Cladding layer surface and deep defect detection method based on pulse laser

Abstract

The invention relates to the technical field of metal surface coating preparation, and discloses a cladding layer surface and deep defect detection method based on pulse laser, which comprises the following steps: clamping and fixing a workpiece; pulse laser excitation and bimodal signal synchronous excitation are carried out; synchronously acquiring and recording dual-channel signals; thermal wave signal processing and surface / near-surface flaw analysis; performing ultrasonic signal processing and deep defect analysis; and carrying out information fusion and three-dimensional flaw reconstruction. According to the invention, the pulse laser beam is utilized to excite the thermal wave and the ultrasonic wave at the same time, various flaws from the surface to the deep layer can be covered through one-time detection, the limitation of a single detection mode is solved, and meanwhile, the depth information of the flaws can be obtained by analyzing the phase diagram of the thermal wave signal; three-dimensional distribution of flaws can be reconstructed by combining ultrasonic time-of-flight (ToF) analysis and infrared sequence images, and accurate positioning and quantitative evaluation are realized.

Description

Technical Field

[0001] This invention relates to the field of metal surface coating inspection technology, and more specifically, to a method for detecting surface and deep defects in cladding layers based on pulsed lasers. Background Technology

[0002] Laser cladding technology is widely used in the repair, remanufacturing, and surface strengthening of critical components. However, during the cladding process, due to factors such as fluctuations in process parameters and differences in material properties, defects such as pores, incomplete fusion, and cracks are easily generated inside and on the surface of the cladding layer. These defects seriously affect the service performance and service life of the workpiece.

[0003] Currently, common cladding layer detection methods such as penetrant testing are only applicable to surface-opening defects, while magnetic particle testing is suitable for surface and near-surface defects in ferromagnetic materials but is insensitive to deep defects. X-ray testing, although capable of detecting internal defects, suffers from drawbacks including expensive equipment, radiation risks, and low detection rates for crack-like planar defects. Ultrasonic testing typically requires a coupling agent, and its coupling effect is poor and the signal-to-noise ratio is low on rough cladding layer surfaces.

[0004] Therefore, proposing a method for detecting surface and deep defects in cladding layers based on pulsed lasers has significant practical implications. Summary of the Invention

[0005] In view of this, the present invention proposes a method for detecting surface and deep defects in cladding layers based on pulsed laser, aiming to solve at least one of the problems in the background art.

[0006] This invention proposes a method for detecting surface and deep defects in cladding layers based on pulsed lasers, comprising the following steps: The workpiece is pre-processed and then clamped and fixed on the moving platform; A laser beam is emitted toward the region to be detected, and the temperature field change sequence of the surface of the region to be detected and the out-of-plane vibration displacement signal of the center of the region to be detected are collected to obtain an infrared thermal image sequence and an ultrasonic time domain signal. The infrared thermal image is first processed and analyzed to identify the location and two-dimensional morphology of surface and near-surface defects in the area to be detected; The ultrasonic time-domain signal is subjected to a second processing analysis to identify deep defect information; Spatial registration is performed on the location and two-dimensional morphology of surface and near-surface defects in the area to be detected, as well as the information on deep defects. For the surface and near-surface defects, depth information is calculated and assigned. For the deep defects, their projection position on the horizontal plane is associated with the infrared thermal image, and their depth is calculated and assigned. Then, all defect points are aggregated into a three-dimensional spatial point set. The three-dimensional spatial point set is reconstructed to obtain a visualized three-dimensional defect model of the region to be detected.

[0007] Preferably, the pretreatment specifically involves: polishing the area to be tested with sandpaper and then ultrasonically cleaning it with anhydrous ethanol.

[0008] Preferably, the pulse width of the laser beam is 10ns-100ns, and the energy density is 0.1mJ / cm2-10mJ / cm2.

[0009] Preferably, the frame rate for acquiring the temperature field change sequence of the surface of the area to be detected is 100-500Hz, and the acquisition duration is 0.5-5s.

[0010] Preferably, the sampling rate for acquiring the out-of-plane vibration displacement signal of the point in the region to be detected is 100MHz, and the acquisition duration is 50-200us.

[0011] Preferably, the first processing analysis is as follows: A fast Fourier transform is performed on the temperature-time curve of each pixel in the infrared thermal image to obtain the phase-frequency spectrum of the corresponding pixel. The low-frequency components in the phase-frequency spectrum are selected to generate the corresponding phase diagram; By analyzing the abnormal regions in the phase image and combining them with image segmentation algorithms, the location and two-dimensional morphology of surface and near-surface defects are identified.

[0012] Preferably, the second processing analysis is as follows: The ultrasonic time-domain signal is bandpass filtered, and the filtered signal is subjected to wavelet transform to obtain the time spectrum. By analyzing the abnormal echo signals that appear in the time spectrum, reflected waves or scattered waves from deep defects are identified.

