Method and system for detecting welding compactness of diamond compact

Abstract

The invention relates to the technical field of diamond detection, and discloses a method and a system for detecting the welding compactness of a diamond compact. The method comprises the following steps: carrying out multi-modal data acquisition on a diamond compact welding sample; the data are input into a feature enhancement recognition algorithm to be processed, and three defect feature vector sets are obtained; performing spatial registration and data fusion on the feature vector set to obtain a fusion defect distribution diagram and a multi-modal defect feature matrix; and calculating a compactness evaluation index according to the defect characteristic matrix, and evaluating the welding quality grade. The problem that a traditional detection algorithm does not optimize special physical characteristics of the diamond composite material is solved, and meanwhile the problems that multi-modal detection data is difficult to effectively fuse and lacks objective evaluation standards are solved.

Description

Technical Field

[0001] This application relates to the field of diamond testing technology, and in particular to a method and system for testing the weld tightness of diamond composite sheets. Background Technology

[0002] Diamond composite sheets are engineering materials with high hardness, wear resistance, and thermal conductivity, widely used in oil drilling, mining, and precision machining. The welding quality of diamond composite sheets, especially the density of the weld interface, directly affects the product's service life and performance. Currently, the welding quality of diamond composite sheets is mainly inspected using methods such as ultrasonic testing, X-ray fluoroscopy, and thermal imaging analysis.

[0003] However, existing detection technologies have significant shortcomings. The sensitivity of a single detection method varies greatly depending on the type of defect, easily leading to missed detections and misjudgments. Ultrasonic testing is sensitive to interface cracks but has a low detection rate for small pores; X-ray imaging can detect density differences but has limited resolution; thermal imaging analysis can detect abnormal heat conduction but has poor depth resolution. Furthermore, traditional detection methods do not adequately consider the unique physical properties of diamond composite materials, such as high acoustic impedance, high density, and high thermal conductivity, resulting in poor signal quality and difficulties in data interpretation. Simultaneously, data obtained from different detection methods are difficult to integrate effectively, lacking a comprehensive evaluation system.

[0004] To address the aforementioned issues, a multimodal testing method is needed that can adapt to the unique material properties of diamond composite sheets, enabling effective processing and fusion of various test data. Simultaneously, an objective and quantitative evaluation system needs to be established to comprehensively assess the compactness of the weld interface. Furthermore, the spatial registration problem of different modal data needs to be solved to ensure the accuracy and consistency of the test results. Summary of the Invention

[0005] This application provides a method and system for detecting the weld tightness of diamond composite sheets, which solves the problem that traditional detection algorithms are not optimized for the special physical properties of diamond composite materials, and also solves the problems of difficulty in effectively integrating multimodal detection data and lack of objective evaluation standards.

[0006] In a first aspect, this application provides a method for detecting the weld tightness of diamond composite sheets. The method includes: acquiring multimodal data from the welded sample of the diamond composite sheet to be tested, obtaining an original ultrasonic dataset, an original X-ray dataset, and an original thermal imaging dataset; inputting the original ultrasonic dataset, original X-ray dataset, and original thermal imaging dataset into a diamond composite sheet weld interface feature enhancement and recognition algorithm for processing, obtaining an ultrasonic defect feature vector set, an X-ray defect feature vector set, and a thermal imaging defect feature vector set; performing spatial registration and data fusion on the ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set based on the diamond composite sheet weld interface feature enhancement and recognition algorithm, obtaining a fused defect distribution map and a multimodal defect feature matrix; and calculating a quantitative evaluation index of the weld interface tightness based on the multimodal defect feature matrix to obtain a weld quality grade assessment result.

[0007] Optionally, the diamond composite welded sample to be tested undergoes multimodal data acquisition to obtain a raw ultrasonic dataset, a raw X-ray dataset, and a raw thermal imaging dataset, including: The diamond composite sheet welding sample was placed on a multidimensional detection platform, and the sample surface was cleaned to obtain a cleaned sample. The ultrasonic testing parameters, X-ray fluoroscopy parameters, and thermal imaging analysis parameters are set for the cleaned sample to obtain a set of testing parameters. The sample is marked according to the set of detection parameters, and reference points are marked at the four corners and the center of the sample to obtain a test sample with reference points; The ultrasonic testing unit is used to perform a C-scan on the test sample with reference points, and a zigzag scanning path is used to cover the entire welding area to obtain the original ultrasonic dataset. The test sample with the reference point is placed in the X-ray fluoroscopy unit and rotated 360° for scanning. A projection image is acquired at 0.5° intervals to obtain the original X-ray dataset. The test sample with reference points is placed in the thermal imaging analysis unit, and the thermal excitation source is used to uniformly irradiate the welding area. Thermal image sequences are continuously acquired during and after the excitation process to obtain the original thermal imaging dataset.

[0008] Optionally, the step of inputting the original ultrasonic dataset, original X-ray dataset, and original thermal imaging dataset into a diamond composite sheet welding interface feature enhancement and recognition algorithm for processing to obtain ultrasonic defect feature vector sets, X-ray defect feature vector sets, and thermal imaging defect feature vector sets includes: The original ultrasonic dataset is subjected to bandpass filtering and waveform denoising to obtain preprocessed ultrasonic data; The acoustic impedance ratio correction factor of the diamond-metal interface is calculated based on the preprocessed ultrasonic data. The acoustic impedance ratio correction factor is determined by the ratio of the density and sound velocity of diamond to the density and sound velocity of the metal matrix, and the acoustic impedance ratio correction factor value is obtained. Based on the acoustic impedance ratio correction factor, a diamond characteristic compensation transformation function is constructed. The preprocessed ultrasonic data is then subjected to nonlinear transformation processing through the diamond characteristic compensation transformation function to obtain the compensated ultrasonic signal. The compensated ultrasonic signal is subjected to high-frequency feature extraction and Hilbert transform to extract the acoustic reflection intensity distribution map of the welding interface. The ultrasonic defect feature vector set is obtained by adaptive threshold segmentation. The X-ray attenuation correction model in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original X-ray dataset. Nonlinear transformation is performed through the density correction function. Three-dimensional reconstruction and nonlinear phase reconstruction are performed on the transformed image to obtain the X-ray defect feature vector set. An anisotropic thermal diffusion model from the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original thermal imaging dataset to calculate the corrected temperature field. Time-domain analysis and principal component analysis are then performed on the corrected thermal imaging sequence to obtain the thermal imaging defect feature vector set.

[0009] Optionally, the step of calculating the acoustic impedance ratio correction factor for the diamond-metal interface based on the preprocessed ultrasonic data, wherein the acoustic impedance ratio correction factor is determined by the ratio of the density and sound velocity of diamond to the density and sound velocity of the metal matrix, and the resulting acoustic impedance ratio correction factor value includes: Obtain the density and sound velocity values ​​of diamond material to obtain the basic parameters of diamond; The density and sound velocity of the metal matrix are measured to obtain the basic parameters of the metal matrix; The acoustic impedance of diamond is calculated by multiplying the density value and the sound velocity value in the basic diamond parameters. The acoustic impedance of the metal matrix is ​​calculated by multiplying the density value and the sound velocity value in the basic parameters of the metal matrix. Divide the acoustic impedance value of the diamond by the acoustic impedance value of the metal matrix to obtain the acoustic impedance ratio correction factor value. Based on the acoustic impedance ratio correction factor, the amplitude adjustment coefficient and nonlinear correction coefficient in the diamond characteristic compensation transformation function are determined, thus obtaining the complete parameters of the diamond characteristic compensation transformation function.

