X-ray based online detection method and device for aramid fiber tape defects

Abstract

The present application relates to the field of aramid fiber belt defect detection, and discloses an X-ray-based online defect detection method and device for aramid fiber belts. The method comprises: performing potential defect analysis on physical property images and material parameter distribution maps to obtain spatial distribution data of potential defect areas; performing physical property quantitative analysis on the spatial distribution data of potential defect areas to obtain defect physical descriptors, which include density anomaly characteristics, molecular chain orientation disorder degree, and thermal anomaly characteristics; performing multi-dimensional feature extraction and weight distribution on defect physical descriptors according to preset material performance standards to obtain quality scores and physical performance parameters of aramid fiber belts, and generating defect detection reports based on the quality scores and physical performance parameters. The present application improves the comprehensiveness, accuracy, and reliability of aramid fiber belt defect detection.

Description

Technical Field

[0001] The present application relates to the technical field of aramid fiber belt defect detection, and in particular to an X-ray-based online detection method and device for aramid fiber belt defects. Background Art

[0002] Aramid fiber tapes are widely used in aerospace, military equipment, and high-performance composite materials due to their excellent mechanical properties, heat resistance, and chemical resistance. Traditional methods for defect detection of aramid fiber tapes mainly include visual inspection, optical microscope observation, and mechanical property testing. These methods can detect obvious surface defects and large-sized internal defects to a certain extent, but the detection effect on small defects and internal structural abnormalities is limited. With the continuous improvement of the quality requirements of aramid fiber tapes in industrial applications, X-ray detection technology has gradually been introduced into the defect detection of aramid fiber tapes. X-ray detection technology is non-destructive, fast, and high-resolution, and can achieve accurate imaging of the internal structure of aramid fiber tapes.

[0003] However, existing X-ray-based defect detection methods for aramid fiber tapes still have some shortcomings. First, most methods only use single-mode X-ray images, which cannot fully reflect the physical properties of aramid fiber tapes. Secondly, there is a lack of in-depth quantitative analysis of the physical properties of the detected defects, making it difficult to accurately evaluate the impact of defects on the performance of aramid fiber tapes. Furthermore, existing methods usually use fixed detection parameters and evaluation criteria, which are difficult to adapt to the detection needs of aramid fiber tapes of different specifications and application scenarios. In addition, there is a lack of a mechanism to directly link defect detection results with material properties, which makes it impossible to provide accurate quality control and optimization suggestions for the production process. Summary of the invention

[0004] (1) Technical problem: In order to overcome the deficiencies of the prior art, the present application provides an X-ray-based online defect detection method and device for aramid fiber tapes, which are used to solve the technical problems that the aramid fiber tape defect detection method in the prior art cannot fully reflect the physical properties of the aramid fiber tape, it is difficult to accurately evaluate the impact of defects on the performance of the aramid fiber tape, and the use of fixed detection parameters and evaluation standards is difficult to adapt to the detection needs of aramid fiber tapes of different specifications and application scenarios.

[0005] (2) Technical solution: In a first aspect, the present application provides an X-ray-based online defect detection method for aramid fiber tapes, the X-ray-based online defect detection method for aramid fiber tapes comprising: analyzing the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters; based on the X-ray imaging parameters, performing multimodal physical property imaging on a preset aramid fiber tape to obtain a multimodal data set including an X-ray transmission image, an X-ray dark field image, a polarized optical image, and a thermal image; performing multiscale image analysis on the multimodal data set to obtain a physical property image and a material parameter distribution map; performing potential defect analysis on the physical property image and the material parameter distribution map to obtain spatial distribution data of a potential defect area; performing physical property quantitative analysis on the spatial distribution data of the potential defect area to obtain a defect physical descriptor, the defect physical descriptor including density anomaly characteristics, molecular chain orientation disorder degree, and thermal anomaly characteristics; performing multidimensional feature extraction and weight allocation on the defect physical descriptor according to preset material performance standards to obtain a quality score and physical performance parameters of the aramid fiber tape, and generating a defect detection report based on the quality score and the physical performance parameters.

[0006] In a second aspect, the present application provides an X-ray-based online defect detection device for aramid fiber belts, the X-ray-based online defect detection device for aramid fiber belts comprising: an acquisition module for analyzing the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters; an imaging module for performing multimodal physical property imaging of a preset aramid fiber belt based on the X-ray imaging parameters to obtain a multimodal data set including an X-ray transmission image, an X-ray dark field image, a polarized optical image, and a thermal image; an analysis module for performing multiscale image analysis on the multimodal data set to obtain a physical property image and a material parameter distribution map; and a recognition module for performing multiscale image analysis on the multimodal data set to obtain a physical property image and a material parameter distribution map. A block is used to perform potential defect analysis on the physical property image and the material parameter distribution map to obtain spatial distribution data of the potential defect area; a quantification module is used to perform physical property quantitative analysis on the spatial distribution data of the potential defect area to obtain defect physical descriptors, wherein the defect physical descriptors include density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics; an extraction module is used to perform multi-dimensional feature extraction and weight allocation on the defect physical descriptors according to preset material performance standards to obtain the quality score and physical performance parameters of the aramid fiber tape, and generate a defect detection report based on the quality score and the physical performance parameters.

[0007] (3) Beneficial effects: In the technical solution provided by this application, the interaction between X-rays and aramid fibers is deeply analyzed to obtain optimized X-ray imaging parameters, which lays a solid foundation for subsequent multimodal imaging and significantly improves the imaging quality and detection sensitivity. Secondly, multimodal physical property imaging technology, including X-ray transmission imaging, X-ray dark field imaging, polarization optical imaging and thermal imaging, is used to comprehensively capture the density distribution, microstructure, molecular chain orientation and thermal properties of the aramid fiber tape, greatly enriching the information dimension of defect detection. The application of multiscale image analysis technology effectively extracts image features at different scales and enhances the recognition ability of tiny defects and large-scale structural anomalies. In the process of potential defect analysis, the defect area is accurately located and described through adaptive threshold segmentation and regional growing algorithm, which improves the accuracy and reliability of defect detection. The physical properties of the potential defect area are quantitatively analyzed to obtain defect physical descriptors including density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics, which not only realizes the qualitative identification of defects, but also provides quantitative physical property evaluation, laying the foundation for subsequent performance prediction. Finally, by extracting multi-dimensional features and assigning weights to the physical descriptors of defects, combined with the preset material performance standards, the quality scores and physical performance parameters of the aramid fiber tapes were obtained, achieving a seamless connection from defect detection to performance evaluation. This comprehensive evaluation method not only provides a more objective and comprehensive quality evaluation, but also provides precise guidance for the optimization of the production process. The generation of defect detection reports integrates defect distribution maps and comprehensive quality evaluation results, providing an intuitive and detailed decision-making basis for production management and quality control. Through multimodal data fusion, multi-scale analysis and physical property quantification, the comprehensiveness, accuracy and reliability of aramid fiber tape defect detection have been significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0009] Figure 1 This is a schematic diagram of an embodiment of an X-ray-based online detection method for aramid fiber belt defects in an embodiment of the present application; Figure 2 It is a schematic diagram of an embodiment of an X-ray-based online defect detection device for aramid fiber belts in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The embodiments of the present application provide an X-ray-based online detection method and device for defects in aramid fiber belts. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0011] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the online detection method of aramid fiber belt defects based on X-rays includes: step S101, analyzing the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters; step S102, based on the X-ray imaging parameters, performing multi-modal physical property imaging on the preset aramid fiber belt to obtain a multi-modal data set including X-ray transmission images, X-ray dark field images, polarized optical images and thermal images; step S103, performing multi-scale image analysis on the multi-modal data set to obtain a physical property image and a material parameter distribution map; step S104, Perform potential defect analysis on the physical property image and material parameter distribution map to obtain spatial distribution data of the potential defect area; step S105, perform physical property quantitative analysis on the spatial distribution data of the potential defect area to obtain defect physical descriptors, which include density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics; step S106, perform multi-dimensional feature extraction and weight allocation on the defect physical descriptors according to preset material performance standards to obtain the quality score and physical performance parameters of the aramid fiber tape, and generate a defect detection report based on the quality score and physical performance parameters.