[0013] Preferably, the method for identifying the abnormal echo signal is as follows: The time spectrum and the reference time spectrum are differentially calculated within the same time-frequency window. When a new time-frequency energy concentration region appears between the direct wave and the bottom echo, and the energy amplitude of the region exceeds a preset threshold, the region is identified as an abnormal echo signal.

[0014] Preferably, when correlating the projection position of the deep defect on the horizontal plane with the infrared thermal image and calculating its depth, the depth is calculated using the following formula:

[0015] Where d is the depth of the deep defect, v is the propagation speed of the ultrasonic wave in the cladding material, and Δt is the time difference between the defect echo and the initial surface wave / direct wave.

[0016] Preferably, the method for detecting surface and deep defects in the cladding layer based on pulsed laser further includes the following steps: Texture features are extracted from the generated phase map, and time-domain and frequency-domain features are extracted from the ultrasonic time-domain signal; The extracted texture features, temporal features, and frequency domain feature vectors are input into a deep learning classification model to obtain the defect type.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention utilizes a pulsed laser beam to simultaneously excite thermal waves and ultrasonic waves, enabling a single detection to cover various defects from the surface to the depths, thus overcoming the limitations of a single detection mode.

[0018] (2) The entire detection process of this invention does not require coupling agent, thus avoiding the problems caused by contact detection.

[0019] (3) By analyzing the phase diagram of the thermal wave signal, the present invention can obtain the depth information of the defect; by combining ultrasonic time-of-flight (ToF) analysis and infrared sequence images, the three-dimensional distribution of the defect can be reconstructed, so as to achieve accurate positioning and quantitative evaluation.

[0020] (4) The present invention uses the pulse phase method (PPT) to process thermal image sequences, which effectively suppresses the interference of uneven surface emissivity. Detailed Implementation

[0021] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as a limitation of the present invention, but rather as a more detailed description of certain aspects, features, and embodiments of the present invention. It should be understood that the terminology used in this invention is merely for describing particular embodiments and is not intended to limit the present invention.

[0022] Furthermore, regarding the numerical ranges in this invention, it should be understood that each intermediate value between the upper and lower limits of the range is also specifically disclosed. Every smaller range between any stated value or intermediate value within a stated range, and any other stated value or intermediate value within said range, is also included within this invention. The upper and lower limits of these smaller ranges may be independently included or excluded from the range.

[0023] Unless otherwise stated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. While only preferred methods and materials have been described herein, any methods and materials similar or equivalent to those described herein may be used in the implementation or testing of this invention. All references to this specification are incorporated by way of citation to disclose and describe methods and / or materials associated with those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.

[0024] Various modifications and variations can be made to the specific embodiments described in this specification without departing from the scope or spirit of the invention, as will be apparent to those skilled in the art. Other embodiments derived from this specification will also be obvious to those skilled in the art. This specification and embodiments are merely exemplary.

[0025] The terms “include,” “including,” “have,” “contain,” etc., used in this article are all open-ended terms, meaning that they include but are not limited to.

[0026] This invention provides a method for detecting surface and deep defects in cladding layers based on pulsed lasers, comprising the following steps: The workpiece is pre-processed and then clamped and fixed on the moving platform; A laser beam is emitted toward the region to be detected, and the temperature field change sequence of the surface of the region to be detected and the out-of-plane vibration displacement signal of the center of the region to be detected are collected to obtain an infrared thermal image sequence and an ultrasonic time domain signal. The infrared thermal image is first processed and analyzed to identify the location and two-dimensional morphology of surface and near-surface defects in the area to be detected; The ultrasonic time-domain signal is subjected to a second processing analysis to identify deep defect information; Spatial registration is performed on the location and two-dimensional morphology of surface and near-surface defects in the area to be detected, as well as the information on deep defects. For the surface and near-surface defects, depth information is calculated and assigned. For the deep defects, their projection position on the horizontal plane is associated with the infrared thermal image, and their depth is calculated and assigned. Then, all defect points are aggregated into a three-dimensional spatial point set. The three-dimensional spatial point set is reconstructed to obtain a visualized three-dimensional defect model of the region to be detected.

[0027] The preferred tools used in the pulsed laser-based method for detecting surface and deep defects in cladding layers according to this invention are: a three-dimensional moving platform, a pulsed laser, an infrared thermal imager, a laser ultrasonic interferometer, a synchronous trigger, and a control computer.

[0028] Specifically, the area to be inspected of the workpiece is pre-processed and then clamped and fixed on a three-dimensional moving platform; The pulsed laser is controlled by the synchronous trigger to emit a short-pulse, low-energy laser beam (ensuring no damage to the cladding layer) towards the area to be tested in the cladding layer.

