Metal microcrack ultrasonic detection method

Abstract

The invention belongs to the technical field of metal crack detection, and discloses a metal microcrack ultrasonic detection method, which comprises the following modules: an ultrahigh frequency composite array probe with a center frequency dynamic adjustable range of 5-50 MHz; the dual-mode excitation and receiving module is used for independently generating a pulse reflection mode excitation signal and a diffraction mode chirp frequency modulation signal; the phase-controlled synthetic aperture dynamic focusing module is used for realizing + / -60-degree deflection and sub-wavelength level focusing of acoustic beams; the nonlinear harmonic enhancement module is used for exciting a crack nonlinear effect by superposing 1-2MHz fundamental waves; and a time-space-frequency three-domain data fusion module. According to the method, signal-to-noise ratio collapse caused by the diffraction effect in a traditional method is overcome, more importantly, the system has low-frequency penetrability and high-frequency resolution at the same time due to 80%-120% of broadband coverage of chirp frequency modulation signals, the diffraction signal gain is increased by 20 dB or above, a near-surface blind area is compressed to be within 0.2 mm, and the system can be applied to the field of radar imaging. And the core pain point of missing detection of the near-surface microcracks of the thin-wall component is solved.

Description

Technical Field

[0001] This invention belongs to the field of metal crack detection technology, specifically an ultrasonic detection method for metal microcracks. Background Technology

[0002] Ultrasonic testing for microcracks in metals is a non-destructive testing technique widely used in quality control and defect detection of metallic materials, particularly in industries such as aerospace and machinery manufacturing. This technology involves emitting high-frequency ultrasonic waves into a metal component. The presence of microcracks is determined by the reflection or scattering signals generated when these waves encounter cracks, pores, or other defects as they propagate within the metal. Specifically, an ultrasonic probe transmits high-frequency sound waves to the metal surface. When these waves encounter discontinuous structures such as cracks, they generate echoes. The probe receives these echo signals and analyzes data such as the intensity, frequency, and propagation time of the echoes to determine the location, size, and nature of the cracks.

[0003] When using the conventional pulse reflection method, there are inherent contradictions in physical scale, resulting in insufficient signal-to-noise ratio and directional limitations. The size of microcracks is close to or even smaller than the length of commonly used ultrasonic waves, causing the sound waves to mainly diffract rather than be effectively reflected. The echo energy is extremely weak and is easily submerged by material microstructure noise and electronic noise. At the same time, due to the strict requirements of the beam incident angle, the crack echo signal in the direction not perpendicular to the sound beam is significantly attenuated or even missing, resulting in a significant direction-dependent detection blind zone. In addition, the near-surface blind zone formed by the residual vibration of the emitted pulse and the interface echo coverage further limits the reliable detection capability of this method for micro-defects in high-incidence areas.

[0004] However, the diffraction time-of-flight method has limitations in capturing and processing the extremely weak diffraction signals generated at the tip of microcracks, and the resulting near-surface resolution dilemma. The energy of the diffracted waves at the tip of microcracks is extremely low. In addition, the signal attenuates with distance in the material and is sensitive to the probe distance, which places stringent requirements on the instrument gain and probe sensitivity. Ultimately, the lower limit of detection sensitivity is constrained by physical laws and is difficult to improve. At the same time, high-precision depth quantification depends on accurately measuring the nanosecond-level time difference between the diffracted wave and the through wave. This is directly limited by the instrument's sampling rate, clock accuracy, and noise level, resulting in a significant increase in the axial positioning error of small-sized cracks. Summary of the Invention

[0005] The purpose of this invention is to provide an ultrasonic detection method for metal microcracks to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an ultrasonic detection method for metal microcracks, comprising the following modules:

[0007] The ultra-high frequency composite array probe has a dynamically adjustable center frequency range of 5-50MHz.

[0008] A dual-mode excitation and reception module is used to independently generate pulse reflection mode excitation signals and diffraction mode chirped frequency modulation signals;

[0009] A phased-array synthetic aperture dynamic focusing module enables ±60° deflection of the acoustic beam and subwavelength-level focusing.