[0010] Optionally, the diamond composite sheet welding interface feature enhancement and recognition algorithm performs spatial registration and data fusion on the ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set to obtain a fused defect distribution map and a multimodal defect feature matrix, including: Based on the reference points marked on the test sample, the spatial coordinates of the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set are transformed by the heterogeneous data space mapping function in the diamond composite sheet welding interface feature enhancement and recognition algorithm to obtain three-mode defect data in a unified coordinate system. The rigid registration module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the three-modal defect data under the unified coordinate system. The spatial transformation matrix is ​​calculated based on the registration control point selection strategy corrected by acoustic impedance characteristics to obtain preliminary registration data. The preliminary registration data is input into the elastic registration fine-tuning module of the diamond composite sheet welding interface feature enhancement and recognition algorithm. Deformation compensation is performed using the B-spline deformation field description function constrained by diamond material properties to obtain accurate registration data. The multimodal evidence theory fusion module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the precise registration data. Based on the diamond-metal interface characteristics, a confidence weight is assigned to the detection results of different modes to obtain a fused defect distribution map. The ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set in the precise registration data are combined and the feature extractor in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to construct a multi-dimensional feature space containing acoustic, density, and thermal information to obtain the original feature matrix. The feature selection module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original feature matrix. The contribution of each feature is calculated by the diamond welding defect sensitivity quantification function, and key features are selected based on the importance ranking of random forest to obtain the multimodal defect feature matrix.

[0011] Optionally, the ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set from the precise registration data are combined and the feature extractor in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to construct a multi-dimensional feature space containing acoustic, density, and thermal information, resulting in an original feature matrix, including: Acoustic feature parameters, including maximum reflection intensity, average reflection intensity, reflection waveform width, phase change degree, and edge sharpness, are extracted from the ultrasonic defect feature vector set to obtain an acoustic feature subset; Density feature parameters, including defect area, defect perimeter, equivalent diameter, circularity coefficient, and density gradient value, are extracted from the X-ray defect feature vector set to obtain a density feature subset. Thermal feature parameters, including thermal time constant, maximum temperature difference, thermal diffusivity, thermal response delay time, and thermal conduction uniformity index, are extracted from the thermal imaging defect feature vector set to obtain a subset of thermal features. The acoustic feature subset, density feature subset, and thermal feature subset are concatenated to form a joint feature vector. Each potential defect region corresponds to a joint feature vector, resulting in a multidimensional feature vector set. The feature normalization module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the multidimensional feature vector set to standardize features of different dimensions, thereby obtaining a standardized feature vector set. The standardized feature vector set is integrated into a feature matrix structure, where each row represents a potential defect region and each column represents a feature parameter. Redundant features are eliminated through feature cross-correlation analysis to obtain the original feature matrix.

[0012] Optionally, the step of calculating the quantitative evaluation index of the weld interface density based on the multimodal defect feature matrix to obtain the weld quality grade evaluation result includes: Key feature parameters are extracted from the multimodal defect feature matrix, and the ratio of the total defect area to the detection area is calculated to obtain the defect area ratio as the first evaluation index of weld tightness. The defect depth, area, and signal intensity data in the multimodal defect feature matrix are normalized, and different weighting coefficients are set according to the defect type. The severity of each defect is nonlinearly weighted and summed to obtain the defect severity index as the second evaluation index of weld tightness. Information entropy is calculated for the thermal field distribution data and acoustic field distribution data in the multimodal defect feature matrix. Interface uniformity is evaluated by calculating the entropy difference and gradient distribution variance, and the welding interface uniformity index is obtained as the third evaluation index of welding density. Based on the acoustic reflection characteristics and X-ray density gradient data in the multimodal defect feature matrix, a support vector regression model is constructed, and the training interface is combined with the strength prediction function to output the predicted bonding strength value as the fourth evaluation index of weld tightness. A fuzzy membership function set is constructed, and the first, second, third, and fourth evaluation indicators of welding density are converted into fuzzy membership values. A variable weight adaptive algorithm is applied to dynamically adjust the weights of each indicator, and a comprehensive welding quality score is obtained by weighted summation. Cluster analysis is performed on historical inspection data to determine the optimal classification threshold. Based on the comprehensive welding quality score, the welding quality is divided into four levels to obtain the welding quality level evaluation result.

[0013] Secondly, this application provides a detection system for the weld tightness of diamond composite sheets, the detection system for the weld tightness of diamond composite sheets comprising: The acquisition module is used to acquire multimodal data from the diamond composite welded sample to be tested, and obtain the original ultrasonic dataset, the original X-ray dataset, and the original thermal imaging dataset. The input module is used to input the original ultrasonic dataset, the original X-ray dataset, and the original thermal imaging dataset into the diamond composite sheet welding interface feature enhancement and recognition algorithm for processing, so as to obtain the ultrasonic defect feature vector set, the X-ray defect feature vector set, and the thermal imaging defect feature vector set; The fusion module is used to perform spatial registration and data fusion on the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set based on the diamond composite sheet welding interface feature enhancement and recognition algorithm, so as to obtain a fused defect distribution map and a multimodal defect feature matrix. The calculation module is used to calculate the quantitative evaluation index of the density of the welding interface based on the multimodal defect feature matrix, and obtain the welding quality grade evaluation result.

[0014] Thirdly, a testing device for the weld tightness of diamond composite sheets is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the testing device for the weld tightness of diamond composite sheets to perform the aforementioned testing method for the weld tightness of diamond composite sheets.

[0015] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the above-described method for detecting the weld tightness of diamond composite sheets.

[0016] The technical solution provided in this application acquires raw ultrasonic datasets, raw X-ray datasets, and raw thermal imaging datasets through multimodal data acquisition technology. This overcomes the limitation of limited information from single detection methods, enabling comprehensive and multi-angle detection of the weld interface and improving defect detection rate and accuracy. The three raw datasets are then input into a diamond composite weld interface feature enhancement and recognition algorithm for processing. This algorithm is specifically optimized for the unique physical properties of diamond composites, such as high acoustic impedance, high density, and high thermal conductivity. Through specific algorithm modules such as acoustic impedance ratio correction factors, X-ray attenuation correction models, and anisotropic thermal diffusion models, it effectively enhances the feature representation of the diamond-metal interface, solving the problem of poor signal quality in traditional algorithms when processing diamond composites. Based on a diamond composite sheet welding interface feature enhancement and recognition algorithm, spatial registration and data fusion are performed on ultrasonic defect feature vector sets, X-ray defect feature vector sets, and thermal imaging defect feature vector sets. Through heterogeneous data spatial mapping functions, rigid registration, and elastic registration fine-tuning techniques, precise spatial alignment of different modal data is achieved, overcoming the technical obstacle of multimodal data fusion in traditional methods. This results in a more comprehensive fused defect distribution map and multimodal defect feature matrix. Based on the multimodal defect feature matrix, quantitative evaluation indicators of weld interface compactness are calculated, establishing a multi-index evaluation system including defect area ratio, defect severity index, weld interface uniformity index, and interface bonding strength index. Using artificial intelligence algorithms such as fuzzy theory and cluster analysis, the welding quality level is objectively and quantitatively assessed, solving the problems of strong subjectivity and inconsistent standards in traditional evaluation methods.