[0029] It is understandable that the laser pulse acts on the surface of the cladding layer, producing two physical effects: Thermal effect: Laser energy is absorbed by the surface, causing a local instantaneous temperature rise and forming a heat wave that propagates inward.

[0030] Mechanical effect: Due to the thermoelastic effect, the instantaneous thermal expansion in this region excites broadband ultrasonic waves (mainly including longitudinal waves, transverse waves and surface waves) in the material.

[0031] Specifically, an infrared thermal imager is used to continuously record the temperature field change sequence of the surface of the area to be detected, i.e., an infrared thermal image sequence; a laser ultrasonic interferometer is used to align with the center point of the area to be detected and record the out-of-plane vibration displacement signal caused by the propagation of ultrasonic waves at that point, i.e., an ultrasonic time-domain signal.

[0032] Specifically, the acquired infrared thermal image sequences and ultrasonic time-domain signals are transmitted to a control computer for processing and analysis.

[0033] In this invention, the pretreatment specifically involves: polishing the area to be tested with sandpaper and then ultrasonically cleaning it with anhydrous ethanol.

[0034] Specifically, the surface of the cladding layer is uniformly polished using sandpaper with a grit size of 400# to 1000# to reduce surface roughness. Then, the workpiece is cleaned with anhydrous ethanol and an ultrasonic cleaner to remove surface grease and dust.

[0035] Understandably, surface contaminants such as oil can: alter the surface's thermophysical properties (e.g., emissivity, absorptivity), leading to distortion of the thermal field distribution during pulsed laser excitation and severely interfering with the accurate measurement of the surface temperature field by infrared thermal imagers; and form a damping layer in the laser-ultrasonic effect, absorbing some laser energy, suppressing and attenuating the effective excitation of ultrasonic waves, while also interfering with the detection of surface vibrations by laser interferometers. These interferences can directly introduce irrelevant noise signals and even mask real anomalies caused by internal defects, ultimately leading to misjudgments or missed detections. Therefore, rigorous surface cleaning is a necessary prerequisite for ensuring the validity and accuracy of subsequent dual-modal (thermal wave and ultrasonic) detection data.

[0036] In this invention, the pulse width of the laser beam is 10ns-100ns, and the energy density is 0.1mJ / cm². 2-10mJ / cm 2 .

[0037] Understandable is that 0.1 mJ / cm 2 The lower limit of energy density ensures that sufficiently strong thermal and ultrasonic signals can be excited, enabling effective detection by infrared thermal imagers and laser interferometers while maintaining a high signal-to-noise ratio; 10 mJ / cm 2 The upper limit of energy density ensures that the laser energy is far below the damage threshold of the cladding material (for typical nickel-based, cobalt-based, or iron-based alloy cladding layers, the ablation threshold is usually higher than 100 mJ / cm). 2 This ensures that the testing process is non-destructive.

[0038] In this invention, the frame rate for acquiring the temperature field change sequence of the surface of the area to be detected is 100-500Hz, and the acquisition duration is 0.5-5s.

[0039] Specifically, the infrared thermal imager continuously records the temperature field change sequence of the surface of the area to be detected at a high frame rate (100Hz-500Hz) starting from the moment the pulsed laser is emitted, for a duration of 0.5-5s, which is the infrared thermal image sequence.

[0040] Understandably, the high frame rate (100Hz-500Hz) setting aims to accurately sample the initial dynamics of heat wave diffusion from the surface to the interior of the cladding layer, ensuring the ability to distinguish rapid temperature changes caused by minor near-surface defects. The continuous recording time of 0.5 to 5 seconds ensures the observation of the complete thermal diffusion process as the heat wave propagates deep into the cladding layer and returns to the surface due to larger internal defects or incomplete interlayer fusion. The recorded infrared thermal image sequence essentially constitutes a complete dataset (x, y, t) containing spatial, temporal, and thermal attribute information. Specifically, this dataset is a three-dimensional structure, where two dimensions (x, y) represent the spatial pixel coordinates of the detection area, and the third dimension (t) represents the time series, i.e., the acquisition time point corresponding to each frame of the thermal image. Therefore, this dataset can be characterized as a data cube F(x, y, t), where each data point F(x... i y j， t k Records the location at a specific spatial position (x) i y j ) and a specific time t k The absolute or relative temperature value.

[0041] This data cube F(x, y, t) is the sole source of raw data for subsequent depth-resolution analysis. By analyzing each pixel (x... i y jThe time series F(t) at point (x, y, f) is subjected to a Fast Fourier Transform (F(t)) to transform it from the time domain F(t) to the frequency domain G(f), thereby introducing the implicit fourth dimension, frequency (f). The resulting core analytical object, the phase-frequency spectrum, is actually a four-dimensional data set Φ(x, y, f), which clearly reveals the phase response of thermal waves at different frequencies (corresponding to different detection depths) at various points in space.

