Intelligent detection method and system for adhesive force of oil coating of skylight aluminum rail
Through intelligent detection methods and systems, combined with confocal laser scanning and multimodal optical detection, a 3D topology map of the aluminum rail surface is generated and the original detection data is obtained, structural alignment and defect identification are solved, and the problems of low adhesion detection efficiency and difficult to trace the deviation links of the sunroof aluminum rail oil coating in the existing technology are solved, efficient adhesion detection and production deviation positioning are achieved, production process is optimized and rework costs are reduced.
Patent Information
- Application Number
- CN202510384643.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-28
- Publication Date
- 2025-06-10
AI Technical Summary
In the prior art, the adhesion detection efficiency of the sunroof aluminum rail oil coating is low, and it is difficult to trace the deviation link, which affects the quality control efficiency and increases the rework cost.
Using intelligent detection methods and systems, a 3D topology map of the aluminum rail surface is generated through a confocal laser scanner, and the original detection data is obtained in combination with multimodal optical detection, structural alignment and defect identification are performed, grid-based adhesion hierarchical detection is performed, defect distribution and adhesion defect distribution are analyzed in fusion, production deviation node sequence is positioned, and subsequent processing operation and maintenance intervention is carried out.
The efficiency of adhesion detection of the sunroof aluminum rail oil coating is improved, and the production deviation links can be accurately positioned, the production process can be optimized, the rework cost can be reduced, and the accuracy and efficiency of quality control can be improved.
Smart Images

Figure CN120121528A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of coating detection, and particularly to an intelligent detection method and system for the adhesion of fuel coatings on skylight aluminum rails. Background Art
[0002] During the manufacturing process of automotive skylights, the surface of aluminum rails often needs to be treated with fuel coatings to enhance its corrosion resistance and extend its service life. However, the existing detection methods for the adhesion of fuel coatings on skylight aluminum rails usually adopt manual or partial sampling methods, which are difficult to reflect the actual adhesion status of the coatings in real time and comprehensively. In addition, deviations in coating quality may occur in multiple links such as spraying, curing, and cleaning in the production process. If these deviation links cannot be detected and traced in a timely and accurate manner, it will not only affect the efficiency and accuracy of quality control, but also increase the cost of subsequent repair or rework. Summary of the Invention
[0003] This application provides an intelligent detection method and system for the adhesion of fuel coatings on skylight aluminum rails, which solves the technical problems of low detection efficiency of the adhesion of fuel coatings on skylight aluminum rails in the prior art and difficulty in tracing deviation links.
[0004] In the first aspect of this application, an intelligent detection method for the adhesion of fuel coatings on skylight aluminum rails is provided. The method includes: Starting a confocal laser scanner according to the structural characteristics of the aluminum rail, comprehensively scanning the skylight aluminum rail to be detected clamped directionally to generate a 3D topology map of the aluminum rail surface; obtaining optical original detection data by performing multimodal optical detection on the skylight aluminum rail to be detected; after structurally aligning the 3D topology map of the aluminum rail surface and the optical original detection data based on feature point matching, obtaining the aluminum rail defect distribution topology through defect identification and marking; referring to the aluminum rail defect distribution topology, performing a grid-based fuel coating adhesion level detection on the skylight aluminum rail to be detected, and outputting a multi-level adhesion defect distribution; fusing and analyzing the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect characteristics; performing production traceback according to the M defect regions of the M comprehensive adhesion defect characteristics to locate the production deviation node sequence; and performing processing and operation and maintenance intervention on the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence.