[0012] It is understandable that the execution subject of the present application may be an X-ray based aramid fiber belt defect online detection device, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0013] Specifically, the interaction between X-rays and aramid fibers is analyzed to obtain the optimal X-ray imaging parameters. This involves modeling the molecular structure and density distribution of aramid fibers, as well as calculating the propagation path of X-rays in aramid fibers. Through the Monte Carlo simulation method, the scattering and absorption effects of X-rays are considered to optimize the X-ray energy spectrum range and the source-detector geometric configuration. The special structure of aramid fibers, such as the high orientation of molecular chains and the hydrogen bond network, are considered, which will affect the attenuation and scattering characteristics of X-rays. Based on the obtained X-ray imaging parameters, multimodal physical property imaging is performed on the preset aramid fiber tape. This includes X-ray transmission imaging, X-ray dark field imaging, polarization optical imaging, and thermal imaging. X-ray transmission imaging provides material density distribution information and can detect density anomalies such as bubbles and impurities. X-ray dark field imaging enhances the sensitivity to small structural changes and is particularly suitable for detecting small changes in fiber orientation. Polarization optical imaging reflects the orientation of molecular chains and can identify stress concentration areas or local orientation anomalies. Thermal imaging provides additional structural information and can detect abnormal points of heat conduction, such as internal cracks or interface debonding. These multimodal image data are collected synchronously in space and time with high precision to form a comprehensive multimodal dataset.

[0014] Multiscale image analysis of multimodal datasets is the next key step. First, the image is decomposed into subbands of different scales using wavelet transform, and then each subband is denoised by adaptive threshold. This process can effectively remove image noise while retaining important structural information. The Canny edge detection algorithm is applied to the denoised image to extract fiber structural features, and the boundaries are enhanced through morphological operations such as dilation and erosion. Subsequently, local binary pattern (LBP) features and gray-level co-occurrence matrix (GLCM) features are calculated to construct texture feature descriptors. LBP features can capture local texture changes, while GLCM features provide global texture statistics. At the same time, the Beer-Lambert law is used to reconstruct the material density distribution from the X-ray transmission image, and the local orientation and crystallinity are estimated from the X-ray dark field image. Finally, this information is integrated through data fusion techniques such as principal component analysis (PCA) or independent component analysis (ICA) to obtain a comprehensive physical property image and material parameter distribution map. After obtaining the physical property image and material parameter distribution map, potential defect analysis is performed. First, the abnormal area is preliminarily divided by adaptive threshold segmentation, and then the potential defect boundary is extended using the region growing algorithm. It can accurately locate defective areas and exclude noise and false positives. The obtained candidate defective areas are subjected to shape analysis to calculate geometric characteristic parameters such as area, perimeter, and circularity. These parameters help to distinguish different types of defects, such as bubbles are usually round, while cracks are elongated. Then the density distribution, orientation information and thermal distribution characteristics are analyzed to calculate the degree of local abnormality. Through the multi-feature fusion scoring mechanism, the shape, density, orientation and thermal characteristics are comprehensively considered to screen out the final potential defective areas, and construct a spatial distribution data structure containing position and boundary information. For the identified potential defective areas, in-depth physical property quantitative analysis is carried out. This includes calculating density abnormality characteristics, molecular chain orientation disorder degree and thermal abnormality characteristics. The density abnormality characteristics are obtained by comparing the density difference between the defective area and the surrounding normal area, and can be used to identify defects such as bubbles and impurities. The degree of molecular chain orientation disorder uses X-ray dark field image and polarized optical image information to calculate orientation entropy and orientation consistency. This feature is crucial for detecting local stress concentration or fiber orientation abnormality. Thermal anomaly features are based on thermal image data, analyzing temperature distribution, temperature gradient and thermal diffusion coefficient, and can detect defects that affect heat conduction, such as internal cracks or interface debonding. These features are normalized and combined to form defect physical descriptors.

[0015] Finally, according to the preset material performance standards, multi-dimensional feature extraction and weight assignment are performed on the defect physical descriptors. This process involves steps such as feature normalization, weight assignment, and nonlinear mapping, and finally the quality score of the aramid fiber tape is obtained. At the same time, key physical performance parameters such as mechanical properties, thermal stability, and durability are estimated based on the defect physical descriptors. Combining the quality score, physical performance parameters, and defect distribution map, a comprehensive defect detection report is generated to provide decision support for production quality control.

[0016] For example, a roll of aramid fiber tape with a length of 1000 meters passed through the inspection system. The system collected 10,000 frames of multimodal images, each with a resolution of 2048x2048 pixels. After multi-scale analysis, the system identified 50 potential defect areas. Through quantitative analysis of physical properties, it was found that 30 of them were density anomalies (average density deviation of 5%), 15 were disordered molecular chain orientation (orientation entropy increased by 20%), and 5 were thermal anomalies (local temperature increased by 3°C). The system calculated that the quality score of the roll of aramid fiber tape was 92 points (out of 100 points), and predicted its tensile strength to be 3.2 GPa and thermal decomposition temperature to be 550°C. This information is integrated into the defect detection report, including the type, location, severity of the defect and the potential impact on material properties.

[0017] In the embodiment of the present application, by conducting an in-depth analysis of the interaction between X-rays and aramid fibers, optimized X-ray imaging parameters are obtained, which lays a solid foundation for subsequent multimodal imaging and significantly improves the imaging quality and detection sensitivity. Secondly, multimodal physical property imaging technology, including X-ray transmission imaging, X-ray dark field imaging, polarized optical imaging and thermal imaging, is used to comprehensively capture the density distribution, microstructure, molecular chain orientation and thermal properties of the aramid fiber belt, greatly enriching the information dimension of defect detection. The application of multiscale image analysis technology effectively extracts image features at different scales and enhances the recognition ability of tiny defects and large-scale structural anomalies. In the process of potential defect analysis, the defect area is accurately located and described through adaptive threshold segmentation and regional growth algorithm, which improves the accuracy and reliability of defect detection. The physical properties of the potential defect area are quantitatively analyzed to obtain defect physical descriptors including density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics, which not only realizes the qualitative identification of defects, but also provides quantitative physical property evaluation, laying the foundation for subsequent performance prediction. Finally, by extracting multi-dimensional features and assigning weights to the physical descriptors of defects, combined with the preset material performance standards, the quality scores and physical performance parameters of the aramid fiber tapes were obtained, achieving a seamless connection from defect detection to performance evaluation. This comprehensive evaluation method not only provides a more objective and comprehensive quality evaluation, but also provides precise guidance for the optimization of the production process. The generation of defect detection reports integrates defect distribution maps and comprehensive quality evaluation results, providing an intuitive and detailed decision-making basis for production management and quality control. Through multimodal data fusion, multi-scale analysis and physical property quantification, the comprehensiveness, accuracy and reliability of aramid fiber tape defect detection have been significantly improved.