[0042] It is by analyzing this four-dimensional dataset Φ(x, y, f) that the frequency components that are insensitive to surface emissivity can be effectively separated, and the phase contrast that is directly related to the depth of the defect can be extracted, thus laying an irreplaceable data foundation for the detection and depth estimation of defects from the surface to the depth.

[0043] In this invention, the sampling rate for the out-of-plane vibration displacement signal of the point in the region to be detected is 100MHz, and the sampling duration is 50-200us.

[0044] Specifically, after the pulsed laser is emitted, the laser ultrasonic interferometer is immediately aligned with the center point of the area to be tested and records the out-of-plane vibration displacement signal caused by the propagation of ultrasonic waves at the point with an ultra-high sampling rate (100MHz), i.e., the ultrasonic time domain signal, with a duration of 50μs-200μs.

[0045] Understandably, the ultra-high sampling rate of 100MHz is used to ensure distortion-free recording of the complete waveform of high-frequency ultrasound (typically ranging from several megahertz to tens of megahertz), with sufficient temporal resolution to accurately measure the flight time of ultrasound at minute geometric scales. The set recording time window of 50μs to 200μs aims to fully cover the entire journey of ultrasound from the excitation point to the bottom of the cladding layer and back to the surface, ensuring the acquisition of defect reflections or transmissions from the deepest region of interest. The out-of-plane vibration displacement signal recorded by the laser interferometer is a direct manifestation of the interaction between ultrasound and internal defects (such as pores or lack of fusion): when ultrasound encounters defects with abrupt changes in acoustic impedance, reflection, scattering, and mode conversion occur. These physical phenomena lead to anomalous waveform distortion, additional echo envelopes, or specific modal components in the time-domain signal captured at the receiving point. This high-fidelity ultrasonic time-domain signal is the sole sample for subsequent time-frequency analysis, defect echo identification, and depth estimation, and is crucial for achieving highly sensitive detection of deep volumetric and interface defects.

[0046] In this invention, the first processing analysis is as follows: A fast Fourier transform is performed on the temperature-time curve of each pixel in the infrared thermal image to obtain the phase-frequency spectrum of the corresponding pixel. The low-frequency components in the phase-frequency spectrum are selected to generate the corresponding phase diagram; By analyzing the abnormal regions in the phase image and combining them with image segmentation algorithms, the location and two-dimensional morphology of surface and near-surface defects are identified.

[0047] Specifically, in the phase diagram, areas with defects will exhibit a different phase from those without defects due to changes in the thermal diffusion rate, thus appearing as obvious bright or dark spots.

[0048] Specifically, the image segmentation algorithm is preferably the Otsu thresholding method or the region growing method.

[0049] Understandably, the pulse phase method transforms the temperature decay history of each pixel from the time domain to the frequency domain using Fast Fourier Transform. Leveraging the high sensitivity of the phase angle to changes in the material's internal thermal properties, it effectively suppresses thermal signal interference caused by surface roughness, oxidation state, and emissivity inhomogeneity. Subsequently, extracting specific low-frequency phase maps is based on the physical characteristics of thermal wave propagation: low-frequency thermal waves have a deeper detection depth, exhibiting significant phase lag or lead for thermal barriers (such as pores and lack of fusion) or thermal short circuits (such as cracks) existing beneath the surface, thus forming abnormal contrast regions in the phase map corresponding to the defect morphology. Finally, by automatically identifying and extracting the contours of these characteristic phase anomalies using image segmentation algorithms, the precise location, planar morphology, and distribution of defects such as surface microcracks and near-surface pores can be accurately determined, providing a reliable two-dimensional data foundation for subsequent 3D reconstruction and comprehensive evaluation.

[0050] In this invention, the second processing analysis is as follows: The ultrasonic time-domain signal is bandpass filtered, and the filtered signal is subjected to wavelet transform to obtain the time spectrum. By analyzing the abnormal echo signals that appear in the time spectrum, reflected waves or scattered waves from deep defects are identified.

[0051] Specifically, the ultrasonic time-domain signal is bandpass filtered to suppress noise and highlight defect echoes; the filtered signal is then subjected to wavelet transform to obtain the time spectrum; and by analyzing the abnormal echo signals appearing in the time spectrum, reflected or scattered waves from deep defects are identified.