[0010] The nonlinear harmonic enhancement module excites the nonlinear effect of cracks by superimposing a 1-2MHz fundamental wave;

[0011] Time-space-frequency three-domain data fusion module.

[0012] As a further technical solution of the present invention, the ultra-high frequency composite array probe adopts aluminum nitride piezoelectric composite array elements and integrates an acoustic metamaterial matching layer and a backing damping sound-absorbing layer, with a bandwidth ≥150% and a near-surface detection blind zone ≤0.2mm.

[0013] As a further technical solution of the present invention, the bandwidth of the diffraction mode chirped frequency modulation signal is 80%-120% of the probe center frequency, the time width is 10-100μs, and the coherence gain of the diffraction signal is ≥20dB through a matched filter.

[0014] As a further technical solution of the present invention, the nonlinear harmonic enhancement module superimposes the fundamental wave in the pulse reflection channel, and the receiving end filters out the fundamental wave component through a band-stop filter to extract the 2-4MHz second harmonic signal.

[0015] As a further technical solution of the present invention, the phased array synthetic aperture dynamic focusing module performs the following steps:

[0016] S1: Calculate the potential orientation distribution of microcracks;

[0017] S2: Dynamically adjust the array aperture and deflection angle so that the beam focal diameter is ≤λ / 2 (λ is the current center frequency wavelength).

[0018] S3: Perform synthetic aperture scanning on the focused area.

[0019] As a further technical solution of the present invention, the time-space-frequency three-domain data fusion module fuses the following data through a weighting function:

[0020] Spatiotemporal domain: Pulse reflection data processed by SAFT;

[0021] Frequency domain: Nonlinear harmonic separation data;

[0022] Time domain: Broadband cross-correlation delay data of diffraction modes.

[0023] As a further technical solution of the present invention, the weighting function satisfies: ,in denoted as the time-frequency domain signal-to-noise ratio, where A is the normalized amplitude of the signal, and α, β, and γ are adaptive coefficients.

[0024] As a further technical solution of the present invention, the embedded deep learning classifier adopts the U-Net network architecture, the input is a three-dimensional data volume (depth × azimuth × frequency) after three-domain fusion, and the output is a probability distribution map of crack size and orientation.

[0025] As a further technical solution of the present invention, it also includes a system-integrated FPGA+GPU heterogeneous computing architecture to achieve real-time processing latency of ≤50ms for dual-mode data streams, and embeds a reference test block feedback mechanism to automatically compensate for coupling fluctuations.

[0026] An ultrasonic detection method for metal microcracks includes the following steps:

[0027] S1: The ultra-high frequency probe scans the detection area and simultaneously activates pulse reflection and diffraction modes;

[0028] S2: Perform nonlinear harmonic separation and dynamic focusing on the pulse reflection data;

[0029] S3: Perform matched filtering and time delay measurement on the chirped signal of the diffraction mode;

[0030] S4: Fuse the data from the three domains and input them into a deep learning classifier to generate a crack quantification report.

[0031] The beneficial effects of this invention are as follows:

[0032] (1) This invention achieves a leap in nanoscale acoustic sensitivity through the synergistic innovation of high-frequency subwavelength focusing and broadband signal processing. The high-frequency array reduces the wavelength to the micrometer level, while the subwavelength focusing accurately concentrates the acoustic energy at the crack tip, directly overcoming the signal-to-noise ratio collapse caused by diffraction effect in traditional methods. More importantly, the wide bandwidth coverage of the chirped frequency signal of 80%-120% enables the system to have both low-frequency penetration and high-frequency resolution. It not only increases the diffraction signal gain by more than 20dB, but also compresses the near-surface blind zone to within 0.2mm, solving the core pain point of missing near-surface microcracks in thin-walled components.

[0033] (2) This invention constructs a three-dimensional collaborative diagnostic paradigm of time-space-frequency, achieving omnidirectional crack tracking through ±60° dynamic deflection of the sound beam. Adaptive focusing eliminates the directional blind spots of traditional methods, while nonlinear harmonic technology opens up a dimension of material damage perception independent of geometric morphology. Second harmonics have a specific response to crack opening and closing behavior, making closed cracks, which have long plagued the industry, clearly visible in strong background noise. The most innovative aspect is the three-domain data fusion strategy: spatiotemporal domain data accurately locates the defect coordinates, frequency domain data reveals the internal stress state of the material, and time domain data captures the dynamic behavior of microcrack tips.

