Please refer to the following for details. Figure 1 The details and their corresponding descriptions are not repeated here.

[0160] In this embodiment, the signal processing module 400 performs continuous wavelet transform on the scattered signal to extract the time-domain envelope and dominant frequency component of the scattered signal; aligns the arrival times of the defective main lobes of the surface wave mode and the transverse wave mode in the time domain; weights and superimposes the complementary frequency bands of the surface wave mode and the transverse wave mode in the frequency domain to obtain a complex-domain fused signal; and performs degradation correction on the complex-domain fused signal based on the point spread function analytically generated by the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm to output a three-dimensional image of the curve. Please refer to the following for details. Figure 1 The details and their corresponding descriptions are not repeated here.

[0161] Figure 6 This is an electronic device 20 provided in one embodiment of this application. For example... Figure 6 As shown, the electronic device 20 includes at least the following components: a processor 21 and a memory 22.

[0162] In this embodiment, the memory 22 is used to store executable instructions of the processor 21, which, when configured to execute instructions, implement... Figure 1 The nondestructive testing method based on dual-modal collaboration is shown.

[0163] In this application embodiment, a computer-readable storage medium includes instructions that instruct a device to perform actions such as... Figure 1 The nondestructive testing method based on dual-modal collaboration is shown.

[0164] In one embodiment of this application, the program operating in the electronic device 20 may be a program that controls a central processing unit (CPU) or similar device to achieve the functions described in the above-described embodiments of the present invention (a program that enables the computer to function). Information processed by these devices is then temporarily stored in random access memory (RAM) during processing, and subsequently stored in various ROMs such as read-only memory (Flash ROM) and hard disk drives (HDDs), and read, corrected, and written by the CPU as needed.

[0165] It should be noted that a portion of the electronic device 20 described above can also be implemented using a computer. In this case, the program for implementing the control function can be recorded on a computer-readable recording medium, and the program recorded on the recording medium can be read into the computer system and executed.

[0166] It should be noted that the "computer system" mentioned here refers to the computer system built into electronic device 20, which employs hardware including an operating system and peripheral devices. Furthermore, "computer-readable recording media" refers to removable media such as floppy disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage devices such as hard drives built into the computer system.

[0167] Furthermore, a "computer-readable recording medium" can include: a medium that dynamically stores a program for a short period of time, such as a communication line used when transmitting a program via a network such as the Internet or a communication line such as a telephone line; or a medium that stores a program for a fixed period of time, such as volatile memory within a computer system that serves as a server or client in this case. In addition, the aforementioned program can be a program used to implement the above-mentioned functions, or it can be a program that can implement the above-mentioned functions by combining with programs already recorded in the computer system.

[0168] Furthermore, the electronic device 20 in the above embodiments can also be implemented as an assembly (device group) composed of multiple devices. Each device constituting the device group can possess some or all of the functions or functional blocks of the electronic device 20 in the above embodiments. As a device group, it is sufficient to have all the functions or functional blocks of the electronic device 20.

[0169] The dual-modal collaborative nondestructive testing method and system 10 provided in this application can fully utilize the high sensitivity of surface waves to shallow microcracks and the strong penetrating power of transverse waves to deep structural defects. Through time-frequency domain fusion and deconvolution imaging, the blind zones of each mode are mutually compensated, achieving efficient detection of defects in functionally graded laser cladding layers from the surface to the depth. At the same time, based on the grain noise statistical model and the interface effect compensation dual-modal scattering theory, anisotropic grain noise and interface discontinuity interference in heterogeneous structures can be significantly suppressed, improving the detection signal-to-noise ratio and achieving sub-millimeter-level imaging resolution. The scheme of dynamic adaptive filtering and multi-scale numerical simulation optimization can also maintain high accuracy and reliability in online detection of complex working conditions and large-area components, effectively meeting the stringent requirements for quality assessment of key components in aerospace, marine engineering equipment, and nuclear power industries.

[0170] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A nondestructive testing method based on dual-modal synergy, applied to defect detection of functionally graded laser cladding layers, characterized in that, The method includes: A scanning probe is used to emit a scanning signal, which is then incident obliquely onto the functionally graded laser cladding layer in the form of pulses. The scanning signal includes surface waves and transverse waves. Receive scattered signals from defects or interfaces; The scattered signal is subjected to continuous wavelet transform to extract the time-domain envelope and dominant frequency component of the scattered signal; Align the arrival times of the defect main lobes of the surface wave mode and the transverse wave mode in the time domain; The complementary frequency bands of the surface wave mode and the transverse wave mode are weighted and superimposed in the frequency domain to obtain a complex domain fused signal; The point spread function generated analytically based on the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm are used to perform degradation correction on the complex domain fused signal to output a three-dimensional image of the curve.

2. The nondestructive testing method based on dual-modal collaboration according to claim 1, characterized in that, The scattered signals include surface wave-surface wave, surface wave-transverse wave, transverse wave-transverse wave, and transverse wave-surface wave.