[0018] In a specific embodiment, the process of executing step S101 may specifically include the following steps: (1) numerically calculating the molecular structure and density distribution of the aramid fiber to determine the attenuation characteristics of the aramid fiber, and performing Monte Carlo simulation on the attenuation characteristics to generate X-ray propagation path data in the aramid fiber; (2) optimizing the X-ray energy spectrum range based on the propagation path data to determine the target X-ray energy spectrum range, and discretizing the target X-ray energy spectrum range to form a discrete energy spectrum point set; (3) parameterizing the X-ray tube voltage and current based on the discrete energy spectrum point set to determine the initial X-ray (4) Correct the initial X-ray source parameters according to the scattered radiation correction factor, determine the corrected X-ray source parameters, perform a matching analysis on the corrected X-ray source parameters and the detector response characteristics, and set the detector parameters; (5) Calculate the geometric layout of the X-ray imaging system based on the corrected X-ray source parameters and detector parameters, determine the spatial position relationship of the source, sample, and detector, and perform a balance calculation of the spatial resolution and penetration ability on the spatial position relationship to generate the X-ray imaging parameters.

[0019] Specifically, the molecular structure and density distribution of aramid fiber are numerically calculated to determine its attenuation characteristics. Due to its unique molecular structure, aramid fiber has high strength and heat resistance, which directly affect its X-ray attenuation performance. The attenuation characteristics can be expressed by the following formula:

[0020] ,in, is the transmitted X-ray intensity, is the incident X-ray intensity, is the linear attenuation coefficient, is the material thickness. In the embodiment of the present application, for aramid fiber, μ is a function of energy E, which can be expressed as:

[0021] ,here, is the density of aramid fiber, is the mass fraction of the ith element, is the mass attenuation coefficient of the ith element at energy E.

[0022] The Monte Carlo simulation method is used to generate the propagation path data of X-rays in aramid fibers. This method simulates the propagation process of a large number of photons by random sampling, taking into account the interaction between photons and aramid fiber molecules, including photoelectric effect, Compton scattering and Rayleigh scattering. The simulation results provide key information such as the depth penetration of X-rays in aramid fibers and the distribution of scattering angles. Based on the propagation path data, the X-ray energy spectrum range is optimized and calculated. The optimization goal is to maximize image contrast and signal-to-noise ratio while minimizing radiation dose. The target X-ray energy spectrum range is determined by analyzing the penetration ability and attenuation differences of X-rays of different energies. For example, for a typical aramid fiber tape, the energy range may be between 20-80 keV. This continuous energy spectrum range is discretized to form a discrete energy spectrum point set, such as . Among them, each Represents a specific energy point.

[0023] According to the discrete energy spectrum point set, the X-ray tube voltage and current are parameterized. X-ray tube voltage V and maximum photon energy The relationship can be expressed as:

[0024] ,in, is the electron charge. The current I is proportional to the X-ray intensity. By adjusting V and I, the initial X-ray source parameters are determined. Subsequently, the scattered radiation effect of these initial parameters is evaluated and the scattered radiation correction factor is calculated. :

[0025] , is the total receiving intensity, is the main ray intensity.

[0026] The initial X-ray source parameters are corrected using the scattered radiation correction factor to obtain the corrected X-ray source parameters. These corrected parameters are matched with the detector response characteristics to set appropriate detector parameters, such as integration time and gain. The quantum efficiency of the detector is a key parameter that varies with the X-ray energy:

[0027] , is the linear attenuation coefficient of the detector material, and d is the detector thickness. Finally, according to the corrected X-ray source parameters and detector parameters, the geometric layout of the X-ray imaging system is calculated to determine the spatial position relationship between source-sample-detector. This step needs to consider the balance between spatial resolution R and penetration ability P, which can be expressed by the following relationship:

[0028] , is the focal spot size, M is the geometric magnification, and t is the sample thickness. By adjusting these parameters, the final X-ray imaging parameter set is generated.

[0029] For example, assume a specific aramid fiber tape with a density of 1.44 g / cm3 and mainly composed of C, H, N, and O elements. Through initial calculations, its linear attenuation coefficient at 50 keV was determined to be approximately 0.3 cm-1. Monte Carlo simulation of the propagation of 1 million photons gave an average penetration depth of 2.5 cm. Based on these data, the optimized X-ray energy spectrum range was determined to be 30-70 keV, discretized into 10 energy points. The X-ray tube voltage was set to 75 kV and the current was 100 mA. The calculated scattered radiation correction factor was 1.15. After matching the detector parameters, the optimal source-sample distance was determined to be 50 cm and the sample-detector distance was 10 cm, achieving a spatial resolution of 5μm and a penetration of 90%. These parameters together constitute the optimal X-ray imaging parameter set for this aramid fiber tape.

[0030] In a specific embodiment, the process of executing step S102 may specifically include the following steps: (1) decomposing the X-ray imaging parameters, extracting the X-ray source parameters and the detector parameters, and setting the voltage and current of the X-ray source according to the X-ray source parameters; (2) adjusting the response characteristics of the X-ray detector according to the detector parameters, and performing an X-ray transmission scan on the preset aramid fiber tape to generate an X-ray transmission image; (3) inserting a phase grating between the X-ray source and the X-ray detector to perform X-ray dark field imaging, and reconstructing the X-ray dark field image by a phase stepping method; (4) adjusting the polarization direction of the polarized light source, performing a polarization optical scan on the preset aramid fiber tape, and calculating a synthetic polarization optical image by Stokes parameters; (5) starting the infrared thermal imager, performing a thermal imaging scan on the preset aramid fiber tape, and performing temperature calibration and noise suppression on the collected thermal image data to generate a thermal image; (6) spatially registering the X-ray transmission image, the X-ray dark field image, the polarization optical image, and the thermal image, and fusing the multimodal images based on the registration result to obtain a multimodal data set.

[0031] Specifically, the X-ray imaging parameters are decomposed to extract the X-ray source parameters and detector parameters. The X-ray source parameters include tube voltage, tube current, focal spot size, etc., while the detector parameters include pixel size, sensitivity, dynamic range, etc. According to the extracted X-ray source parameters, the voltage and current of the X-ray source are set, which directly affects the energy distribution and intensity of the X-rays. For example, for an aramid fiber belt, the tube voltage may be set to 60 kV and the tube current may be set to 100 mA to obtain appropriate penetration and image contrast. The response characteristics of the X-ray detector are adjusted according to the detector parameters. This includes setting the integration time, gain, and dynamic range. For example, the integration time may be set to 100 ms and the gain may be set to 2 to balance the signal-to-noise ratio and image acquisition speed. After the adjustment is completed, an X-ray transmission scan is performed on the preset aramid fiber belt to generate an X-ray transmission image. The X-ray passes through the fiber belt, and its intensity decays according to the Beer-Lambert law. The detector captures the transmitted X-ray and converts it into a digital signal to form a grayscale image.