[0034] (3) This invention creates a closed-loop processing link with hardware and software collaboration. The FPGA+GPU heterogeneous architecture compresses the dual-mode signal processing delay to the 50ms level, making millisecond-level crack dynamic tracking possible. The deep neural network performs pixel-level analysis of the three-dimensional data volume, realizing the automated quantitative analysis of submicron cracks. The detection efficiency is improved by two orders of magnitude compared to manual detection. The embedded feedback mechanism is particularly important. By calibrating the coupling state in real time through the reference test block, it overcomes the detection drift problem caused by the coupling fluctuation of the probe in the industrial field and ensures long-term monitoring stability. These three elements together constitute a new paradigm of "end-to-end intelligent detection". The hardware layer meets the real-time requirements of the production line, the algorithm layer achieves expert-level recognition accuracy, and the system layer ensures reliable operation at the field level. Attached Figure Description

[0035] Figure 1 This is a flowchart illustrating the overall system architecture of the present invention.

[0036] Figure 2 This is a diagram of the layered structure of the composite probe of the present invention;

[0037] Figure 3 This is a flowchart of the dual-mode signal processing of the present invention;

[0038] Figure 4 This is the dynamic focusing control logic diagram of the present invention;

[0039] Figure 5 This is a block diagram illustrating the principle of three-domain data fusion in this invention.

[0040] Figure 6 This is a real-time processing pipeline diagram for the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] like Figures 1 to 6 As shown in the embodiment of the present invention, an ultrasonic detection method for metal microcracks includes the following modules:

[0043] The ultra-high frequency composite array probe 101 has a dynamically adjustable center frequency range of 5-50MHz.

[0044] The dual-mode excitation and reception module 102 is used to independently generate pulse reflection mode excitation signal and diffraction mode chirped frequency modulation signal;

[0045] The phased-array synthetic aperture dynamic focusing module 103 enables ±60° deflection of the acoustic beam and subwavelength-level focusing.

[0046] The nonlinear harmonic enhancement module 104 excites the nonlinear effect of cracks by superimposing a 1-2MHz fundamental wave;

[0047] Time-space-frequency three-domain data fusion module 105.

[0048] The ultra-high frequency dynamically adjustable probe significantly improves the ability to distinguish micro-cracks in metals. The unique dual-mode signal design simultaneously captures the geometric features and subtle nonlinear responses of the crack. Combined with wide-angle beam deflection and microscopic focusing technology, it enables precise localization of defects in complex areas. By using the low-frequency fundamental wave to excite high-order harmonic characteristics, it effectively enhances the identification sensitivity of closed cracks or stress concentration areas. Finally, it integrates three-dimensional information of signal propagation time, spatial location, and spectral characteristics, breaking through the limitations of traditional methods and greatly improving the reliable detection capability and quantitative assessment accuracy of early micro-damage and hidden defects.

[0049] Among them, the ultra-high frequency composite array probe 101 adopts aluminum nitride piezoelectric composite array elements and integrates an acoustic metamaterial matching layer 101a and a backing damping sound-absorbing layer 101b, with a bandwidth ≥150% and a near-surface detection blind zone ≤0.2mm.

[0050] The core structure of this probe significantly enhances its detection performance and application value. The high-frequency piezoelectric composite unit ensures wideband excitation efficiency, and its composite acoustic matching structure enables efficient energy transfer between ultrasonic waves and the metal surface, effectively suppressing the probe's own reverberation interference. The special sound-absorbing backing design greatly shortens the pulse after-resonance time. Combined with the acoustic optimization layer, it breaks through to reduce the near-field detection blind zone to the sub-millimeter level, making it possible to accurately identify micro-defects on the surface of thin-walled components. The wideband response characteristics provide sufficient signal spectrum support for imaging the fine structure of materials. The overall technical indicators lay a key hardware foundation for high-resolution non-destructive testing in complex environments.