[0017] The diamond composite sheet welding interface feature enhancement and recognition algorithm in this invention takes into account the physical properties of diamond composite materials. Through specific functional modules such as acoustic impedance ratio correction, nonlinear transformation, and phase reconstruction, it realizes accurate processing and feature enhancement of diamond-metal interface signals. In the multimodal data fusion stage, the algorithm overcomes the spatial inconsistency of data obtained from different physical principles through heterogeneous data mapping and multi-level registration strategies. In the evaluation stage, the algorithm accurately captures key feature parameters related to weld compactness through feature extraction and screening mechanisms. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of an embodiment of the method for detecting the weld tightness of diamond composite sheets in this application; Figure 2 This is a schematic diagram of an embodiment of the detection system for the weld tightness of diamond composite sheets in this application. Figure 3 This is a schematic block diagram of the structure of the testing equipment for the welding density of diamond composite sheets in an embodiment of the present invention. Detailed Implementation

[0020] This application provides a method and system for detecting the weld tightness of diamond composite sheets. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the method for detecting the weld tightness of diamond composite sheets in this application includes: Step S101: Perform multimodal data acquisition on the diamond composite welded sample to be tested to obtain the original ultrasonic dataset, the original X-ray dataset, and the original thermal imaging dataset. Step S102: Input the original ultrasonic dataset, original X-ray dataset, and original thermal imaging dataset into the diamond composite sheet welding interface feature enhancement and recognition algorithm for processing to obtain ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set; Step S103: Based on the diamond composite sheet welding interface feature enhancement and recognition algorithm, spatial registration and data fusion are performed on the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set to obtain a fused defect distribution map and a multimodal defect feature matrix. Step S104: Calculate the quantitative evaluation index of the density of the welding interface based on the multimodal defect feature matrix to obtain the welding quality grade evaluation result.

[0022] It is understood that the executing entity of this application can be a testing system for the weld tightness of diamond composite sheets, or it can be a terminal or a server; no specific limitation is made here. This application's embodiments use a server as an example for illustration.

[0023] Specifically, a multi-dimensional detection platform was used to acquire multimodal data from the welded diamond composite sheet samples. The sample was placed on the platform, its surface was cleaned, and detection parameters were set. Reference points were marked at the four corners and center of the sample. Then, an ultrasonic detection unit was used for C-scanning, an X-ray fluoroscopy unit for 360° rotational scanning, and a thermal imaging analysis unit for thermal image sequence acquisition. Finally, the original ultrasonic dataset, original X-ray dataset, and original thermal imaging dataset were obtained.

[0024] The collected raw datasets are input into a diamond composite sheet welding interface feature enhancement and recognition algorithm for processing. First, the raw ultrasonic data undergoes bandpass filtering and waveform denoising. Then, the acoustic impedance ratio correction factor of the diamond-metal interface is calculated, and a diamond characteristic compensation transformation function is constructed for nonlinear transformation processing. Next, the acoustic reflection intensity distribution map of the welding interface is extracted through high-frequency feature extraction and Hilbert transform. Finally, an ultrasonic defect feature vector set is obtained through adaptive threshold segmentation. Simultaneously, the raw X-ray data undergoes a nonlinear transformation using an X-ray attenuation correction model, and the raw thermal imaging data is processed using an anisotropic thermal diffusion model to calculate the corrected temperature field, yielding X-ray defect feature vector sets and thermal imaging defect feature vector sets, respectively.

[0025] Based on the diamond composite sheet welding interface feature enhancement recognition algorithm, spatial registration and data fusion are performed on the three defect feature vector sets mentioned above. The defect data of the three modalities are transformed to a unified coordinate system through a heterogeneous data space mapping function, and then rigid registration and elastic registration fine-tuning are performed to obtain accurately registered data. Next, a multimodal evidence theory fusion module is applied to assign confidence weights to the detection results of different modalities, generating a fused defect distribution map. Simultaneously, the three defect feature vector sets in the accurately registered data are combined to construct a multidimensional feature space containing acoustic, density, and thermal information. Finally, a multimodal defect feature matrix is ​​obtained through feature filtering. Based on the multimodal defect feature matrix, a quantitative evaluation index of the weld interface density is calculated to obtain the weld quality grade assessment result. Specifically, by calculating indicators such as defect area ratio, defect severity index, weld interface uniformity index, and interface bonding strength index, a fuzzy membership function set is constructed to convert these indicators into fuzzy membership values. A variable weight adaptive algorithm is used to dynamically adjust the weights of each indicator. Finally, based on the optimal classification threshold determined by cluster analysis, the weld quality is divided into four levels: excellent, good, qualified, and unqualified.

[0026] In one specific embodiment, the process of performing step S101 may specifically include the following steps: The diamond composite sheet welding sample was placed on a multidimensional detection platform, and the sample surface was cleaned to obtain a cleaned sample. The ultrasonic testing parameters, X-ray fluoroscopy parameters, and thermal imaging analysis parameters are set for the cleaned sample to obtain a set of testing parameters. The sample is marked according to the set of detection parameters, and reference points are marked at the four corners and the center of the sample to obtain a test sample with reference points; The ultrasonic testing unit is used to perform a C-scan on the test sample with reference points, and a zigzag scanning path is used to cover the entire welding area to obtain the original ultrasonic dataset. The test sample with the reference point is placed in the X-ray fluoroscopy unit and rotated 360° for scanning. A projection image is acquired at 0.5° intervals to obtain the original X-ray dataset. The test sample with reference points is placed in the thermal imaging analysis unit, and the thermal excitation source is used to uniformly irradiate the welding area. Thermal image sequences are continuously acquired during and after the excitation process to obtain the original thermal imaging dataset.

[0027] Specifically, the diamond composite sheet welded sample is placed on a multidimensional detection platform, which is a comprehensive testing device integrating an ultrasonic testing unit, an X-ray fluoroscopy unit, and a thermal imaging analysis unit. The sample surface is cleaned with anhydrous ethanol to remove surface contaminants, and then allowed to stabilize at room temperature to obtain a cleaned sample. Testing parameters are set for the cleaned sample according to the specific specifications of the diamond composite sheet. Ultrasonic testing parameters include detection frequency, pulse width, gain, and scan step; X-ray fluoroscopy parameters include tube voltage, tube current, exposure time, and detector resolution; thermal imaging analysis parameters include thermal excitation source power, excitation time, acquisition frame rate, and total acquisition time. These parameters combine to form a testing parameter set, ensuring the standardization and repeatability of the testing process. The sample is marked according to the set testing parameters, with reference points marked at the four corners and center. These reference points are crucial for subsequent multimodal data registration and are typically marked with 0.5 mm diameter silver paint dots to ensure clear identification in all three testing modes. After marking, a test sample with reference points is obtained.

[0028] An ultrasonic testing unit was used to perform a C-scan on the sample with reference points. C-scanning is a two-dimensional surface scanning method that uses a zigzag scanning path to cover the entire welding area. The probe is perpendicular to the sample surface, maintaining a uniform coupling agent thickness. After the ultrasonic waves enter the sample, they are reflected at the material interface. The reflected signals are recorded to form a raw ultrasonic dataset, containing reflected wave amplitude, phase information, and time information. The sample with reference points was then placed in an X-ray imaging unit for a 360° rotation scan. During the scan, the X-ray source remained fixed, and the sample rotated 1 / 2 revolution at 0.5° intervals. One projection image was acquired at each angular position, for a total of 720 projection images. These projection images recorded the attenuation of X-rays at different angles, forming a raw X-ray dataset.