[0052] Understandably, the role of the second processing analysis is to expand a single ultrasonic time-domain signal in the time-frequency domain to simultaneously acquire the joint characteristics of the signal in both time and frequency dimensions. This effectively separates and identifies weak ultrasonic echoes hidden in strong background noise caused by deep defects, and enables preliminary identification of the defect type. Specifically, the time-frequency spectrum obtained through wavelet transform can clearly reveal the specific mode changes produced when ultrasonic waves encounter defects of different depths and types during propagation: echoes from deep volumetric defects (such as pores) typically exhibit localized energy concentrations at specific time points, with frequency bands consistent with the main wave; while echoes from planar defects (such as incomplete fusion or cracks) may cause waveform distortion, enhancement of specific frequency components, or the appearance of new frequency components after mode conversion. This time-frequency analysis greatly improves the signal-to-noise ratio and discernibility of defect echoes, providing indispensable two-dimensional feature basis for subsequent accurate calculation of ultrasonic flight time and quantitative assessment of defect depth and its mechanical interaction properties.

[0053] In this invention, the method for identifying the abnormal echo signal is as follows: The time spectrum and the reference time spectrum are differentially calculated within the same time-frequency window. When a new time-frequency energy concentration region appears between the direct wave and the bottom echo, and the energy amplitude of the region exceeds a preset threshold, the region is identified as an abnormal echo signal.

[0054] Specifically, before testing, ultrasonic signals are collected on a standard test block that is known to be defect-free and has the same geometry as the material to be tested, using the same method, as a reference signal and reference time spectrum.

[0055] In this invention, when associating the projection position of the deep defect on the horizontal plane with the infrared thermal image and calculating its depth, the depth is calculated using the following formula:

[0056] Where d is the depth of the deep defect, v is the propagation speed of the ultrasonic wave in the cladding material, and Δt is the time difference between the defect echo and the initial surface wave / direct wave.

[0057] In this invention, the method for detecting surface and deep defects in cladding layers based on pulsed lasers further includes the following steps: Texture features are extracted from the generated phase map, and time-domain and frequency-domain features are extracted from the ultrasonic time-domain signal; The extracted texture features, temporal features, and frequency domain feature vectors are input into a deep learning classification model to obtain the defect type.

[0058] Specifically, from the phase image, local binary pattern (LBP) texture features and Haralick features based on gray-level co-occurrence matrix are extracted for each identified defect region to quantify its texture characteristics; morphological features, including area, perimeter, roundness and aspect ratio, are extracted from the location and two-dimensional morphology of the surface and near-surface defects in the region to be detected. From the ultrasonic time-domain signal and its time spectrum corresponding to the defect area, extract time-domain features (including peak amplitude of defect echo and time difference with the initial wave), frequency-domain features (including center frequency and energy of spectral sub-bands), and time-frequency-domain features (including wavelet coefficient energy). All extracted features are combined to form a high-dimensional, unified multimodal fusion feature vector; The multimodal fused feature vector is input into a convolutional neural network (CNN) classification model pre-trained with a large number of known defect samples. This model automatically calculates the probability that the defect belongs to a preset category (including "porosity," "crack," and "lack of fusion"). The category corresponding to the highest probability value output by the model is taken as the final classification result. When the highest probability value is higher than the confidence threshold of 0.95, the result is considered valid. If it is lower than the threshold, the defect is marked as "unknown type" and the system prompts for manual review.

[0059] Example 1 Workpiece under inspection: An Inconel 718 nickel-based superalloy turbine blade that has undergone laser cladding repair. The cladding layer thickness is approximately 1.5 mm.

[0060] Testing equipment: Pulsed laser: Nd:YAG pulsed laser, wavelength 1064nm, pulse width 20ns, maximum single pulse energy 200mJ.

[0061] Infrared thermal imager: Mid-wave infrared thermal imager, spectral range 3-5μm, frame rate 200 Hz, resolution 640×512 pixels.

[0062] Laser ultrasonic interferometer: A laser vibration meter based on the principle of multi-frequency heterodyne interference, with a measurement bandwidth ≥50MHz and a sampling rate of 100MHz.

[0063] Synchronous trigger: Digital delay pulse generator.

[0064] Data processing computer: A workstation equipped with a high-performance GPU.

[0065] Three-dimensional moving platform: an electric translation stage with controllable movement accuracy of 10μm.

[0066] Inspection steps: S1. Use sandpaper with a grit size of 400# to 1000# to uniformly polish the surface of the cladding layer of the workpiece to reduce the surface roughness. Then use anhydrous ethanol and an ultrasonic cleaner to clean the workpiece to remove surface grease and dust. S2. Fix the cleaned workpiece on the three-dimensional moving platform, adjust the output spot diameter of the pulsed laser to 5 mm, and adjust the detection spot of the laser ultrasonic interferometer to be located at the geometric center of the excitation region.