[0051] The bandwidth of the chirped frequency modulated signal in the diffraction mode is 80%-120% of the center frequency of the probe, and the time width is 10-100μs. The coherence gain of the diffraction signal is ≥20dB through a matched filter.

[0052] This signal design significantly enhances the detection capability of microcrack diffraction features. The broadband time-frequency coupling characteristic breaks through the limitations of traditional narrow pulse signals. By exciting the high-frequency scattering response at the crack tip through a frequency scanning mechanism, it enhances the sensitivity to capture weak diffraction waves. Matched filtering processing realizes a signal compression technique with a large time-bandwidth product, accurately focusing the diffraction energy point in the time domain and effectively suppressing the detection background noise by two orders of magnitude. The long-duration linear frequency modulation structure provides a sufficient time window for the interaction between the sound wave and the microcrack structure, improving the spatial sampling rate for defect orientation identification.

[0053] In this module, the nonlinear harmonic enhancement module 104 superimposes the fundamental wave in the pulse reflection channel, and the receiver filters out the fundamental wave component through a band-stop filter to extract the 2-4MHz second harmonic signal.

[0054] By cleverly exciting and separating the unique high-order harmonic response of microcracks, it is significantly different from the traditional linear ultrasonic testing principle. Special waveform processing technology effectively suppresses strong fundamental frequency interference and accurately extracts the nonlinear acoustic fingerprint generated by the opening and closing of micron-level cracks inside the metal. This mechanism has a unique advantage in identifying closed cracks and grain boundary slip that are difficult to capture by conventional methods, making it possible to visualize early fatigue damage in stress concentration areas. Its core technological breakthrough lies in converting the nonlinear effect of materials into a quantifiable detection signal, providing a new diagnostic dimension for the study of microscopic damage evolution in metal components.

[0055] The phase-controlled synthetic aperture dynamic focusing module 103 performs the following operations:

[0056] S1: Calculate the potential orientation distribution of microcracks;

[0057] S2: Dynamically adjust the array aperture and deflection angle so that the beam focal diameter is ≤λ / 2 (λ is the current center frequency wavelength).

[0058] S3: Perform synthetic aperture scanning on the focused area.

[0059] By intelligently optimizing the beam path based on a defect prediction model, the limitations of traditional fixed-angle scanning are overcome, enabling ultrasonic waves to actively track microcracks with complex spatial distributions. The beam's microscopic focusing capability surpasses the conventional diffraction limit, achieving a resolution level at the material grain scale. The synchronously executed dynamic aperture control and scanning imaging form a closed-loop detection system, ensuring that defects of any orientation receive the best signal-to-noise ratio echo response. The core value of this technology lies in improving acoustic detection accuracy to the microscopic deformation level, effectively solving the problem of missed detection of multi-directional cracks.

[0060] The time-space-frequency three-domain data fusion module 105 fuses the following data using a weighting function:

[0061] Spatiotemporal domain: Pulse reflection data processed by SAFT;

[0062] Frequency domain: Nonlinear harmonic separation data;

[0063] Time domain: Broadband cross-correlation delay data of diffraction modes.

[0064] By integrating complementary information from three dimensions of ultrasonic waves—time transmission sequence, spatial distribution characteristics, and spectral response characteristics—a holographic defect analysis model is constructed. The spatiotemporal signal accurately maps the geometric morphology of the crack, the harmonic spectrum characterizes the nonlinear damage state of the material, and the diffraction characteristics reveal the details of the microstructure. The weighted algorithm intelligently balances the confidence of data in each dimension, effectively overcoming the effects of noise interference and signal attenuation in metal components.

[0065] The weighting function satisfies: ,in denoted as the time-frequency domain signal-to-noise ratio, where A is the normalized amplitude of the signal, and α, β, and γ are adaptive coefficients.

[0066] By constructing a dynamic weight allocation strategy based on signal quality and intensity characteristics, it has the ability to adjust in real time to adapt to environmental noise, highlight the reliable data contribution in areas with superior signal-to-noise ratio, and compensate for the effective weight of weak diffraction signals. The exponential function structure accurately balances the nonlinear relationship between signal intensity and detection confidence, preventing misjudgment caused by high-amplitude noise interference. The core lies in establishing a quantitative evaluation system so that crack echoes with different physical properties can obtain the optimal fusion ratio according to actual detection conditions, ensuring that microcrack features are most completely expressed in the multi-dimensional data space.