[0005] In the second aspect of this application, an intelligent detection system for the adhesion of fuel coatings on skylight aluminum rails is provided. The system includes: A scanning module, configured to start a confocal laser scanner according to the structural characteristics of the aluminum rail, comprehensively scan the to-be-detected skylight aluminum rail clamped directionally, and generate a 3D topology map of the aluminum rail surface; a first detection module, configured to obtain optical original detection data by performing multi-modal optical detection on the to-be-detected skylight aluminum rail; a defect identification module, configured to obtain the aluminum rail defect distribution topology through defect identification marks after structurally aligning the 3D topology map of the aluminum rail surface and the optical original detection data based on feature point matching; a second detection module, configured to perform grid-based oil coating adhesion level detection on the to-be-detected skylight aluminum rail with reference to the aluminum rail defect distribution topology, and output a multi-level adhesion defect distribution; an analysis module, configured to fuse and analyze the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect features; a backtracking module, configured to perform production backtracking according to the M defect regions of the M comprehensive adhesion defect features to locate the production deviation node sequence; an operation and maintenance module, configured to perform processing operation and maintenance intervention on the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence.
[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, start a confocal laser scanner according to the structural characteristics of the aluminum rail, comprehensively scan the to-be-detected skylight aluminum rail clamped directionally, and generate a 3D topology map of the aluminum rail surface. Then, obtain optical original detection data by performing multi-modal optical detection on the to-be-detected skylight aluminum rail. After structurally aligning the 3D topology map of the aluminum rail surface and the optical original detection data based on feature point matching, obtain the aluminum rail defect distribution topology through defect identification marks. Next, perform grid-based oil coating adhesion level detection on the to-be-detected skylight aluminum rail with reference to the aluminum rail defect distribution topology, and output a multi-level adhesion defect distribution; fuse and analyze the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect features. Then, perform production backtracking according to the M defect regions of the M comprehensive adhesion defect features to locate the production deviation node sequence. Finally, perform processing operation and maintenance intervention on the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence. This solves the technical problems in the prior art that the detection efficiency of the oil coating adhesion of the skylight aluminum rail is relatively low and it is difficult to trace the deviation link, and by identifying adhesion defects at different levels to locate the production deviation link, the technical effect of improving the detection efficiency is achieved. Description of the Drawings
[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0008] Figure 1 Schematic flow chart of an intelligent detection method for the adhesion of the oil coating on the skylight aluminum rail provided by the embodiment of the present application; Figure 2 Schematic structural diagram of an intelligent detection system for the adhesion of the oil coating on the skylight aluminum rail provided by the embodiment of the present application.
[0009] Explanation of reference numerals: Scanning module 11, First detection module 12, Defect identification module 13, Second detection module 14, Analysis module 15, Backtracking module 16, Operation and maintenance module 17. Detailed implementation manners
[0010] By providing an intelligent detection method and system for the adhesion of the oil coating on the skylight aluminum rail, the present application solves the technical problems in the prior art that the detection efficiency of the adhesion of the oil coating on the skylight aluminum rail is relatively low and it is difficult to trace the deviation links.
[0011] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.
[0012] It should be noted that the terms "include" and "have" are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.
[0013] Embodiment 1, as Figure 1 shown, the present application provides an intelligent detection method for the adhesion of the oil coating on the skylight aluminum rail, wherein the method includes: Start a confocal laser scanner according to the structural characteristics of the aluminum rail, and comprehensively scan the skylight aluminum rail to be detected clamped directionally to generate a 3D topological map of the aluminum rail surface.
[0014] The structural characteristics of the aluminum rail refer to the geometric shape characteristics presented by the skylight aluminum rail in terms of length, curved surface, concave and convex parts, installation hole positions, and local stiffeners.