[0032] In order to obtain more abundant material information, a phase grating is inserted between the X-ray source and the detector to perform X-ray dark field imaging. The phase grating is usually composed of a periodic structure that can produce interference and diffraction effects of X-rays. The X-ray dark field image can be reconstructed by the phase stepping method, that is, gradually moving the grating and collecting images at each position. This image can show the slight structural changes and edge information of the material, which is very effective for detecting subtle defects in the aramid fiber tape. The polarization direction of the polarized light source is adjusted to perform polarization optical scanning on the preset aramid fiber tape. Aramid fiber has obvious optical anisotropy due to its highly oriented molecular structure. By rotating the polarization direction of the polarized light source, polarization images at different angles can be obtained. Then, the polarization optical image is synthesized using the Stokes parameter calculation method. The Stokes parameters include four components, which describe the total intensity, linear polarization degree and circular polarization degree of light respectively. By calculating these parameters, a comprehensive image reflecting the molecular orientation and stress distribution of the aramid fiber can be obtained. In addition, the infrared thermal imager is started to perform thermal imaging scanning on the preset aramid fiber tape. Thermal imaging can reflect the thermal distribution of materials and help detect internal structural defects or stress concentration areas. The collected thermal image data needs to be temperature calibrated and noise suppressed. Temperature calibration usually uses a black body radiation source as a reference to establish the correspondence between pixel values ​​and actual temperatures. Noise suppression can use methods such as median filtering or wavelet transform to improve the signal-to-noise ratio of the image. After these processes, the final thermal image is generated. Finally, the X-ray transmission image, X-ray dark field image, polarization optical image and thermal image are spatially registered. The purpose of spatial registration is to align images of different modalities to the same coordinate system for subsequent fusion and analysis. The registration process usually includes steps such as feature point extraction, transformation matrix calculation and image resampling. Based on the registration results, the multimodal images are fused using a data fusion algorithm to obtain the final multimodal data set. The fusion algorithm can use methods such as weighted averaging, principal component analysis or deep learning to make full use of the complementary information of each modality image.

[0033] For example, in an aramid fiber tape inspection, the X-ray source was set to a tube voltage of 65 kV, a tube current of 120 mA, and a detector integration time of 80 ms. X-ray transmission scanning obtained a 16-bit grayscale image with a resolution of 2048 × 2048 pixels. X-ray dark field imaging used a phase grating with a period of 4 μm, and reconstructed the dark field image by 5 phase steps. Polarization optical scanning was performed in four polarization directions of 0°, 45°, 90°, and 135°, and a polarization image reflecting the molecular orientation was calculated. Thermal imaging used a long-wave infrared camera with a resolution of 640 × 480 pixels and a temperature resolution of 0.05°C. Spatial registration used an affine transformation model, and the registration accuracy reached the sub-pixel level. The fused multimodal data set contains comprehensive information such as density distribution, microstructure information, molecular orientation, and thermal distribution.

[0034] In a specific embodiment, the process of executing step S103 may specifically include the following steps: (1) performing wavelet transform on the multimodal data set to decompose it into sub-band images of different scales, and performing noise suppression on each sub-band image to obtain a denoised multi-scale image; (2) performing edge detection on the denoised multi-scale image to extract fiber structure features, and enhancing the fiber boundaries through morphological operations to obtain an enhanced structure feature map; (3) performing texture analysis on the enhanced structure feature map to calculate local binary pattern features, and constructing a texture feature descriptor based on gray level co-occurrence matrix features; (4) calculating the material density distribution based on the gray level distribution of the X-ray transmission image, and estimating the local orientation and crystallinity of the material based on the X-ray dark field image information to obtain a preliminary material parameter distribution map; (5) performing polarization degree and polarization direction analysis on the polarized optical image to extract molecular chain orientation information, and constructing a comprehensive physical property image based on the temperature distribution data of the thermal image; (6) performing data fusion on the preliminary material parameter distribution map and the comprehensive physical property image to obtain a physical property image and a material parameter distribution map.

[0035] Specifically, wavelet transform is applied to the multimodal data set to decompose the image into sub-band images of different scales. Wavelet transform can effectively decompose image information in the spatial and frequency domains for multi-resolution, which is conducive to subsequent feature extraction and analysis. For each sub-band image, an adaptive threshold denoising algorithm is used for noise suppression, such as the BayesShrink method, which adaptively selects the threshold according to the statistical characteristics of each sub-band, effectively removes noise while retaining image details, thereby obtaining a denoised multi-scale image. Edge detection is performed on the denoised multi-scale image to extract fiber structural features. The Canny edge detection algorithm is used here, which includes steps such as Gaussian filtering, gradient calculation, non-maximum suppression and double threshold processing, and can accurately detect the edges of fibers. In order to further enhance the fiber boundaries, morphological operations such as a combination of dilation and erosion are used. For example, a dilation operation is first performed to connect the disconnected edges, and then an erosion operation is performed to refine the boundary lines, and finally an enhanced structural feature map is obtained.

[0036] The enhanced structural feature map is texture analyzed and the local binary pattern (LBP) feature is calculated. The LBP algorithm generates binary code by comparing the grayscale values ​​of the central pixel and its neighboring pixels. This code can effectively describe the local texture information. At the same time, the gray level co-occurrence matrix (GLCM) feature is calculated. GLCM reflects the spatial relationship between pixel pairs in the image, from which statistical features such as contrast, correlation, energy and homogeneity can be extracted. The LBP feature and the GLCM feature are combined to construct a texture feature descriptor. According to the grayscale distribution of the X-ray transmission image, the material density distribution is calculated using the Beer-Lambert law to obtain the density distribution of the material. At the same time, the X-ray dark field image information is used to estimate the local orientation and crystallinity of the material. The X-ray dark field image can provide information about the internal microstructure of the material. By analyzing the intensity distribution and directionality of the dark field image, the local orientation and crystallinity of the fiber can be estimated. Combining this information, a preliminary material parameter distribution map is obtained. The polarization degree and polarization direction of the polarized optical image are analyzed to extract the molecular chain orientation information. Polarized optical images contain birefringence information of materials. By analyzing the changes in image intensity under different polarization directions, the degree of polarization and polarization direction can be calculated. These parameters directly reflect the orientation state of the molecular chains in aramid fibers. At the same time, combined with the temperature distribution information of thermal images, comprehensive physical property images are constructed. Thermal images reflect the thermal properties of materials and can reveal internal structural defects or stress concentration areas.

[0037] Finally, the preliminary material parameter distribution map and the comprehensive physical property image are fused to obtain the final physical property image and material parameter distribution map. The data fusion adopts a method combining weighted average method and principal component analysis (PCA). The weighted average method takes into account the reliability and importance of different modal data, while PCA can extract the main features in the data and reduce redundant information.