[0067] Among them, the embedded deep learning classifier 106 adopts the U-Net network architecture. The input is a three-dimensional data volume (depth × azimuth × frequency) after three-domain fusion, and the output is a probability distribution map of crack size and orientation.

[0068] By employing a deep neural network architecture to achieve end-to-end parsing of fused data, the system intelligently mines deep crack patterns in a three-dimensional feature space. Its unique encoding and decoding structure precisely decouples defect signals from complex background noise, enabling pixel-level segmentation of sub-millimeter damage. Through autonomous learning of physical laws in massive samples, the system maps multi-dimensional ultrasonic features into precise geometric parameters and spatial orientation probability fields.

[0069] This includes a system-integrated FPGA+GPU heterogeneous computing architecture, which enables real-time processing of dual-mode data streams with a latency of ≤50ms, and embeds a reference test block feedback mechanism 107 to automatically compensate for coupling fluctuations.

[0070] By using heterogeneous computing units to collaboratively achieve parallel processing of multimodal signals, the system meets the stringent real-time requirements of industrial online detection. The embedded feedback unit automatically corrects measurement deviations introduced by changes in probe coupling state, ensuring the stability and repeatability of long-term monitoring data. The overall solution overcomes the core bottleneck of traditional ultrasonic equipment being difficult to deploy on production lines, namely the problems of time delay and coupling error, making millisecond-level dynamic tracking of sub-millimeter cracks a reality.

[0071] An ultrasonic detection method for metal microcracks includes the following steps:

[0072] S1: The ultra-high frequency probe scans the detection area and simultaneously activates pulse reflection and diffraction modes;

[0073] S2: Perform nonlinear harmonic separation and dynamic focusing on the pulse reflection data;

[0074] S3: Perform matched filtering and time delay measurement on the chirped signal of the diffraction mode;

[0075] S4: Fuse the data from the three domains and input them into a deep learning classifier to generate a crack quantification report.

[0076] By synchronously acquiring and processing multimodal signals, the geometric morphology, spatial orientation, and nonlinear characteristics of the crack are captured simultaneously in a single scan. The innovative closed-loop processing flow transforms acoustic physical effects into quantifiable damage indicators. Combined with the three-dimensional spatial analysis capability of deep neural networks, it enables automatic identification and parameter reconstruction of sub-millimeter-level cracks. The entire solution breaks through the dependence of traditional ultrasonic testing on operator experience and stably outputs quantitative damage maps under industrial noise backgrounds, solving the industry challenges of closed crack identification and micron-level damage tracking.

[0077] Through the synergistic innovation of high-frequency subwavelength focusing and broadband signal processing, a leap in nanoscale acoustic sensitivity has been achieved. The high-frequency array reduces the wavelength to the micrometer level, while the subwavelength focusing precisely concentrates the acoustic energy at the crack tip, directly overcoming the signal-to-noise ratio collapse caused by diffraction effects in traditional methods. More importantly, the 80%-120% broadband coverage of the chirped frequency signal enables the system to simultaneously possess low-frequency penetration and high-frequency resolution. This not only increases the diffraction signal gain by more than 20dB, but also compresses the near-surface blind zone to within 0.2mm, solving the core pain point of missed detection of near-surface microcracks in thin-walled components.

[0078] By constructing a three-dimensional collaborative diagnostic paradigm of time, space, and frequency, omnidirectional crack tracking is achieved through ±60° dynamic deflection of the sound beam. Adaptive focusing eliminates the directional blind spots of traditional methods, while nonlinear harmonic technology opens up a dimension of material damage perception independent of geometric morphology. Second harmonics have a specific response to crack opening and closing behavior, making closed cracks, which have long plagued the industry, clearly visible in strong background noise. The most innovative aspect is the three-domain data fusion strategy: spatiotemporal domain data accurately locates the defect coordinates, frequency domain data reveals the internal stress state of the material, and time domain data captures the dynamic behavior of microcrack tips.