[0015] According to the geometric structure characteristics of the skylight aluminum rail, first select a suitable clamping device to ensure the stability and standardized placement of the aluminum rail during the scanning process. Use an adjustable mechanical fixture to fix the skylight aluminum rail on the detection platform, ensuring that the clamping position does not affect the scanning of key surface areas, and orient it according to the standard reference plane of the aluminum rail (such as the assembly reference or machining reference) to ensure the consistency of all samples to be detected. At the same time, adopt a multi-axis adjustable support structure to adjust the clamping angle according to the length and curvature of the aluminum rail to ensure that the scanning range covers the entire surface. Subsequently, according to the material characteristics of the aluminum rail (such as the reflectivity of the aluminum alloy surface) and the scanning accuracy requirements, select a suitable confocal laser scanner and set relevant parameters, including the selection of the light source wavelength, which is suitable for high-reflection metal surfaces (such as 405nm or 532nm), and the XY-direction step accuracy is usually set in the range of 10μm - 50μm to ensure the high-precision capture of surface features. In addition, to adapt to the possible curved surface areas of the aluminum rail, adjust the focal depth range of the confocal system to ensure that the focus always falls on the surface during the scanning process, and select a suitable scanning mode according to the requirements, such as line scanning or surface scanning, and set a reasonable scanning speed to optimize the data acquisition efficiency. When officially performing 3D scanning, calculate the optimal scanning path according to the geometric shape of the aluminum rail to ensure no blind area coverage, and adopt a multiple overlapping scanning method to improve data integrity. The confocal laser scanner gradually obtains the surface data of the aluminum rail during the scanning process, and each sampling point records the XYZ space coordinates and surface reflection intensity information, thus forming high-density point cloud data. Subsequently, preprocess the collected original point cloud data, including denoising and filtering operations, to remove external environmental interference signals, such as dust or reflection noise in non-detection areas, and further improve the data quality. After completing the data cleaning, use the Delaunay triangulation method based on the point cloud data to generate a high-precision 3D surface topology model and store it in a standard format (such as STL or OBJ files), and finally obtain a complete 3D topology map of the aluminum rail surface, providing basic data support for subsequent adhesion detection and defect analysis.
[0016] By performing multi-modal optical detection on the skylight aluminum rail to be detected, optical original detection data is obtained.
[0017] Perform multimodal optical inspection on the skylight aluminum rail to be inspected. First, select appropriate optical inspection equipment according to the inspection accuracy requirements and the optical characteristics of the aluminum rail oil coating, including a near-infrared camera, an ultraviolet fluorescence imaging device, and a necessary light source adjustment system. During the near-infrared inspection, based on the spectral absorption characteristics of the aluminum rail coating material, select an appropriate near-infrared band, and call the target near-infrared camera that meets the band setting to align with the surface of the aluminum rail and take pictures and scans to obtain the first original inspection data. During the ultraviolet fluorescence imaging, adjust the light source intensity to evenly irradiate the surface of the aluminum rail oil coating with ultraviolet light, excite the fluorescence signal, and scan it with a high-sensitivity fluorescence imaging device to obtain the second original inspection data. Subsequently, spatially align and store the first original inspection data and the second original inspection data to ensure that the data is analyzed subsequently under the same coordinate system, and finally form complete optical original inspection data to provide basic data support for adhesion defect detection.
[0018] Furthermore, by performing multimodal optical inspection on the skylight aluminum rail to be inspected, optical original inspection data is obtained. The method includes: After selecting the band of the near-infrared camera based on the inspection accuracy, call the target near-infrared camera that meets the target band setting; align the target near-infrared camera with the oil coating surface of the skylight aluminum rail to be inspected and take pictures and scans to output the first original inspection data; after adjusting the light source intensity, run the ultraviolet fluorescence imaging device to irradiate and scan the oil coating surface of the skylight aluminum rail to be inspected to output the second original inspection data; spatially align and store the first original inspection data and the second original inspection data as the optical original inspection data.
[0019] First, determine the near-infrared band suitable for oil coating inspection according to the inspection accuracy requirements and the spectral response characteristics of the aluminum rail oil coating, and call the target near-infrared camera that meets the band setting. Subsequently, accurately align the target near-infrared camera with the oil coating surface of the skylight aluminum rail to be inspected, and perform step-by-step picture taking and scanning along a preset scanning path to capture the reflection or absorption characteristics of the coating in the near-infrared band, and output the first original inspection data. After completing the near-infrared inspection, dynamically adjust the intensity of the inspection environment light source to ensure uniform irradiation of the aluminum rail surface with ultraviolet light, and run the ultraviolet fluorescence imaging device to perform fluorescence excitation scanning on the oil coating surface of the aluminum rail to capture the fluorescence response characteristics and output the second original inspection data. To ensure the consistency of multimodal optical inspection data, spatially align the obtained first original inspection data and the second original inspection data, and perform coordinate registration and storage based on the aluminum rail structure characteristics so that they can be used for subsequent analysis under the same reference coordinate system, and finally form complete optical original inspection data to provide multi-dimensional information support for defect detection of the adhesion of the oil coating.