[0038] For example, in an aramid fiber band inspection, a 3-layer wavelet decomposition was performed on a multimodal image of 2048×2048 pixels to obtain 10 sub-band images. The BayesShrink method was applied to denoise each sub-band, and the noise level was reduced by about 75%. The high and low thresholds of Canny edge detection were set to the 90% and 45% quantiles of the image gradient amplitude, respectively. The LBP feature used a circular neighborhood with a radius of 3 to extract a 256-dimensional feature vector. The GLCM calculation considered four directions of 0°, 45°, 90°, and 135°, and extracted 16 statistical features. The grayscale value range of the X-ray transmission image was 0-65535. By looking up the pre-established attenuation coefficient table, the grayscale value was converted to a density value, and a density distribution map with a density range of 1.2-1.6 g / cm³ was obtained. The molecular chain orientation angle range obtained by polarization optical analysis was 0-180°, and the polarization degree range was 0-1. The temperature range of the thermal image was 20-35°C, and the resolution was 0.05°C. The final fused material parameter distribution map includes multiple parameters such as density, orientation, and crystallinity.

[0039] In a specific embodiment, the process of executing step S104 may specifically include the following steps: (1) performing adaptive threshold segmentation on the physical property image and the material parameter distribution map to obtain abnormal regions, and expanding the boundaries of the abnormal regions through a region growing algorithm to obtain preliminary potential defect regions; (2) performing shape analysis on the preliminary potential defect regions, calculating geometric feature parameters, and screening the preliminary potential defect regions according to preset shape rules to obtain potential defect regions after shape screening; (3) performing feature extraction on the potential defect regions after shape screening in the physical property image, calculating local physical property values, and extracting material parameter values ​​of the corresponding regions in the material parameter distribution map; (4) performing statistical analysis on the local physical property values ​​and material parameter values, calculating the degree of difference from the surrounding normal regions, and obtaining an abnormality index; (5) performing multi-feature fusion on the geometric feature parameters and abnormality index of the potential defect regions after shape screening, calculating a comprehensive score, and screening according to the comprehensive score to obtain potential defect regions; (6) performing spatial positioning and contour extraction on the potential defect regions, constructing a data structure containing position coordinates and boundary information, and obtaining spatial distribution data of the potential defect regions.

[0040] Specifically, the Otsu adaptive threshold algorithm is used, which determines the optimal threshold T by maximizing the inter-class variance, which can be expressed as:

[0041] ,in, and are the probabilities of foreground and background, respectively, and is the average gray value of the foreground and background. In this way, the abnormal area can be effectively separated from the background. The region growing algorithm is used to expand the boundary of the abnormal area. The region growing algorithm starts from the initial seed point and gradually incorporates adjacent similar pixels into the region. The similarity judgment criterion can be defined as:

[0042] , where I(x,y) is the gray value of the pixel to be judged, is the average gray value of the current area, is the standard deviation of the area, and k is the control parameter. Through this process, the preliminary potential defect area is obtained. The shape analysis of the preliminary potential defect area is performed to calculate the geometric characteristic parameters. These parameters include area A, perimeter P, circularity C, etc. The circularity can be calculated by the following formula:

[0043] , the closer the circularity value is to 1, the closer the shape is to a circle. According to the preset shape rules, such as area threshold, circularity range, etc., the preliminary potential defect area is screened to obtain the potential defect area after shape screening. For the potential defect area after shape screening, the local physical property values, such as density, orientation, etc., are extracted from the physical property image, and the material parameter values ​​of the corresponding area are extracted from the material parameter distribution map. These values ​​can be characterized by calculating the average and standard deviation of the pixels in the area.

[0044] Subsequently, the local physical property values ​​and material parameter values ​​are statistically analyzed, and the degree of difference from the surrounding normal area is calculated to obtain the abnormality index. Finally, the geometric characteristic parameters and abnormality index of the potential defect area after shape screening are multi-feature fused to calculate the comprehensive score. The weighted summation method can be used:

[0045] ,here, is the i-th feature, is the corresponding weight. Screening is performed based on the comprehensive score to obtain the final potential defect area. These potential defect areas are spatially located and contour extracted, and a data structure containing position coordinates and boundary information is constructed to obtain the spatial distribution data of the potential defect area. This data structure may include defect center coordinates, boundary point list, area, perimeter and other information.

[0046] For example, in an aramid fiber belt inspection, a 2048×2048 pixel physical property image was processed. The optimal threshold calculated by the Otsu algorithm was 127 (within the grayscale range of 0-255). The region growing algorithm used k=2.5 as a control parameter to expand the initial segmented area. In shape analysis, the area threshold was set to 100 square pixels and the circularity range was 0.6-1.0. After shape screening, 50 potential defect areas were obtained. In the physical property analysis, it was found that the density value of a typical defect area was 5% lower than that of the surrounding normal area, and the orientation angle deviation exceeded 15°. The density anomaly calculated by the Z-score method was -2.3, and the orientation anomaly was 3.1. In the comprehensive score, the weights of density, orientation and shape features were 0.4, 0.4 and 0.2 respectively, and the comprehensive score of the defect area was calculated to be 0.85 (full score is 1). Finally, the defect was confirmed as a potential defect with a center coordinate of (1024, 768), an area of ​​150 square pixels, and a perimeter of 44 pixels.

[0047] In a specific embodiment, the process of executing step S105 may specifically include the following steps: (1) performing regional segmentation on the spatial distribution data of the potential defect area, extracting the contour information of each defect area, and calculating the area, perimeter and shape factor of each defect area to obtain defect geometric features; (2) calculating the average density and density variance in each defect area according to the grayscale information corresponding to each defect area in the X-ray transmission image, and comparing them with the surrounding normal area to obtain density anomaly features; (3) analyzing the molecular chain orientation distribution in the defect area through the image parameter information corresponding to each defect area in the X-ray dark field image and the polarized optical image, calculating the orientation entropy and orientation consistency, and obtaining the degree of molecular chain orientation disorder; (4) calculating the temperature distribution characteristics of each defect area based on the data corresponding to each defect area in the thermal image, including the average temperature deviation, temperature gradient and thermal diffusion coefficient, to obtain thermal anomaly features; (5) normalizing the defect geometric features, density anomaly features, molecular chain orientation disorder degree and thermal anomaly features to obtain defect physical descriptors.

[0048] Specifically, the spatial distribution data of the potential defect area is segmented, and the edge tracking algorithm is used to extract the contour information of each defect area. In this process, the 8-connectivity principle is used to start from the edge point and track the boundary pixels in a clockwise or counterclockwise direction until returning to the starting point, thereby obtaining complete contour information. Based on the extracted contour information, the geometric features of each defect area are calculated, including area, perimeter and shape factor. The area is obtained by calculating the number of pixels inside the contour, and the perimeter is the number of pixels on the contour. The shape factor is a dimensionless parameter that describes the degree of regularity of the shape, which is calculated by the relationship between the area and the perimeter. These geometric features together constitute the geometric description of the defect, providing a basis for subsequent analysis.