[0079] By creating a closed-loop processing link that integrates hardware and software, the FPGA+GPU heterogeneous architecture compresses the dual-mode signal processing latency to the 50ms level, making millisecond-level crack dynamic tracking possible. Meanwhile, the deep neural network performs pixel-level analysis of the three-dimensional data volume, enabling automated quantitative detection of submicron cracks. The detection efficiency is improved by two orders of magnitude compared to manual detection. Crucially, the embedded feedback mechanism, which uses a reference test block to calibrate the coupling state in real time, overcomes the detection drift problem caused by probe coupling fluctuations in industrial settings, ensuring long-term monitoring stability. These three elements together constitute a new paradigm of "end-to-end intelligent detection": the hardware layer meets the real-time requirements of the production line, the algorithm layer achieves expert-level recognition accuracy, and the system layer ensures reliable operation at the field level.

[0080] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An ultrasonic detection method for metal microcracks, characterized in that: Includes the following modules: The ultra-high frequency composite array probe (101) has a dynamically adjustable center frequency range of 5-50MHz. A dual-mode excitation and reception module (102) is used to independently generate pulse reflection mode excitation signals and diffraction mode chirped frequency modulation signals; The phased-array synthetic aperture dynamic focusing module (103) enables ±60° deflection of the acoustic beam and subwavelength-level focusing. The nonlinear harmonic enhancement module (104) excites the nonlinear effect of cracks by superimposing a 1-2MHz fundamental wave; Time-space-frequency three-domain data fusion module (105).

2. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: The ultra-high frequency composite array probe (101) adopts aluminum nitride piezoelectric composite array elements and integrates an acoustic metamaterial matching layer (101a) and a backing damping sound-absorbing layer (101b), with a bandwidth ≥150% and a near-surface detection blind zone ≤0.2mm.

3. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: The bandwidth of the diffraction mode chirped frequency modulation signal is 80%-120% of the probe center frequency, and the time width is 10-100μs. The coherence gain of the diffraction signal is ≥20dB through a matched filter.

4. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: The nonlinear harmonic enhancement module (104) superimposes the fundamental wave in the pulse reflection channel, and the receiving end filters out the fundamental wave component through a band-stop filter to extract the 2-4MHz second harmonic signal.

5. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: The phased array synthetic aperture dynamic focusing module (103) performs the following steps: S1: Calculate the potential orientation distribution of microcracks; S2: Dynamically adjust the array aperture and deflection angle so that the beam focal diameter is ≤λ / 2 (λ is the current center frequency wavelength). S3: Perform synthetic aperture scanning on the focused area.

6. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: The time-space-frequency three-domain data fusion module (105) fuses the following data using a weighting function: Spatiotemporal domain: Pulse reflection data processed by SAFT; Frequency domain: Nonlinear harmonic separation data; Time domain: Broadband cross-correlation delay data of diffraction modes.

7. The ultrasonic detection method for metal microcracks according to claim 6, characterized in that: The weighting function satisfies: ,in denoted as the time-frequency domain signal-to-noise ratio, where A is the normalized amplitude of the signal, and α, β, and γ are adaptive coefficients.

8. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: Also includes: An embedded deep learning classifier (106) is adopted with a U-Net network architecture. The input is a three-dimensional data volume (depth × azimuth × frequency) after three-domain fusion, and the output is a probability distribution map of crack size and orientation.

9. The ultrasonic detection method for metal microcracks according to claim 1, characterized in that: It also includes a system-integrated FPGA+GPU heterogeneous computing architecture to achieve real-time processing latency of ≤50ms for dual-mode data streams, and embeds a reference test block feedback mechanism (107) to automatically compensate for coupling fluctuations.

10. The ultrasonic detection method for metal microcracks according to any one of claims 1-9, characterized in that: Includes the following steps: S1: The ultra-high frequency probe scans the detection area and simultaneously activates pulse reflection and diffraction modes; S2: Perform nonlinear harmonic separation and dynamic focusing on the pulse reflection data; S3: Perform matched filtering and time delay measurement on the chirped signal of the diffraction mode; S4: Fuse the data from the three domains and input them into a deep learning classifier to generate a crack quantification report.

Images