[0020] Furthermore, the method further includes: Preset the heating distance interval scale; with the heating distance interval scale as a constraint, randomly select multiple heating nodes on the skylight aluminum rail to be detected; during the multi-node local heating of the skylight aluminum rail to be detected according to the multiple heating nodes, use a thermal imager to record the heat diffusion and obtain the third original detection data; after spatially registering the third original detection data according to the multiple heating nodes, use the third original detection data to update the optical original detection data.
[0021] On the basis of multi-modal optical detection, in order to further enhance the detection ability of the adhesion defects of the oil coating, local heating and thermal imaging analysis are also required. First, according to the geometric dimensions of the skylight aluminum rail and the thermal conduction characteristics of the coating material, preset a reasonable heating distance interval scale to ensure the uniformity of heat diffusion and the detection accuracy. On this basis, with the heating distance interval scale as a constraint, randomly select multiple heating nodes on the surface of the skylight aluminum rail to be detected to ensure coverage of different regions of the coating. Subsequently, for the selected heating nodes, use a controllable heat source (such as an infrared laser or a hot air heating device) to locally heat the surface of the skylight aluminum rail, and use a high-precision thermal imager to record the temperature change and heat diffusion during the heating process, so as to obtain the third original detection data. This thermal imaging data can reflect the thermal conductivity of the oil coating, the uniformity of heat diffusion and the influence of potential defects (such as microcracks or peeling). After completing the heat diffusion record, perform spatial registration on the obtained third original detection data to make its coordinate reference consistent with the near-infrared and ultraviolet fluorescence detection data, ensuring the accuracy of data fusion. Finally, use the third original detection data to update the optical original detection data, so as to further improve the recognition ability of the coating adhesion defects on the basis of the original near-infrared and ultraviolet fluorescence detections, and provide richer data support for subsequent comprehensive analysis and defect classification.
[0022] After structurally aligning the 3D topology map of the aluminum rail surface and the optical original detection data based on feature point matching, obtain the aluminum rail defect distribution topology through defect identification marks.
[0023] First, construct the spatial coordinate system of the aluminum rail. Based on the 3D topological map, extract key structural feature points, such as curvature mutation points, bolt hole edges, and guide rail cross-section feature points, to establish a stable reference coordinate set. Subsequently, preprocess the optical original detection data (including near-infrared, ultraviolet fluorescence, and thermal imaging detection data). Using edge detection, feature point extraction, and image registration algorithms, locate the feature points corresponding to the 3D topological map in the optical detection data. Utilize an optimized feature point matching algorithm (such as SIFT or ORB) to calculate the spatial transformation matrix between the optical data and the 3D topological map, and through iterative optimization, achieve high-precision structural alignment between the two, ensuring that the optical detection data can be accurately mapped into the coordinate system of the 3D topological map. After alignment, based on the fused multi-modal data, use a deep learning defect detection model or an image processing-based defect recognition algorithm to analyze the defect features of the optical detection data, identify the defect types of the oil coating, including bulges, abnormal thickness, micro-cracks, and peeling areas, etc., and project the detected defect positions onto the 3D topological map, binding the defect identification to the aluminum rail spatial coordinate system, and finally generating a complete aluminum rail defect distribution topology, providing accurate defect information for subsequent adhesion evaluation and production process optimization.
[0024] Furthermore, after performing structural alignment on the 3D topological map of the aluminum rail surface and the optical original detection data based on feature point matching, obtain the aluminum rail defect distribution topology through defect identification and marking. The method includes: Establish an aluminum rail spatial coordinate system based on the 3D topological map of the aluminum rail surface; use the aluminum rail spatial coordinate system to perform feature point registration iteration on the optical original detection data and the 3D topological map of the aluminum rail surface to generate a spatially unified reference data pool; perform multi-modal image segmentation processing on the spatially unified reference data pool to separate the oil coating feature area; through multi-dimensional defect feature recognition of the oil coating feature area and binding the defect position identification to the aluminum rail spatial coordinate system, generate the aluminum rail defect distribution topology.