[0067] S3. Control the pulsed laser at 5mJ / cm² using a synchronous trigger. 2 The energy density emits a laser pulse with a pulse width of 50 ns, which is irradiated onto the area to be detected. The infrared thermal imager starts recording immediately after the laser pulse is triggered, and continuously acquires data for 3 seconds at a frame rate of 200 Hz to obtain infrared thermal sequence data containing 600 frames of infrared images. At the same time, the laser ultrasonic interferometer is started synchronously, and records the out-of-plane vibration displacement signal for 100 μs at a sampling rate of 100 MHz to obtain the ultrasonic time domain signal. S4. Perform a fast Fourier transform on the temperature-time curve of each pixel (640×512 pixels) in the 600-frame infrared thermal image sequence to obtain the phase-frequency spectrum. Select the 0.5Hz phase map for display and analysis. In this phase map, several obvious phase anomalous regions can be seen. Use the Otsu automatic thresholding method to process the 0.5 Hz phase map and successfully identify two suspected defective regions: one is linear (Region A) and the other is nearly circular (Region B).

[0068] S5. The ultrasonic time-domain signal is bandpass filtered from 1 MHz to 20 MHz to suppress low-frequency vibration noise and high-frequency electronic noise. A continuous wavelet transform is then performed on the filtered signal. The result is compared with the time-frequency spectrum of a reference signal acquired in a flawless substrate region. A significant anomalous echo envelope is found at 35.2 μs. Given that the longitudinal wave velocity of ultrasound in an Inconel 718 is 5800 m / s, according to the formula...

[0069] Where d is the depth of the deep defect, v is the propagation speed of the ultrasonic wave in the cladding layer material, and Δt is the time difference between the defect echo and the initial surface wave / direct wave. The calculated depth of the defect is approximately (5800 × 35.2 × 10⁻⁶). -6 ) / 2≈0.102mm.

[0070] S6. The identified linear defect Region A and near-circular defect Region B are spatially correlated with the abnormal echo signal with a depth of 0.102 mm found in step S5. The analysis confirms that the echo signal originates from Region B. The depth of Region A is initially marked as a surface defect (depth 0 mm), and Region B is assigned the calculated depth of 0.102 mm. Using a three-dimensional interpolation algorithm, a visualized three-dimensional defect model containing the spatial location and basic shape of the two defects is generated.

[0071] S7. Extract LBP texture features and Haralick features from the 0.5 Hz phase map regions corresponding to Region A and Region B, respectively. Extract time-frequency domain features (wavelet coefficient energy) from the ultrasound signal corresponding to Region B. Combine the features into two fused feature vectors corresponding to Region A and Region B, respectively. Input the two fused feature vectors into the pre-trained CNN classification model.

[0072] Model output probability vector: Region A: [Porosity: 0.05, Cracks: 0.92, Lack of Fusion: 0.03] - Classified as "Cracks".

[0073] Region B: [Porosity: 0.88, Cracks: 0.08, Lack of Fusion: 0.04] - Classified as "Porosity".

[0074] Both results have a confidence level higher than the 0.95 decision threshold, so the results are valid.

[0075] Example 2 Test workpiece: A 304 stainless steel thin-layer cladding specimen prepared on a low-carbon steel substrate, with a cladding thickness of approximately 0.5 mm.

[0076] Testing equipment: Pulsed laser: Nd:YAG pulsed laser, wavelength 1064nm, pulse width 10ns, maximum single pulse energy 200mJ.

[0077] Infrared thermal imager: Mid-wave infrared thermal imager, spectral range 8-14μm, frame rate 100 Hz, resolution 320×240 pixels.

[0078] Laser ultrasonic interferometer: laser vibration meter, sampling rate 100MHz.

[0079] Synchronous trigger: Digital delay pulse generator.

[0080] Data processing computer: A workstation equipped with a high-performance GPU.

[0081] Three-dimensional moving platform: an electric translation stage with controllable movement accuracy of 10μm.

[0082] Inspection steps: S1. Use sandpaper with a grit size of 400# to 1000# to uniformly polish the surface of the cladding layer of the workpiece to reduce the surface roughness. Then use anhydrous ethanol and an ultrasonic cleaner to clean the workpiece to remove surface grease and dust. S2. Fix the cleaned workpiece on the three-dimensional moving platform, adjust the output spot diameter of the pulsed laser to 3mm, and adjust the detection spot of the laser ultrasonic interferometer to be located at the geometric center of the excitation region.

[0083] S3. Control the pulsed laser at 0.1 mJ / cm² using a synchronous trigger. 2 The energy density of the laser pulse is emitted and irradiates the area to be detected. The infrared thermal imager starts recording immediately after the laser pulse is triggered, and continuously acquires data for 0.5 seconds at a frame rate of 100Hz to obtain infrared thermal sequence data containing 50 frames of infrared images. At the same time, the laser ultrasonic interferometer is started synchronously and records the out-of-plane vibration displacement signal for 50μs at a sampling rate of 100MHz to obtain the ultrasonic time domain signal. S4. Perform a fast Fourier transform on the temperature-time curve of each pixel in the 50-frame infrared thermal image sequence to obtain the phase-frequency spectrum. Select the 1Hz phase map for display and analysis. In this phase map, a weak phase anomalous region can be seen. Use the region growing method to segment the anomalous region and identify a nearly circular suspected defect region (Region A).