[0029] The sample with reference points is placed in the thermal imaging analysis unit, ensuring the thermal excitation source uniformly illuminates the welding area. Thermal imaging detection is based on the differences in the thermal conductivity of materials. When the thermal excitation source irradiates the sample, heat propagates within it. The thermal conductivity of the welding defect area differs from that of the normal area, resulting in differences in surface temperature distribution. Thermal image sequences are continuously acquired during and after the excitation process to record the changes in sample surface temperature over time, forming the original thermal imaging dataset.

[0030] In one specific embodiment, the process of performing step S102 may specifically include the following steps: The original ultrasonic dataset is subjected to bandpass filtering and waveform denoising to obtain preprocessed ultrasonic data; The acoustic impedance ratio correction factor of the diamond-metal interface is calculated based on the preprocessed ultrasonic data. The acoustic impedance ratio correction factor is determined by the ratio of the density and sound velocity of diamond to the density and sound velocity of the metal matrix, and the acoustic impedance ratio correction factor value is obtained. Based on the acoustic impedance ratio correction factor, a diamond characteristic compensation transformation function is constructed. The preprocessed ultrasonic data is then subjected to nonlinear transformation processing through the diamond characteristic compensation transformation function to obtain the compensated ultrasonic signal. The compensated ultrasonic signal is subjected to high-frequency feature extraction and Hilbert transform to extract the acoustic reflection intensity distribution map of the welding interface. The ultrasonic defect feature vector set is obtained by adaptive threshold segmentation. The X-ray attenuation correction model in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original X-ray dataset. Nonlinear transformation is performed through the density correction function. Three-dimensional reconstruction and nonlinear phase reconstruction are performed on the transformed image to obtain the X-ray defect feature vector set. An anisotropic thermal diffusion model from the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original thermal imaging dataset to calculate the corrected temperature field. Time-domain analysis and principal component analysis are then performed on the corrected thermal imaging sequence to obtain the thermal imaging defect feature vector set.

[0031] Specifically, bandpass filtering and waveform denoising are performed on the original ultrasonic dataset. Bandpass filtering is a signal processing technique that sets a frequency range of 15-25MHz based on the acoustic conduction characteristics of diamond composite materials, preserving the effective signal while filtering out environmental and system noise. Waveform denoising employs wavelet thresholding to reduce noise in the ultrasonic A-scan signal, resulting in preprocessed ultrasonic data.

[0032] After preprocessing, the acoustic impedance ratio correction factor (AIRR) of the diamond-metal interface is calculated. The AIRR is a crucial parameter for sound wave propagation between different media. By obtaining the density (approximately 3520 kg / m³) and sound velocity (approximately 18000 m / s) of the diamond material and measuring the density and sound velocity of the metal matrix, the acoustic impedance values ​​(the product of density and sound velocity) of both materials are calculated. The diamond AIRR is then divided by the metal matrix AIRR to obtain the AIRR value. This factor describes the transmission characteristics of ultrasound at the diamond-metal interface. Based on the AIRR value, a diamond characteristic compensation transformation function is constructed, with the form F(x) = α·x·exp(β·x²), where x is the original signal amplitude, α is the amplitude adjustment coefficient, and β is the nonlinear correction coefficient. These two coefficients are dynamically determined based on the AIRR. The preprocessed ultrasonic data is then subjected to a nonlinear transformation using this compensation transformation function to compensate for signal distortion caused by the high acoustic impedance difference at the diamond-metal interface, resulting in a compensated ultrasonic signal.

[0033] High-frequency feature extraction and Hilbert transform were performed on the compensated ultrasonic signal. High-frequency feature extraction involved bandpass filtering optimized for diamond composite materials in the 18-25MHz frequency range, focusing on extracting the unique acoustic response characteristics of the diamond composite material. Hilbert transform, a signal processing method, was used to extract the signal envelope, obtaining the acoustic reflection intensity distribution map of the weld interface. Adaptive threshold segmentation was applied, with the threshold set to 2.5 times the average intensity of the background region. Regions with intensity greater than the threshold were marked as potential defect areas, and their geometric and signal features were extracted to form an ultrasonic defect feature vector set.

[0034] An X-ray attenuation correction model was applied to the original X-ray dataset. This model addresses the impact of high diamond density on X-ray imaging through a density correction function G(I) = A nonlinear transformation is performed, where I is the image grayscale value (an integer ranging from 0 to 255, representing the brightness value of each pixel in the image), and k is the density correction exponent (typically 1.3-1.7, determined based on the density range of the diamond composite sheet). This function enhances the contrast of low-density areas while reducing the saturation of high-density areas through nonlinear transformation. The transformed image is then subjected to 3D reconstruction (using the FDK algorithm, i.e., the Feldkamp-Davis-Kress algorithm, a commonly used cone-beam CT 3D reconstruction algorithm) and nonlinear phase reconstruction processing to enhance the edge details of the diamond-metal interface. Finally, welding defect areas are identified, their geometric and density features are extracted, and an X-ray defect feature vector set is obtained.

[0035] An anisotropic thermal diffusion model was applied to the original thermal imaging dataset. This model considers the high thermal conductivity and anisotropic characteristics of diamond composite materials, calculating the corrected temperature field T'(x,y,t) = TB(x,y,t) × H(x,y), where T'(x,y,t) is the corrected temperature field (in °C, representing the temperature value at time t at spatial coordinates (x,y), TB(x,y,t) is the original acquired temperature field data (in °C), and H(x,y) is the thermal conduction correction function (dimensionless, a coefficient used to correct thermal conduction characteristics at different locations). This correction function considers the influence of the non-uniformity of diamond particle distribution on thermal conduction. Temporal analysis was performed on the corrected thermal imaging sequence to calculate the heating and cooling curves for each pixel, as well as principal component analysis (a dimensionality reduction technique that decomposes the thermal image sequence into a time function and a spatial distribution function) to extract feature parameters of areas with abnormal thermal conduction, resulting in a set of thermal imaging defect feature vectors.

[0036] In one specific embodiment, the process of calculating the acoustic impedance ratio correction factor of the diamond-metal interface on the preprocessed ultrasonic data may specifically include the following steps: Obtain the density and sound velocity values ​​of diamond material to obtain the basic parameters of diamond; The density and sound velocity of the metal matrix are measured to obtain the basic parameters of the metal matrix; The acoustic impedance of diamond is calculated by multiplying the density value and the sound velocity value in the basic diamond parameters. The acoustic impedance of the metal matrix is ​​calculated by multiplying the density value and the sound velocity value in the basic parameters of the metal matrix. Divide the acoustic impedance value of the diamond by the acoustic impedance value of the metal matrix to obtain the acoustic impedance ratio correction factor value. Based on the acoustic impedance ratio correction factor, the amplitude adjustment coefficient and nonlinear correction coefficient in the diamond characteristic compensation transformation function are determined, thus obtaining the complete parameters of the diamond characteristic compensation transformation function.

[0037] Specifically, the density and sound velocity of diamond materials are obtained through material database queries or experimental measurements. Diamond typically has a density of 3520 kg / m³ and a sound velocity of 18000 m / s; these two parameters together constitute the fundamental parameters of diamond. These fundamental parameters reflect the basic acoustic properties of diamond materials and are the basis for calculating acoustic impedance. The density and sound velocity of the metal matrix are measured using direct measurement methods. The density of the metal matrix is ​​measured using Archimedes' displacement method, where the mass difference of the sample in air and water is measured using a precision balance to calculate the density value. The sound velocity is measured using the transmission method, where a high-frequency ultrasonic probe emits ultrasonic waves, and the time it takes for the waves to travel through the sample is recorded. The sound velocity is then calculated by combining this with the sample thickness. A common cemented carbide matrix has a density of 14500 kg / m³ and a sound velocity of 6100 m / s; these two parameters constitute the fundamental parameters of the metal matrix.