[0049] Then, according to the grayscale information corresponding to each defect area in the X-ray transmission image, the average density and density variance in each defect area are calculated. The grayscale value of the X-ray transmission image is inversely proportional to the material density, so the density information can be obtained by grayscale value inversion. The average density and density variance of the defect area are calculated and compared with the surrounding normal area to obtain the density anomaly characteristics. In this process, the local contrast enhancement algorithm is used to improve the detection sensitivity of density anomalies. The image parameter information corresponding to each defect area in the X-ray dark field image and the polarized optical image is used to analyze the orientation distribution of the molecular chain in the defect area. The X-ray dark field image can reflect the microstructural changes inside the material, while the polarized optical image directly reflects the orientation state of the molecular chain. The degree of molecular chain orientation disorder is obtained by calculating the orientation entropy and orientation consistency. The orientation entropy describes the uniformity of the orientation distribution, and the orientation consistency reflects the concentration of the orientation. These two parameters together describe the degree of abnormality of the molecular chain arrangement in the defect area.

[0050] At the same time, based on the data corresponding to each defect area in the thermal image, the temperature distribution characteristics of each defect area are calculated. This includes the average temperature deviation, temperature gradient and thermal diffusion coefficient. The average temperature deviation reflects the temperature difference between the defect area and the surrounding normal area, the temperature gradient describes the spatial distribution of temperature changes, and the thermal diffusion coefficient reflects the ability of the material to conduct heat. These parameters constitute the thermal anomaly characteristics, which can effectively identify thermal performance anomalies caused by internal structural changes or material inhomogeneities. Finally, the defect geometry characteristics, density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics are normalized to obtain standardized defect physical descriptors. The normalization process uses the minimum-maximum normalization method to map each eigenvalue to the [0,1] interval to ensure that features of different dimensions can be compared and comprehensively analyzed.

[0051] For example, in an aramid fiber tape inspection, a potential defect area with an area of ​​500 square pixels and a circumference of 80 pixels was detected. The calculated shape factor is 0.98, which is close to 1, indicating that the defect shape is approximately circular. X-ray transmission image analysis shows that the average grayscale value of this area is 10% lower than that of the surrounding normal area, and the corresponding density is about 5% higher than that of the normal area, with a density variance of 0.02g / cm3. The orientation entropy obtained by X-ray dark field image and polarized optical image analysis is 0.8, and the orientation consistency is 0.6, indicating a high degree of molecular chain orientation disorder. Thermal image analysis shows that the average temperature of this area is 2°C higher than the surrounding area, the temperature gradient is 0.1°C / pixel, and the thermal diffusion coefficient is 15% lower than that of the normal area. After normalization, the defect physical descriptor obtained is [0.95, 0.85, 0.8, 0.7], corresponding to the normalized values ​​of geometric features, density anomaly, orientation disorder and thermal anomaly, respectively.

[0052] In a specific embodiment, the process of executing step S106 may specifically include the following steps: (1) performing feature normalization processing on the defect physical descriptor to obtain a normalized feature vector; (2) assigning a weight coefficient to each feature component in the normalized feature vector according to the material performance standard to obtain a weighted feature vector; (3) performing nonlinear mapping on the weighted feature vector to convert the multidimensional feature into a one-dimensional score to obtain the quality score of the aramid fiber tape; (4) based on the density anomaly feature, the degree of molecular chain orientation disorder and the thermal anomaly feature in the defect physical descriptor, the mechanical properties, thermal stability and durability of the aramid fiber tape are estimated by regression analysis to obtain physical performance parameters; (5) the aramid fiber tape is graded according to the quality score, and a comprehensive quality assessment result is generated according to the physical performance parameters; (6) based on the spatial distribution data of the potential defect area, a defect distribution map is generated, and the defect distribution map is fused with the comprehensive quality assessment result to obtain a defect detection report.

[0053] Specifically, the minimum-maximum normalization method is used to process each feature in the defect physical descriptor, map it to the range of 0 to 1, and obtain the normalized feature vector. This process ensures that features of different dimensions and ranges can be effectively compared and comprehensively analyzed. According to the pre-established material performance standards, weight coefficients are assigned to each feature component in the normalized feature vector. This step takes into account the degree of influence of different features on the performance of the aramid fiber tape and assigns higher weights to more important features. The weight allocation adopts the analytic hierarchy process (AHP), and the final weight coefficient is determined by expert judgment and consistency test. This method can effectively balance subjective judgment and objective data to obtain a scientific and reasonable weight allocation result. Subsequently, the weighted feature vector is nonlinearly mapped to convert the multidimensional features into a one-dimensional score to obtain the quality score of the aramid fiber tape. The sigmoid function is used as the nonlinear mapping function here. The sigmoid function is characterized by being able to compress the input to between 0 and 1, and has good smoothness and nonlinear characteristics, which is suitable for the calculation of comprehensive scores. By adjusting the parameters of the function, the sensitivity and distribution characteristics of the score can be controlled.

[0054] Next, based on the density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics in the defect physical descriptors, the mechanical properties, thermal stability and durability of the aramid fiber tape were estimated through multivariate regression analysis to obtain the physical performance parameters. The regression model was established based on a large amount of historical data and experimental results, and the regression coefficient was determined by the least squares method. Then, the aramid fiber tape was graded according to the quality score, and the comprehensive quality evaluation results were generated in combination with the physical performance parameters. The grade division adopts the fixed threshold method to divide the score interval into four levels: excellent, good, qualified and unqualified. The comprehensive quality evaluation results include not only the grade information, but also the specific values ​​of each physical performance parameter and the comparison with the standard value, providing a comprehensive reference for production management and quality control.

[0055] Finally, based on the spatial distribution data of the potential defect area, a defect distribution map is generated. The defect distribution map is in the form of a heat map, which intuitively displays the location and density distribution of defects on the aramid fiber tape. The heat map uses color gradients to indicate the severity of the defects, making the defect distribution clear at a glance. The defect distribution map is fused with the comprehensive quality assessment results to obtain the final defect detection report. During the data fusion process, the defect distribution information is combined with the quality assessment results to form a comprehensive description of the quality status, including the location, type, severity and impact assessment of the defects on the overall performance.

[0056] For example, in an aramid fiber tape inspection, a 100-meter-long fiber tape was analyzed. The defect physical descriptor contains four features: geometric features, density anomalies, orientation disorder, and thermal anomalies. The original eigenvalues ​​are 0.8, 0.6, 0.7, and 0.5, respectively. After normalization, the normalized eigenvectors are 0.75, 0.5, 0.625, and 0.25. According to the material performance standards, the weight coefficients assigned are 0.3, 0.25, 0.25, and 0.2, and the weighted eigenvectors obtained are 0.225, 0.125, 0.15625, and 0.05. Through nonlinear mapping, the quality score is calculated to be 0.82. The physical performance parameters estimated by multivariate regression analysis include: tensile strength 3.2GPa (standard value ≥3GPa), thermal decomposition temperature 520°C (standard value ≥500°C), and durability index 0.95 (standard value ≥0.9). Based on the score and performance parameters, the section of aramid fiber tape was rated as "good". The defect distribution map showed a large defect cluster at 70 meters. This information was integrated into the final defect inspection report, providing precise guidance for production quality control. This comprehensive report not only points out the location and severity of the defects, but also gives specific performance parameters and quality grades.