[0084] S5. The ultrasonic time-domain signal is bandpass filtered from 0.5MHz to 15MHz to suppress low-frequency vibration noise and high-frequency electronic noise. A continuous wavelet transform is then performed on the filtered signal. The result is compared with the time-spectrum of a reference signal acquired in a flawless substrate region. A identifiable anomalous echo is found at 12.5μs. Given that the longitudinal wave velocity of ultrasound in 304 stainless steel is 5640m / s, according to the formula...

[0085] Where d is the depth of the deep defect, v is the propagation speed of the ultrasonic wave in the cladding layer material, and Δt is the time difference between the defect echo and the initial surface wave / direct wave. The calculated depth of the defect is approximately (5640 × 12.5 × 10⁻⁶). -6 ) / 2≈0.035mm.

[0086] S6. Spatial correlation is performed between the identified Region A and the discovered abnormal echo signal with a depth of 0.035 mm to confirm that they are the same defect. Region A is assigned the calculated depth of 0.035 mm, and the position and outline of the defect are marked in three-dimensional space.

[0087] S7. Extract LBP texture features from the phase map region corresponding to Region A, extract the peak amplitude and center frequency of the defect echo from the corresponding ultrasonic signal, combine the features into a fused feature vector, and input the fused feature vector into a pre-trained CNN classification model.

[0088] The model output probability vector is: [porosity: 0.91, crack: 0.07, lack of fusion: 0.02] - judged as "porosity".

[0089] Example 3 The workpiece being inspected is a cobalt-based tungsten carbide composite cladding layer prepared on the surface of a large high-chromium cast iron roll. The cladding layer thickness is approximately 4.0 mm.

[0090] Testing equipment: Pulsed laser: Nd:YAG pulsed laser, wavelength 1064nm, pulse width 100ns.

[0091] Infrared thermal imager: Mid-wave infrared thermal imager, spectral range 3-5μm, frame rate 500Hz, resolution 1280×1024 pixels.

[0092] Laser ultrasonic interferometer: a high-performance laser vibrometer with a sampling rate of 100MHz.

[0093] Synchronous trigger: Digital delay pulse generator.

[0094] Data processing computer: A workstation equipped with a high-performance GPU.

[0095] Three-dimensional moving platform: an electric translation stage with controllable movement accuracy of 10μm.

[0096] Inspection steps: S1. Use sandpaper with a grit size of 400# to 1000# to uniformly polish the surface of the cladding layer of the workpiece to reduce the surface roughness. Then use anhydrous ethanol and an ultrasonic cleaner to clean the workpiece to remove surface grease and dust. S2. Fix the cleaned workpiece on the three-dimensional moving platform, adjust the output spot diameter of the pulsed laser to 10mm, and adjust the detection spot of the laser ultrasonic interferometer to be located at the geometric center of the excitation region.

[0097] S3. Control the pulsed laser at 10 mJ / cm² using a synchronous trigger. 2The energy density of the laser pulse is used to illuminate the area to be detected. The infrared thermal imager starts recording immediately after the laser pulse is triggered, and continuously acquires data for 5 seconds at a frame rate of 500Hz to obtain infrared thermal sequence data containing 2500 frames of infrared images. At the same time, the laser ultrasonic interferometer is started synchronously and records the out-of-plane vibration displacement signal for 200μs at a sampling rate of 100MHz to obtain the ultrasonic time domain signal. S4. Perform a fast Fourier transform on the temperature-time curve of each pixel in the 2500-frame infrared thermal image sequence to obtain the phase-frequency spectrum. Select the 0.2Hz phase map for display and analysis. In this phase map, a phase anomaly region with a large range and blurred boundaries can be seen, indicating the presence of a deep thermal barrier. Using an image segmentation algorithm (combined with adaptive thresholding and edge detection), the deep anomaly region (Region C) is successfully extracted from the background noise.

[0098] S5. The ultrasonic time-domain signal is bandpass filtered from 0.1MHz to 25MHz to suppress low-frequency vibration noise and high-frequency electronic noise. A continuous wavelet transform is then performed on the filtered signal. The result is compared with the time spectrum of a reference signal acquired in a flawless substrate region. Within a long 200μs time window, a wide and gentle low-frequency echo envelope is found at 142.8μs, whose characteristics match those of a deep interface reflection signal. Given that the average longitudinal wave velocity of ultrasound in a cobalt-based tungsten carbide cladding layer is 5600m / s, according to the formula:

[0099] Where d is the depth of the deep defect, v is the propagation speed of the ultrasonic wave in the cladding layer material, and Δt is the time difference between the defect echo and the initial surface wave / direct wave. The calculated depth of the defect is approximately (5600 × 142.8 × 10⁻⁶). -6 ) / 2≈3.998mm.