[0038] The density value and the sound velocity value from the basic diamond parameters are then multiplied to calculate the acoustic impedance of the diamond. Acoustic impedance is a physical quantity describing the resistance to sound wave propagation in a material, with units of kg / (m²·s). For diamond, the acoustic impedance value equals the density multiplied by the sound velocity, i.e., 3520 kg / m³ × 18000 m / s, yielding 63.36 × 10⁻⁶. 6 kg / (m²·s). This value represents the resistance to sound wave propagation per unit area and is an important parameter for measuring the ratio of sound wave reflection to transmission at an interface.

[0039] Similarly, the acoustic impedance of the metal matrix is ​​calculated by multiplying the density value and the sound velocity value in the basic parameters of the metal matrix. For a cemented carbide matrix, the acoustic impedance is equal to 14500 kg / m³ × 6100 m / s, resulting in 88.45 × 10⁻⁶ m / s. 6 kg / (m²·s). The acoustic impedance of the metal matrix directly affects the reflection coefficient of ultrasound at the diamond-metal interface and is a key indicator for evaluating interface characteristics. Dividing the diamond acoustic impedance by the metal matrix acoustic impedance yields the acoustic impedance ratio correction factor. This value describes the relative relationship between the acoustic impedances of the two materials. For diamond and cemented carbide matrices, the acoustic impedance ratio correction factor is approximately 63.36 ÷ 88.45 ≈ 0.716. The acoustic impedance ratio correction factor is a quantitative expression of the propagation characteristics of ultrasound at the diamond-metal interface and directly affects subsequent signal processing strategies.

[0040] The amplitude adjustment coefficient and nonlinear correction coefficient in the diamond characteristic compensation transformation function are determined based on the acoustic impedance ratio correction factor. When the acoustic impedance ratio correction factor is in the range of 0.7-0.8, the amplitude adjustment coefficient is set to 0.9 and the nonlinear correction coefficient to 0.08. If the acoustic impedance ratio correction factor is less than 0.7, the amplitude adjustment coefficient is increased to 1.1 and the nonlinear correction coefficient is decreased to 0.05. If the acoustic impedance ratio correction factor is greater than 0.8, the amplitude adjustment coefficient is decreased to 0.8 and the nonlinear correction coefficient is increased to 0.12. In this way, the compensation function parameters are dynamically adjusted for different acoustic impedance ratios, ensuring accurate compensation for signal distortion caused by differences in the acoustic properties of materials in the inspection of diamond composite weld interfaces. This solves the technical problem of poor signal quality in traditional ultrasonic testing methods when dealing with interfaces of materials with high acoustic impedance differences.

[0041] In one specific embodiment, the process of executing step S103 may specifically include the following steps: Based on the reference points marked on the test sample, the spatial coordinates of the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set are transformed by the heterogeneous data space mapping function in the diamond composite sheet welding interface feature enhancement and recognition algorithm to obtain three-mode defect data in a unified coordinate system. The rigid registration module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the three-modal defect data under the unified coordinate system. The spatial transformation matrix is ​​calculated based on the registration control point selection strategy corrected by acoustic impedance characteristics to obtain preliminary registration data. The preliminary registration data is input into the elastic registration fine-tuning module of the diamond composite sheet welding interface feature enhancement and recognition algorithm. Deformation compensation is performed using the B-spline deformation field description function constrained by diamond material properties to obtain accurate registration data. The multimodal evidence theory fusion module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the precise registration data. Based on the diamond-metal interface characteristics, a confidence weight is assigned to the detection results of different modes to obtain a fused defect distribution map. The ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set in the precise registration data are combined and the feature extractor in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to construct a multi-dimensional feature space containing acoustic, density, and thermal information to obtain the original feature matrix. The feature selection module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original feature matrix. The contribution of each feature is calculated by the diamond welding defect sensitivity quantification function, and key features are selected based on the importance ranking of random forest to obtain the multimodal defect feature matrix.

[0042] Specifically, based on the reference points marked on the test sample, the spatial coordinates of the ultrasonic defect feature vector sets, X-ray defect feature vector sets, and thermal imaging defect feature vector sets are transformed using a heterogeneous data spatial mapping function. The heterogeneous data spatial mapping function is a coordinate transformation algorithm that converts data acquired under different modes into the same spatial coordinate system. Specifically, it establishes coordinate mapping relationships between different modes through the correspondence of reference points, transforms the positional information in each defect feature vector, and obtains three-mode defect data in a unified coordinate system. Rigid registration is then performed on the three-mode defect data in the unified coordinate system. Rigid registration is a registration method that preserves the shape of the object, allowing only rigid body transformations such as translation and rotation. This step uses a registration control point selection strategy based on acoustic impedance characteristic correction, selecting the region with the largest acoustic impedance gradient as the control point, and calculating the spatial transformation matrix. The spatial transformation matrix describes the mapping relationship from one coordinate system to another, including rotation and translation components. This matrix is ​​used to transform the three-mode data, obtaining preliminary registration data.

[0043] The preliminary registration data is input into the elastic registration fine-tuning module for processing. Elastic registration allows for local deformation of the object, compensating for subtle deformations that may occur during different modal detection processes. This step uses a B-spline deformation field description function constrained by diamond material properties for deformation compensation. A B-spline is a mathematical curve defined by control points and basis functions, which can smoothly describe the local deformation of an object. By calculating the local deformation field, the preliminary registration data is fine-tuned to obtain accurate registration data. The accurate registration data is then processed using the multimodal evidence theory fusion module. Evidence theory is a mathematical tool for handling uncertain information and is suitable for multi-source information fusion. This step assigns confidence weights to different modal detection results based on the diamond-metal interface characteristics. Ultrasonic detection is highly sensitive to cracks and is assigned a higher weight; X-ray detection is highly sensitive to porosity and is also assigned a higher weight; thermal imaging is sensitive to poor interface bonding and is also assigned a higher weight. The fusion result is calculated using the combination rules of evidence theory to obtain a fused defect distribution map.

[0044] The three defect feature vector sets from the precisely registered data are combined and processed using a feature extractor. The feature extractor is an algorithm specifically designed for the welding characteristics of diamond composite sheets, designed to extract useful information from raw data. This step extracts parameters such as maximum reflection intensity and average reflection intensity from acoustic features; parameters such as defect area and defect perimeter from density features; and parameters such as thermal time constant and maximum temperature difference from thermal features. All features are combined to form a multi-dimensional feature space, with each defect region corresponding to a feature vector, and all vectors constitute the original feature matrix.

[0045] Finally, the original feature matrix is ​​processed by the feature selection module. This step calculates the contribution of each feature using a diamond welding defect sensitivity quantification function to evaluate the correlation between features and defect types. Random forest importance ranking is a machine learning method that evaluates feature importance by constructing multiple decision trees and statistically analyzing feature usage frequency. Based on the feature importance ranking, features with high contributions are selected, ultimately yielding the multimodal defect feature matrix.