[0057] The above describes the on-line detection method of aramid fiber belt defects based on X-rays in the embodiment of the present application. The following describes the on-line detection device of aramid fiber belt defects based on X-rays in the embodiment of the present application. Figure 2 In one embodiment of the present application, an online defect detection device for aramid fiber tapes based on X-rays comprises: an acquisition module 201 for analyzing the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters; an imaging module 202 for performing multimodal physical property imaging of a preset aramid fiber tape based on the X-ray imaging parameters to obtain a multimodal data set including an X-ray transmission image, an X-ray dark field image, a polarized optical image, and a thermal image; an analysis module 203 for performing multiscale image analysis on the multimodal data set to obtain a physical property image and a material parameter distribution map; an identification module 204 for identifying the physical property image; and an image processing module 205 for analyzing the physical property image of the preset aramid fiber tape. The physical characteristic image and the material parameter distribution map are used to perform potential defect analysis to obtain spatial distribution data of the potential defect area; the quantification module 205 is used to perform physical characteristic quantitative analysis on the spatial distribution data of the potential defect area to obtain defect physical descriptors, and the defect physical descriptors include density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics; the extraction module 206 is used to perform multi-dimensional feature extraction and weight allocation on the defect physical descriptors according to preset material performance standards, to obtain the quality score and physical performance parameters of the aramid fiber tape, and to generate a defect detection report according to the quality score and the physical performance parameters.

[0058] Through the synergy of the above components, the optimized X-ray imaging parameters are obtained by in-depth analysis of the interaction between X-rays and aramid fibers, which lays a solid foundation for subsequent multimodal imaging and significantly improves the imaging quality and detection sensitivity. Secondly, multimodal physical property imaging technology, including X-ray transmission imaging, X-ray dark field imaging, polarized optical imaging and thermal imaging, is used to comprehensively capture the density distribution, microstructure, molecular chain orientation and thermal properties of aramid fiber tapes, greatly enriching the information dimension of defect detection. The application of multiscale image analysis technology effectively extracts image features at different scales and enhances the recognition ability of tiny defects and large-scale structural anomalies. In the process of potential defect analysis, the defect area is accurately located and described through adaptive threshold segmentation and regional growth algorithm, which improves the accuracy and reliability of defect detection. The physical properties of the potential defect area are quantitatively analyzed to obtain defect physical descriptors including density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics, which not only realizes the qualitative identification of defects, but also provides quantitative physical property evaluation, laying the foundation for subsequent performance prediction. Finally, by extracting multi-dimensional features and assigning weights to the physical descriptors of defects, combined with the preset material performance standards, the quality scores and physical performance parameters of the aramid fiber tapes were obtained, achieving a seamless connection from defect detection to performance evaluation. This comprehensive evaluation method not only provides a more objective and comprehensive quality evaluation, but also provides precise guidance for the optimization of the production process. The generation of defect detection reports integrates defect distribution maps and comprehensive quality evaluation results, providing an intuitive and detailed decision-making basis for production management and quality control. Through multimodal data fusion, multi-scale analysis and physical property quantification, the comprehensiveness, accuracy and reliability of aramid fiber tape defect detection have been significantly improved.

[0059] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An X-ray based online detection method for aramid fiber tape defects, characterized in that: The X-ray-based online defect detection method for aramid fiber tapes includes: analyzing the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters; based on the X-ray imaging parameters, performing multimodal physical property imaging on the preset aramid fiber tapes to obtain a multimodal data set including X-ray transmission images, X-ray dark field images, polarized optical images, and thermal images; performing multiscale image analysis on the multimodal data set to obtain a physical property image and a material parameter distribution map; performing potential defect analysis on the physical property image and the material parameter distribution map to obtain spatial distribution data of potential defect areas; performing physical property quantitative analysis on the spatial distribution data of the potential defect areas to obtain defect physical descriptors, and the defect physical descriptors The characteristics include density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics, including: performing regional segmentation on the spatial distribution data of the potential defect area, extracting the contour information of each defect area, and calculating the area, perimeter and shape factor of each defect area to obtain defect geometric characteristics; calculating the average density and density variance in each defect area according to the grayscale information corresponding to each defect area in the X-ray transmission image, and comparing with the surrounding normal area to obtain density anomaly characteristics; analyzing the molecular chain orientation distribution in the defect area through the image parameter information corresponding to each defect area in the X-ray dark field image and the polarized optical image, calculating the orientation entropy and orientation consistency, and obtaining the degree of molecular chain orientation disorder; based on the grayscale information corresponding to each defect area in the thermal image, calculating the orientation entropy and orientation consistency of the molecular chain in the defect area; ... the orientation consistency of the molecular chain in the defect area; calculating the orientation entropy and the orientation consistency of the molecular chain in the defect area; calculating The data of each defect area are used to calculate the temperature distribution characteristics of each defect area, including the average temperature deviation, the temperature gradient and the thermal diffusion coefficient, to obtain the thermal anomaly characteristics; the defect geometric characteristics, the density anomaly characteristics, the molecular chain orientation disorder degree and the thermal anomaly characteristics are normalized to obtain the defect physical descriptor; according to the preset material performance standards, the multi-dimensional feature extraction and weight distribution of the defect physical descriptor are performed to obtain the quality score and physical performance parameters of the aramid fiber belt, and a defect detection report is generated according to the quality score and the physical performance parameters, including: feature normalization processing of the defect physical descriptor to obtain a normalized feature vector; according to the material performance standards, the normalized feature vector is extracted and weighted. The weight coefficient is assigned to each characteristic component in the quantity to obtain a weighted characteristic vector; the weighted characteristic vector is nonlinearly mapped to convert the multidimensional characteristics into a one-dimensional score to obtain the quality score of the aramid fiber tape; based on the density abnormality characteristics, the degree of molecular chain orientation disorder and the thermal abnormality characteristics in the defect physical descriptor, the mechanical properties, thermal stability and durability of the aramid fiber tape are estimated through regression analysis to obtain physical performance parameters; the aramid fiber tape is graded according to the quality score, and a comprehensive quality assessment result is generated according to the physical performance parameters; a defect distribution map is generated based on the spatial distribution data of the potential defect area, and the defect distribution map is data fused with the comprehensive quality assessment result to obtain the defect detection report.

2. The X-ray based online defect detection method for aramid fiber tape according to claim 1, characterized in that: The analysis of the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters includes: numerically calculating the molecular structure and density distribution of the aramid fibers to determine the attenuation characteristics of the aramid fibers, and performing Monte Carlo simulation on the attenuation characteristics to generate propagation path data of X-rays in the aramid fibers; optimizing the X-ray energy spectrum range according to the propagation path data to determine the target X-ray energy spectrum range, and discretizing the target X-ray energy spectrum range to form a discrete energy spectrum point set; parameterizing the X-ray tube voltage and current according to the discrete energy spectrum point set to determine the initial X-ray source parameters; number, and perform scattered radiation impact assessment on the initial X-ray source parameters to calculate scattered radiation correction factors; correct the initial X-ray source parameters according to the scattered radiation correction factors to determine corrected X-ray source parameters, and perform matching analysis on the corrected X-ray source parameters and detector response characteristics to set detector parameters; calculate the geometric layout of the X-ray imaging system according to the corrected X-ray source parameters and the detector parameters to determine the spatial position relationship of source-sample-detector, and perform a balance calculation of spatial resolution and penetration ability on the spatial position relationship to generate the X-ray imaging parameters.