[0100] S6. Spatial correlation is performed between the identified Region C and the discovered abnormal echo signal with a depth of 3.998 mm to confirm that they are the same defect. Region C is assigned the calculated depth of 3.998 mm, and the lateral distribution range of the interface defect is reconstructed in three-dimensional space.

[0101] S7. Extract Haralick energy and homogeneous features representing large-area, slowly varying characteristics from the phase map region corresponding to Region C. Extract low center frequency and long pulse width features representing interface reflection from the ultrasound signal. Combine the features into a fused feature vector and input the fused feature vector into a pre-trained CNN classification model.

[0102] The model output probability vector is: [porosity: 0.12, cracks: 0.15, lack of fusion: 0.73] - judged as "lack of fusion".

[0103] As can be seen from the detection results of Examples 1-3, the pulsed laser-based method for detecting surface and deep defects in cladding layers of the present invention can cover various defects from the surface to the depth in a single detection, solving the limitations of a single detection mode, and achieving precise positioning and quantitative evaluation.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting surface and deep defects in a cladding layer based on pulsed laser, characterized in that, Includes the following steps: The workpiece is pre-processed and then clamped and fixed on the moving platform; A laser beam is emitted toward the region to be detected, and the temperature field change sequence of the surface of the region to be detected and the out-of-plane vibration displacement signal of the center of the region to be detected are collected to obtain an infrared thermal image sequence and an ultrasonic time domain signal. The infrared thermal image is first processed and analyzed to identify the location and two-dimensional morphology of surface and near-surface defects in the area to be detected; The ultrasonic time-domain signal is subjected to a second processing analysis to identify deep defect information; Spatial registration is performed on the location and two-dimensional morphology of surface and near-surface defects in the area to be detected, as well as the information on deep defects. For the surface and near-surface defects, depth information is calculated and assigned. For the deep defects, their projection position on the horizontal plane is associated with the infrared thermal image, and their depth is calculated and assigned. Then, all defect points are aggregated into a three-dimensional spatial point set. The three-dimensional spatial point set is reconstructed to obtain a visualized three-dimensional defect model of the region to be detected.

2. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 1, characterized in that, The pretreatment specifically involves: polishing the area to be tested with sandpaper and then ultrasonically cleaning it with anhydrous ethanol.

3. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 2, characterized in that, The laser beam has a pulse width of 10ns-100ns and an energy density of 0.1mJ / cm². 2 -10mJ / cm 2 .

4. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 3, characterized in that, The frame rate for acquiring the temperature field change sequence of the surface of the area to be detected is 100-500Hz, and the acquisition duration is 0.5-5s.

5. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 4, characterized in that, The sampling rate for the out-of-plane vibration displacement signal of the point in the area to be detected is 100MHz, and the sampling duration is 50-200us.

6. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 5, characterized in that, The first processing analysis is as follows: A fast Fourier transform is performed on the temperature-time curve of each pixel in the infrared thermal image to obtain the phase-frequency spectrum of the corresponding pixel. The low-frequency components in the phase-frequency spectrum are selected to generate the corresponding phase diagram; By analyzing the abnormal regions in the phase image and combining them with image segmentation algorithms, the location and two-dimensional morphology of surface and near-surface defects are identified.

7. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 6, characterized in that, The second processing analysis is as follows: The ultrasonic time-domain signal is bandpass filtered, and the filtered signal is subjected to wavelet transform to obtain the time spectrum. By analyzing the abnormal echo signals that appear in the time spectrum, reflected waves or scattered waves from deep defects are identified.

8. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 7, characterized in that, The method for identifying the abnormal echo signal is as follows: The time spectrum and the reference time spectrum are differentially calculated within the same time-frequency window. When a new time-frequency energy concentration region appears between the direct wave and the bottom echo, and the energy amplitude of the region exceeds a preset threshold, the region is identified as an abnormal echo signal.

9. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 8, characterized in that, When associating the projection position of the deep defect on the horizontal plane with the infrared thermal image and calculating its depth, the depth is calculated using the following formula: Where d is the depth of the deep defect, v is the propagation speed of the ultrasonic wave in the cladding material, and Δt is the time difference between the defect echo and the initial surface wave / direct wave.

10. The method for detecting surface and deep defects in cladding layers based on pulsed lasers according to claim 9, characterized in that, It also includes the following steps: Texture features are extracted from the generated phase map, and time-domain and frequency-domain features are extracted from the ultrasonic time-domain signal; The extracted texture features, temporal features, and frequency domain feature vectors are input into a deep learning classification model to obtain the defect type.

Images