[0046] In one specific embodiment, the process of constructing a multidimensional feature space containing acoustic, density, and thermal information may specifically include the following steps: Acoustic feature parameters, including maximum reflection intensity, average reflection intensity, reflection waveform width, phase change degree, and edge sharpness, are extracted from the ultrasonic defect feature vector set to obtain an acoustic feature subset; Density feature parameters, including defect area, defect perimeter, equivalent diameter, circularity coefficient, and density gradient value, are extracted from the X-ray defect feature vector set to obtain a density feature subset. Thermal feature parameters, including thermal time constant, maximum temperature difference, thermal diffusivity, thermal response delay time, and thermal conduction uniformity index, are extracted from the thermal imaging defect feature vector set to obtain a subset of thermal features. The acoustic feature subset, density feature subset, and thermal feature subset are concatenated to form a joint feature vector. Each potential defect region corresponds to a joint feature vector, resulting in a multidimensional feature vector set. The feature normalization module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the multidimensional feature vector set to standardize features of different dimensions, thereby obtaining a standardized feature vector set. The standardized feature vector set is integrated into a feature matrix structure, where each row represents a potential defect region and each column represents a feature parameter. Redundant features are eliminated through feature cross-correlation analysis to obtain the original feature matrix.

[0047] Specifically, acoustic feature parameters are extracted from the ultrasonic defect feature vector set. Maximum reflection intensity refers to the maximum amplitude of the ultrasonic reflected signal in the defect area, obtained by finding the peak value of the signal amplitude; average reflection intensity is the average amplitude of all ultrasonic reflected signals in the defect area, obtained by calculating the arithmetic mean of all signal amplitudes in the area; reflected waveform width represents the time span of the ultrasonic echo signal, characterized by the time interval during which the signal amplitude exceeds a threshold; phase abruptness measures the severity of phase change, obtained by calculating the standard deviation of the phase difference between adjacent sampling points; edge sharpness describes the clarity of the defect boundary, quantified by calculating the average gradient at the defect boundary. These five parameters together constitute an acoustic feature subset, comprehensively describing the acoustic characteristics of the weld interface defect. Density feature parameters are extracted from the X-ray defect feature vector set. The defect area is calculated by multiplying the number of pixels contained in the defect area by the actual area of ​​each pixel; the defect perimeter is calculated as the total length of the defect area boundary; the equivalent diameter is the diameter of the defect area when it is equivalent to a circle, calculated using the formula 4 × area / perimeter; the circularity coefficient measures how close the defect shape is to a circle, calculated using 4π × area / perimeter², with a value closer to 1 indicating a more circular shape; the density gradient value describes the steepness of the density change at the defect boundary, quantified by calculating the gradient of grayscale values ​​at the defect boundary. These five parameters constitute a subset of density features, describing the characteristics of welding defects from both geometric and density distribution perspectives.

[0048] Subsequently, thermal feature parameters were extracted from the thermal imaging defect feature vector set. The thermal time constant, a parameter describing the speed of the material's thermal response, is obtained by fitting heating or cooling curves; the maximum temperature difference is the maximum temperature difference between the defect area and the surrounding normal area; the thermal diffusivity characterizes the speed of heat propagation in the material, calculated by analyzing the temperature field changes over time; the thermal response delay time is the time that the temperature change in the defect area lags behind that of the normal area; and the thermal conduction uniformity index measures the uniformity of heat conduction, quantified by calculating the entropy value of the thermal image. These five parameters constitute a subset of thermal features, describing the characteristics of the weld interface from a thermodynamic perspective.

[0049] The acoustic, density, and thermal feature subsets are concatenated, which involves sequentially merging the features from the three subsets into a single, longer feature vector. This feature concatenation operation integrates information acquired from different modes to form a joint feature vector. Each potential defect region corresponds to a joint feature vector containing 15 feature parameters, and all joint feature vectors form a multidimensional feature vector set.

[0050] Feature normalization is applied to the multidimensional feature vector set. Normalization is the process of transforming features of different dimensions to the same scale. Commonly used methods include min-max normalization and z-score normalization. Min-max normalization maps feature values ​​to the [0,1] interval, while z-score normalization transforms features into a distribution with a mean of 0 and a standard deviation of 1. For the weld tightness detection of diamond composite sheets, the z-score normalization method is used. Each feature parameter is subtracted from its mean and then divided by its standard deviation to obtain a standardized feature vector set. The standardized feature vector set is integrated into a feature matrix structure, forming a matrix with the number of rows equal to the number of defect regions and the number of columns equal to the number of feature parameters. Highly correlated feature pairs are identified through feature cross-correlation analysis. The correlation coefficient between each pair of features is calculated. When the absolute value of the correlation coefficient is greater than 0.85, features with lower importance are deleted, thereby eliminating redundant features and obtaining a lower-dimensional original feature matrix.

[0051] In one specific embodiment, the process of executing step S104 may specifically include the following steps: Key feature parameters are extracted from the multimodal defect feature matrix, and the ratio of the total defect area to the detection area is calculated to obtain the defect area ratio as the first evaluation index of weld tightness. The defect depth, area, and signal intensity data in the multimodal defect feature matrix are normalized, and different weighting coefficients are set according to the defect type. The severity of each defect is nonlinearly weighted and summed to obtain the defect severity index as the second evaluation index of weld tightness. Information entropy is calculated for the thermal field distribution data and acoustic field distribution data in the multimodal defect feature matrix. Interface uniformity is evaluated by calculating the entropy difference and gradient distribution variance, and the welding interface uniformity index is obtained as the third evaluation index of welding density. Based on the acoustic reflection characteristics and X-ray density gradient data in the multimodal defect feature matrix, a support vector regression model is constructed, and the training interface is combined with the strength prediction function to output the predicted bonding strength value as the fourth evaluation index of weld tightness. A fuzzy membership function set is constructed, and the first, second, third, and fourth evaluation indicators of welding density are converted into fuzzy membership values. A variable weight adaptive algorithm is applied to dynamically adjust the weights of each indicator, and a comprehensive welding quality score is obtained by weighted summation. Cluster analysis is performed on historical inspection data to determine the optimal classification threshold. Based on the comprehensive welding quality score, the welding quality is divided into four levels to obtain the welding quality level evaluation result.

[0052] Specifically, key feature parameters are extracted from the multimodal defect feature matrix to calculate the defect area ratio. The defect area ratio is the ratio of the total defect area to the area of ​​the detection region. It is calculated by statistically analyzing the number of pixels in all regions marked as defects in the fused defect distribution map and dividing by the total number of pixels in the entire detection region. The defect area ratio directly reflects the distribution range of defects at the weld interface and serves as the primary evaluation indicator of weld tightness. The defect depth, area, and signal intensity data in the multimodal defect feature matrix are normalized. Normalization transforms features of different dimensions to a unified scale range. The min-max normalization method is used, subtracting the minimum value of each feature value and dividing by the difference between the maximum and minimum values. After normalization, different weighting coefficients are set according to the defect type: 0.8 for porosity defects, 1.0 for crack defects, and 0.9 for poor bonding defects. The severity of each defect is nonlinearly weighted and summed, and an exponential weighting function is used to calculate the defect severity index. This index comprehensively considers the geometric and physical characteristics of the defects and serves as the secondary evaluation indicator of weld tightness.