3. The X-ray based online defect detection method for aramid fiber tape according to claim 1, characterized in that: The method of performing multimodal physical property imaging on the preset aramid fiber belt based on the X-ray imaging parameters to obtain a multimodal data set including an X-ray transmission image, an X-ray dark field image, a polarization optical image, and a thermal image includes: decomposing the X-ray imaging parameters, extracting X-ray source parameters and detector parameters, and setting the voltage and current of the X-ray source according to the X-ray source parameters; adjusting the response characteristics of the X-ray detector according to the detector parameters, and performing X-ray transmission scanning on the preset aramid fiber belt to generate an X-ray transmission image; inserting a X-ray source and an X-ray detector between the X-ray source and the X-ray detector; A phase grating is used to perform X-ray dark field imaging, and the X-ray dark field image is reconstructed by a phase stepping method; the polarization direction of the polarized light source is adjusted, a polarization optical scan is performed on the preset aramid fiber belt, and a synthetic polarization optical image is calculated by Stokes parameter calculation; the infrared thermal imager is started to perform thermal imaging scan on the preset aramid fiber belt, and the temperature calibration and noise suppression are performed on the collected thermal image data to generate a thermal image; the X-ray transmission image, X-ray dark field image, polarization optical image and thermal image are spatially registered, and the multimodal images are fused based on the registration results to obtain a multimodal data set.

4. The X-ray based online defect detection method for aramid fiber tape according to claim 3, characterized in that: The method of performing multi-scale image analysis on a multi-modal data set to obtain a physical property image and a material parameter distribution map includes: performing wavelet transform on the multi-modal data set to decompose it into sub-band images of different scales, and performing noise suppression on each sub-band image to obtain a denoised multi-scale image; performing edge detection on the denoised multi-scale image to extract fiber structure features, and enhancing fiber boundaries through morphological operations to obtain an enhanced structure feature map; performing texture analysis on the enhanced structure feature map to calculate local binary pattern features, and constructing a texture feature descriptor based on grayscale co-occurrence matrix features; calculating material density distribution based on grayscale distribution of X-ray transmission images, and estimating local orientation and crystallinity of materials based on X-ray dark field image information to obtain a preliminary material parameter distribution map; performing polarization degree and polarization direction analysis on polarized optical images to extract molecular chain orientation information, and constructing a comprehensive physical property image based on temperature distribution data of thermal images; and performing data fusion on the preliminary material parameter distribution map and the comprehensive physical property image to obtain a physical property image and a material parameter distribution map.

5. The X-ray based online defect detection method for aramid fiber tape according to claim 1, characterized in that: The method of performing potential defect analysis on the physical property image and the material parameter distribution map to obtain spatial distribution data of the potential defect area includes: performing adaptive threshold segmentation on the physical property image and the material parameter distribution map to obtain abnormal areas, and expanding the boundaries of the abnormal areas by a region growing algorithm to obtain preliminary potential defect areas; performing shape analysis on the preliminary potential defect areas, calculating geometric feature parameters, and screening the preliminary potential defect areas according to preset shape rules to obtain shape-screened potential defect areas; performing feature extraction on the shape-screened potential defect areas in the physical property image, calculating local physical property values, and extracting material parameter values ​​of corresponding areas in the material parameter distribution map; performing statistical analysis on the local physical property values ​​and the material parameter values, calculating the degree of difference from the surrounding normal areas, and obtaining an abnormality index; performing multi-feature fusion on the geometric feature parameters of the shape-screened potential defect areas and the abnormality index, calculating a comprehensive score, and screening according to the comprehensive score to obtain potential defect areas; performing spatial positioning and contour extraction on the potential defect areas, constructing a data structure containing position coordinates and boundary information, and obtaining spatial distribution data of the potential defect areas.

6. An X-ray based online detection device for aramid fiber belt defects, used to implement the X-ray based online detection method for aramid fiber belt defects as described in any one of claims 1 to 5, characterized in that: The X-ray-based online defect detection device for aramid fiber tapes includes: an acquisition module, which is used to analyze the interaction relationship between X-rays and aramid fibers to obtain X-ray imaging parameters; an imaging module, which is used to perform multimodal physical property imaging of a preset aramid fiber tape based on the X-ray imaging parameters to obtain a multimodal data set including an X-ray transmission image, an X-ray dark field image, a polarized optical image, and a thermal image; an analysis module, which is used to perform multiscale image analysis on the multimodal data set to obtain a physical property image and a material parameter distribution map; an identification module, which is used to perform potential defect analysis on the physical property image and the material parameter distribution map to obtain spatial distribution data of potential defect areas; and a quantification module, which is used to quantify the potential defect areas. The physical characteristics of the spatial distribution data of the defect area are quantitatively analyzed to obtain a physical descriptor of the defect, which includes density anomaly characteristics, molecular chain orientation disorder degree and thermal anomaly characteristics, including: performing regional segmentation on the spatial distribution data of the potential defect area, extracting the contour information of each defect area, and calculating the area, perimeter and shape factor of each defect area to obtain defect geometric characteristics; calculating the average density and density variance in each defect area according to the grayscale information corresponding to each defect area in the X-ray transmission image, and comparing them with the surrounding normal areas to obtain density anomaly characteristics; analyzing the defect through the image parameter information corresponding to each defect area in the X-ray dark field image and the polarized optical image The molecular chain orientation distribution in the region is calculated, and the orientation entropy and orientation consistency are calculated to obtain the degree of molecular chain orientation disorder; based on the data corresponding to each defect area in the thermal image, the temperature distribution characteristics of each defect area are calculated, including the average temperature deviation, the temperature gradient and the thermal diffusion coefficient, to obtain the thermal anomaly characteristics; the defect geometric characteristics, the density anomaly characteristics, the molecular chain orientation disorder degree and the thermal anomaly characteristics are normalized to obtain the defect physical descriptor; an extraction module is used to perform multi-dimensional feature extraction and weight allocation on the defect physical descriptor according to the preset material performance standard, to obtain the quality score and physical performance parameters of the aramid fiber tape, and generate a defect description based on the quality score and the physical performance parameters. The defect detection report includes: performing feature normalization processing on the defect physical descriptor to obtain a normalized feature vector; assigning a weight coefficient to each feature component in the normalized feature vector according to the material performance standard to obtain a weighted feature vector; performing nonlinear mapping on the weighted feature vector to convert the multidimensional feature into a one-dimensional score to obtain a quality score of the aramid fiber tape; based on the density abnormality feature, the degree of molecular chain orientation disorder and the thermal abnormality feature in the defect physical descriptor, estimating the mechanical properties, thermal stability and durability of the aramid fiber tape through regression analysis to obtain physical performance parameters; grading the aramid fiber tape according to the quality score, and generating a comprehensive quality assessment result according to the physical performance parameters;Based on the spatial distribution data of the potential defect area, a defect distribution map is generated, and the defect distribution map is fused with the comprehensive quality assessment result to obtain the defect detection report. ;

Images