3. The nondestructive testing method based on dual-modal collaboration according to claim 2, characterized in that, The method further includes: Based on electron backscatter diffraction experimental data, X-ray diffraction experimental data, and phase field simulation results, the grain size, orientation, and reinforcing phase distribution of the functionally graded laser cladding layer along the thickness direction are reconstructed. Based on the grain size change rate, texture intensity, and phase change point, the functionally graded laser cladding layer is divided into multiple sub-layers, and the microstructure model is calibrated within each sub-layer using adaptive grid discretization technology, wherein the sub-layer boundary is determined by the phase change point. A spatial point-pair model is constructed by Monte Carlo random sampling, and the multi-point spatial correlation function and texture-corrected elastic modulus covariance matrix are derived by combining higher-order statistical tensors and non-stationary random field theory. Using the elastic modulus covariance matrix and multi-point spatial correlation function as input, the tensor form Green function of surface waves and transverse waves in multilayer heterogeneous media is analytically solved by the Dyson equation and the quasi-crystal approximation method. The differential scattering cross section is derived by combining the displacement potential function and multi-point correlation quantities, and the integral of the scattering cross section of the two-mode wave is numerically solved by the Monte Carlo method, thereby quantifying the sound velocity dispersion and scattering attenuation characteristics caused by grain noise.

4. The nondestructive testing method based on dual-modal collaboration according to claim 3, characterized in that, The method further includes: A gradient factor is introduced for each of the sub-layers, and the gradient factor is the rate of change of the elastic modulus and density along the thickness direction in the functionally graded cladding layer. The local stiffness matrix of each sublayer is recursively corrected using the gradient factor, and the corrected global stiffness matrix is ​​assembled. Solve the characteristic equation of the global stiffness matrix to obtain the sound velocity dispersion curves and energy attenuation characteristics of the surface wave and the transverse wave in the multilayer structure. The attenuation error caused by grain noise is compensated according to the characteristic equation; Four mode conversion operators are constructed to describe the reflection, transmission, and energy conversion ratios of each mode at the interface.

5. The nondestructive testing method based on dual-modal collaboration according to claim 3, characterized in that, The method further includes: Based on the sound velocity dispersion and scattering attenuation characteristics caused by the grain noise, a grain noise database for each sublayer of the functionally graded cladding layer is constructed. The received scattered signal is noise suppressed by an adaptive filtering algorithm, and the weight coefficients of the adaptive filtering algorithm are updated in real time by the elastic modulus covariance matrix.

6. The nondestructive testing method based on dual-modal collaboration according to claim 4, characterized in that, The method further includes: The scattering problem is transformed into a micro-perturbation problem, and the scattered field is solved iteratively based on the defect location coordinates, size, orientation angle and burial depth. The reflection coefficient, transmission coefficient, and mode conversion ratio at the multilayer interface are analyzed by using Green's function in tensor form and Sommerfeld integral. The formula for generating defect echo prediction is as follows: Among them, A inc,i (ω) represents the incident spectrum, ω is the frequency, θ is the incident angle, and S ij (ω,θ) represents the scattering amplitude of the incident mode i transformed into mode j at frequency ω and incident angle θ, recorded by the two-dimensional mode transformation operator in the frequency domain. j Let d be the wavenumber of mode j, and d be the acoustic path distance from the defect to the center of the probe.

7. The nondestructive testing method based on dual-modal collaboration according to claim 1, characterized in that, The method further includes: The amplitude of each echo signal acquired by the scanning probe is normalized. The instantaneous phase of the signal is extracted using Hilbert transform, and the wedge angle deviation is eliminated by least squares fitting. The propagation time difference of each mode wave is calculated based on the sound velocity dispersion curve, and the time delay of surface wave and shear wave signals is calibrated. Short-time Fourier transform combined with time-frequency masking is applied in the frequency domain to separate interface clutter and improve signal contrast.

8. The nondestructive testing method based on dual-modal collaboration according to claim 1, characterized in that, The method further includes: Based on the sound velocity dispersion and scattering attenuation characteristics caused by the grain noise, an adaptive signal-to-noise ratio (SNR) evaluation function is constructed. The optimal weighting coefficients for weighted superposition of complementary frequency bands of surface wave mode and transverse wave mode in the frequency domain are calculated in real time according to the evaluation function to maximize the overall SNR of the fused signal. The SNR evaluation function is constructed by the grain noise statistical model.

9. The nondestructive testing method based on dual-modal collaboration according to claim 4, characterized in that, The method further includes: In the Richardson-Lucy iterative deconvolution algorithm, the complex domain fused signal is used as the initial estimate, and the defect space grayscale distribution is updated in each iteration by combining the observation data with the convolution residual.

10. A nondestructive testing system based on dual-modal collaboration, applied to defect detection in functionally graded laser cladding layers, the system comprising: A scanning probe is used to emit a scanning signal, which is incident obliquely onto the functionally graded laser cladding layer in the form of a pulse. The scanning signal includes surface waves and transverse waves. A signal receiving module is used to receive scattered signals from defects or interfaces; The signal processing module is used to perform continuous wavelet transform on the scattered signal to extract the time-domain envelope and the dominant frequency component of the scattered signal; The arrival times of the defect main lobes of the surface wave mode and the shear wave mode are aligned in the time domain; the complementary frequency bands of the surface wave mode and the shear wave mode are weighted and superimposed in the frequency domain to obtain a complex domain fused signal; the complex domain fused signal is degraded and corrected based on the point spread function generated analytically by the dual-mode scattering model and the Richardson-Lucy iterative deconvolution algorithm to output a three-dimensional image of the curve.

Images