[0053] Information entropy is calculated for the thermal and acoustic field distribution data in the multimodal defect feature matrix. Information entropy is a physical quantity characterizing the degree of disorder in a system, obtained by calculating the probability density function of the data distribution and obtaining the entropy value. Weld interface uniformity is assessed by calculating the entropy difference and gradient distribution variance. The entropy difference refers to the maximum difference in entropy values ​​between different regions, and the gradient distribution variance refers to the statistical variance of the gradient amplitude. These two parameters reflect the uniformity of heat conduction and acoustic propagation at the weld interface, respectively. A weighted average is used to obtain the weld interface uniformity index, which serves as the third evaluation index for weld tightness. Based on the acoustic reflection characteristics and X-ray density gradient data in the multimodal defect feature matrix, a support vector regression model is constructed. Support vector regression is a machine learning algorithm that finds the best hyperplane to fit the data distribution. During model training, samples with known bonding strength are used as the training set, acoustic reflection characteristics and X-ray density gradient are used as input features, interface bonding strength is used as the output label, radial basis function is used as the kernel function, and cross-validation is used to determine the optimal penalty and kernel parameters. After training, the model is used to predict new samples, and the predicted bonding strength value is output as the fourth evaluation index of weld tightness.

[0054] A fuzzy membership function set was constructed to convert the four evaluation indicators of weld density into fuzzy membership values. The fuzzy membership function, in fuzzy set theory, maps precise numerical values ​​to the [0,1] interval, representing the degree to which an element belongs to the set. A decreasing sigmoid function was used for the defect area ratio; a decreasing Gaussian function for the defect severity index; an increasing sigmoid function for the weld interface uniformity index; and an increasing trapezoidal function for the interface bonding strength index. After conversion, a variable-weight adaptive algorithm was applied to dynamically adjust the weights of each indicator. This algorithm automatically adjusts the weight allocation based on the magnitude of change in each indicator, increasing the weight of an indicator when it changes drastically. Finally, a weighted summation was used to obtain the comprehensive weld quality score.

[0055] Cluster analysis was performed on historical inspection data to determine the optimal classification threshold. Cluster analysis is an unsupervised learning method that divides data into different categories. The K-means clustering algorithm was used, with a set number of categories of 4, to cluster the comprehensive scores of historical samples, obtaining the center points and boundaries of four score intervals. Based on the boundary values, the classification threshold was determined, dividing the comprehensive welding quality score into four levels: excellent, good, acceptable, and unacceptable, thus obtaining the welding quality grade evaluation result.

[0056] The above describes the method for detecting the weld tightness of diamond composite sheets in the embodiments of this application. The following describes the system for detecting the weld tightness of diamond composite sheets in the embodiments of this application. Please refer to [link to relevant documentation]. Figure 2 One embodiment of the detection system for the weld tightness of diamond composite sheets in this application includes: The acquisition module 201 is used to acquire multimodal data of the diamond composite sheet welding sample to be tested, and obtain the original ultrasonic dataset, the original X-ray dataset and the original thermal imaging dataset. Input module 202 is used to input the original ultrasonic dataset, original X-ray dataset and original thermal imaging dataset into the diamond composite sheet welding interface feature enhancement and recognition algorithm for processing, so as to obtain ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set; The fusion module 203 is used to perform spatial registration and data fusion on the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set based on the diamond composite sheet welding interface feature enhancement and recognition algorithm, so as to obtain a fused defect distribution map and a multimodal defect feature matrix. The calculation module 204 is used to calculate the quantitative evaluation index of the density of the welding interface based on the multimodal defect feature matrix, and obtain the welding quality grade evaluation result.

[0057] above Figure 2The detection system for weld tightness of diamond composite sheets in this embodiment of the invention is described in detail from the perspective of modular functional entities. The detection equipment for weld tightness of diamond composite sheets in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0058] Figure 3 This is a schematic diagram of a structural device for detecting the weld tightness of diamond composite sheets according to an embodiment of the present invention. The device 300 can vary considerably depending on its configuration or performance. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the detection device 300 for the weld tightness of diamond composite sheets. Furthermore, the processor 310 may be configured to communicate with the storage media 330 and execute the series of instruction operations in the storage media 330 on the detection device 300 for the weld tightness of diamond composite sheets to implement the steps of the aforementioned detection method for the weld tightness of diamond composite sheets.

[0059] The testing device 300 for the weld tightness of diamond composite sheets may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The illustrated structure of the testing device for the weld tightness of diamond composite sheets does not constitute a limitation on the testing device for the weld tightness of diamond composite sheets provided by the present invention. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0060] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the method for detecting the weld tightness of diamond composite sheets.

[0061] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0062] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a testing device for the weld density of diamond composite sheets (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the weld tightness of diamond composite sheets, characterized in that, The method includes: Multimodal data acquisition was performed on the diamond composite welded sample to be tested to obtain the original ultrasonic dataset, the original X-ray dataset, and the original thermal imaging dataset. The original ultrasonic dataset, original X-ray dataset, and original thermal imaging dataset are input into the diamond composite sheet welding interface feature enhancement and recognition algorithm for processing to obtain ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set. Based on the diamond composite sheet welding interface feature enhancement and recognition algorithm, the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set are spatially registered and data fused to obtain a fused defect distribution map and a multimodal defect feature matrix. The density quantitative evaluation index of the weld interface is calculated based on the multimodal defect feature matrix to obtain the weld quality grade evaluation result.

2. The method for detecting the weld tightness of diamond composite sheets according to claim 1, characterized in that, The diamond composite welded sample to be tested undergoes multimodal data acquisition to obtain raw ultrasonic datasets, raw X-ray datasets, and raw thermal imaging datasets, including: The diamond composite sheet welding sample was placed on a multidimensional detection platform, and the sample surface was cleaned to obtain a cleaned sample. The ultrasonic testing parameters, X-ray fluoroscopy parameters, and thermal imaging analysis parameters are set for the cleaned sample to obtain a set of testing parameters. The sample is marked according to the set of detection parameters, and reference points are marked at the four corners and the center of the sample to obtain a test sample with reference points; The ultrasonic testing unit is used to perform a C-scan on the test sample with reference points, and a zigzag scanning path is used to cover the entire welding area to obtain the original ultrasonic dataset. The test sample with the reference point is placed in the X-ray fluoroscopy unit and rotated 360° for scanning. A projection image is acquired at 0.5° intervals to obtain the original X-ray dataset. The test sample with reference points is placed in the thermal imaging analysis unit, and the thermal excitation source is used to uniformly irradiate the welding area. Thermal image sequences are continuously acquired during and after the excitation process to obtain the original thermal imaging dataset.

3. The method for detecting the weld tightness of diamond composite sheets according to claim 1, characterized in that, The original ultrasonic dataset, original X-ray dataset, and original thermal imaging dataset are input into a diamond composite sheet welding interface feature enhancement and recognition algorithm for processing, resulting in ultrasonic defect feature vector sets, X-ray defect feature vector sets, and thermal imaging defect feature vector sets, including: The original ultrasonic dataset is subjected to bandpass filtering and waveform denoising to obtain preprocessed ultrasonic data; The acoustic impedance ratio correction factor of the diamond-metal interface is calculated based on the preprocessed ultrasonic data. The acoustic impedance ratio correction factor is determined by the ratio of the density and sound velocity of diamond to the density and sound velocity of the metal matrix, and the acoustic impedance ratio correction factor value is obtained. Based on the acoustic impedance ratio correction factor, a diamond characteristic compensation transformation function is constructed. The preprocessed ultrasonic data is then subjected to nonlinear transformation processing through the diamond characteristic compensation transformation function to obtain the compensated ultrasonic signal. The compensated ultrasonic signal is subjected to high-frequency feature extraction and Hilbert transform to extract the acoustic reflection intensity distribution map of the welding interface. The ultrasonic defect feature vector set is obtained by adaptive threshold segmentation. The X-ray attenuation correction model in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original X-ray dataset. Nonlinear transformation is performed through the density correction function. Three-dimensional reconstruction and nonlinear phase reconstruction are performed on the transformed image to obtain the X-ray defect feature vector set. An anisotropic thermal diffusion model from the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original thermal imaging dataset to calculate the corrected temperature field. Time-domain analysis and principal component analysis are then performed on the corrected thermal imaging sequence to obtain the thermal imaging defect feature vector set.

4. The method for detecting the weld tightness of diamond composite sheets according to claim 3, characterized in that, The acoustic impedance ratio correction factor for the diamond-metal interface is calculated based on the preprocessed ultrasonic data. This correction factor is determined by the ratio of the density and sound velocity of diamond to the density and sound velocity of the metal matrix. The numerical value of the acoustic impedance ratio correction factor includes: Obtain the density and sound velocity values ​​of diamond material to obtain the basic parameters of diamond; The density and sound velocity of the metal matrix are measured to obtain the basic parameters of the metal matrix; The acoustic impedance of diamond is calculated by multiplying the density value and the sound velocity value in the basic diamond parameters. The acoustic impedance of the metal matrix is ​​calculated by multiplying the density value and the sound velocity value in the basic parameters of the metal matrix. Divide the acoustic impedance value of the diamond by the acoustic impedance value of the metal matrix to obtain the acoustic impedance ratio correction factor value. Based on the acoustic impedance ratio correction factor, the amplitude adjustment coefficient and nonlinear correction coefficient in the diamond characteristic compensation transformation function are determined, thus obtaining the complete parameters of the diamond characteristic compensation transformation function.

5. The method for detecting the weld tightness of diamond composite sheets according to claim 1, characterized in that, The diamond composite sheet welding interface feature enhancement and recognition algorithm performs spatial registration and data fusion on the ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set to obtain a fused defect distribution map and a multimodal defect feature matrix, including: Based on the reference points marked on the test sample, the spatial coordinates of the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set are transformed by the heterogeneous data space mapping function in the diamond composite sheet welding interface feature enhancement and recognition algorithm to obtain three-mode defect data in a unified coordinate system. The rigid registration module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the three-modal defect data under the unified coordinate system. The spatial transformation matrix is ​​calculated based on the registration control point selection strategy corrected by acoustic impedance characteristics to obtain preliminary registration data. The preliminary registration data is input into the elastic registration fine-tuning module of the diamond composite sheet welding interface feature enhancement and recognition algorithm. Deformation compensation is performed using the B-spline deformation field description function constrained by diamond material properties to obtain accurate registration data. The multimodal evidence theory fusion module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the precise registration data. Based on the diamond-metal interface characteristics, a confidence weight is assigned to the detection results of different modes to obtain a fused defect distribution map. The ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set in the precise registration data are combined and the feature extractor in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to construct a multi-dimensional feature space containing acoustic, density, and thermal information to obtain the original feature matrix. The feature selection module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the original feature matrix. The contribution of each feature is calculated by the diamond welding defect sensitivity quantification function, and key features are selected based on the importance ranking of random forest to obtain the multimodal defect feature matrix.

6. The method for detecting the weld tightness of diamond composite sheets according to claim 5, characterized in that, The ultrasonic defect feature vector set, X-ray defect feature vector set, and thermal imaging defect feature vector set from the precisely registered data are combined and the feature extractor in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to construct a multi-dimensional feature space containing acoustic, density, and thermal information, resulting in the original feature matrix, including: Acoustic feature parameters, including maximum reflection intensity, average reflection intensity, reflection waveform width, phase change degree, and edge sharpness, are extracted from the ultrasonic defect feature vector set to obtain an acoustic feature subset; Density feature parameters, including defect area, defect perimeter, equivalent diameter, circularity coefficient, and density gradient value, are extracted from the X-ray defect feature vector set to obtain a density feature subset. Thermal feature parameters, including thermal time constant, maximum temperature difference, thermal diffusivity, thermal response delay time, and thermal conduction uniformity index, are extracted from the thermal imaging defect feature vector set to obtain a subset of thermal features. The acoustic feature subset, density feature subset, and thermal feature subset are concatenated to form a joint feature vector. Each potential defect region corresponds to a joint feature vector, resulting in a multidimensional feature vector set. The feature normalization module in the diamond composite sheet welding interface feature enhancement and recognition algorithm is applied to the multidimensional feature vector set to standardize features of different dimensions, thereby obtaining a standardized feature vector set. The standardized feature vector set is integrated into a feature matrix structure, where each row represents a potential defect region and each column represents a feature parameter. Redundant features are eliminated through feature cross-correlation analysis to obtain the original feature matrix.

7. The method for detecting the weld tightness of diamond composite sheets according to claim 1, characterized in that, The step of calculating the quantitative evaluation index of the weld interface density based on the multimodal defect feature matrix to obtain the weld quality grade evaluation result includes: Key feature parameters are extracted from the multimodal defect feature matrix, and the ratio of the total defect area to the detection area is calculated to obtain the defect area ratio as the first evaluation index of weld tightness. The defect depth, area, and signal intensity data in the multimodal defect feature matrix are normalized, and different weighting coefficients are set according to the defect type. The severity of each defect is nonlinearly weighted and summed to obtain the defect severity index as the second evaluation index of weld tightness. Information entropy is calculated for the thermal field distribution data and acoustic field distribution data in the multimodal defect feature matrix. Interface uniformity is evaluated by calculating the entropy difference and gradient distribution variance, and the welding interface uniformity index is obtained as the third evaluation index of welding density. Based on the acoustic reflection characteristics and X-ray density gradient data in the multimodal defect feature matrix, a support vector regression model is constructed, and the training interface is combined with the strength prediction function to output the predicted bonding strength value as the fourth evaluation index of weld tightness. A fuzzy membership function set is constructed, and the first, second, third, and fourth evaluation indicators of welding density are converted into fuzzy membership values. A variable weight adaptive algorithm is applied to dynamically adjust the weights of each indicator, and a comprehensive welding quality score is obtained by weighted summation. Cluster analysis is performed on historical inspection data to determine the optimal classification threshold. Based on the comprehensive welding quality score, the welding quality is divided into four levels to obtain the welding quality level evaluation result.

8. A testing system for the weld tightness of diamond composite sheets, characterized in that, For implementing the method for detecting the weld tightness of diamond composite sheets as described in any one of claims 1-7, the detection system for detecting the weld tightness of diamond composite sheets comprises: The acquisition module is used to acquire multimodal data from the diamond composite welded sample to be tested, and obtain the original ultrasonic dataset, the original X-ray dataset, and the original thermal imaging dataset. The input module is used to input the original ultrasonic dataset, the original X-ray dataset, and the original thermal imaging dataset into the diamond composite sheet welding interface feature enhancement and recognition algorithm for processing, so as to obtain the ultrasonic defect feature vector set, the X-ray defect feature vector set, and the thermal imaging defect feature vector set; The fusion module is used to perform spatial registration and data fusion on the ultrasonic defect feature vector set, X-ray defect feature vector set and thermal imaging defect feature vector set based on the diamond composite sheet welding interface feature enhancement and recognition algorithm, so as to obtain a fused defect distribution map and a multimodal defect feature matrix. The calculation module is used to calculate the quantitative evaluation index of the density of the welding interface based on the multimodal defect feature matrix, and obtain the welding quality grade evaluation result.

9. A testing device for the weld tightness of diamond composite sheets, characterized in that, The method includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the method for detecting the weld tightness of diamond composite sheets according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, the processor performs the method for detecting the weld tightness of diamond composite sheets as described in any one of claims 1 to 7.

Images