[0025] Specifically, based on the 3D topology map of the aluminum rail surface, a spatial coordinate system of the aluminum rail is established. Taking the geometric reference plane of the aluminum rail and key structural feature points (such as bolt holes, edge curvature mutation points, etc.) as references, a three-dimensional coordinate mapping relationship is constructed. Subsequently, using this spatial coordinate system of the aluminum rail, feature point registration is performed on the optical original detection data (including near-infrared, ultraviolet fluorescence, and thermal imaging data), and the data alignment accuracy is adjusted through iterative optimization algorithms (such as ICP iterative closest point matching or RANSAC robust fitting). Finally, a unified spatial reference data pool is generated, enabling the fusion analysis of all detection data in a unified coordinate system. On this basis, multi-modal image segmentation processing is performed on the unified spatial reference data pool. Combining deep learning segmentation networks (such as U-Net, Mask R-CNN) or traditional image processing methods based on edge detection, the characteristic region of the oil coating is extracted, and the interference of non-detection regions is excluded. Subsequently, multi-dimensional defect feature recognition is carried out within the separated characteristic region of the oil coating. For different defect types (such as bulges, thickness anomalies, micro-cracks, peeling boundaries, etc.), algorithms such as curvature analysis, fluorescence gradient detection, and thermal conduction anomaly analysis are used for automatic detection, and the identified defect positions are bound to the spatial coordinate system of the aluminum rail, making the defect information have spatial consistency and traceability. Finally, a complete topological map of the aluminum rail defect distribution is generated, providing accurate defect position information and characteristic data for subsequent adhesion defect assessment and production optimization.
[0026] Furthermore, by performing multi-dimensional defect feature recognition on the characteristic region of the oil coating and binding the defect position identifiers to the spatial coordinate system of the aluminum rail to generate the topological map of the aluminum rail defect distribution, the method includes: Adopting a preset curvature mutation threshold, performing curvature mutation detection on the characteristic region of the oil coating to identify the distribution of coating bulge defects; traversing the characteristic region of the oil coating with a preset absorption rate mutation threshold to identify the distribution of thickness anomaly defects; detecting and positioning the distribution of micro-crack defects in the characteristic region of the oil coating according to the second derivative extreme value of the fluorescence intensity gradient data; performing coating thermal conduction anomaly analysis on the characteristic region of the oil coating based on the variation of the thermal conduction path to locate the distribution of the coating peeling boundary; and based on the distribution of the coating bulge defects, thickness anomaly defects, micro-crack defects, and coating peeling boundary, binding the defect position identifiers to the spatial coordinate system of the aluminum rail to generate the topological map of the aluminum rail defect distribution.
[0027] Specifically, a preset curvature mutation threshold is adopted to conduct local curvature analysis on the characteristic region of the oil coating. By calculating the change rate of the surface normal vector or the Gaussian curvature gradient, the abnormal undulations of the coating surface topography are detected, and the spatial distribution of bulge defects is identified. Among them, the characteristic region of the oil coating includes the original data of multi-source detection data falling into this region (3D topology map of the aluminum rail surface, infrared scanning, ultraviolet fluorescence equipment data, thermal imaging data). Subsequently, based on the spectral absorption characteristics in the optical detection data, a preset absorption rate mutation threshold is used to globally traverse the oil coating region. The change in the absorption rate of each detection point is calculated using the spectral attenuation model to accurately identify the region with abnormal thickness and locate its distribution characteristics. At the same time, combined with the ultraviolet fluorescence detection data, the fluorescence intensity gradient is analyzed, and the extreme points of its second derivative are calculated to detect the micro-crack defects inside the coating and ensure the accurate positioning of the crack position and trend. In addition, using the thermal imaging data, based on the principle of variation of the heat conduction path (that is, calculating the difference in the temperature attenuation rate of adjacent heating nodes in the thermal imaging data), a coating heat diffusion model is constructed. By analyzing the local anomalies in the heat flow direction and conduction rate, the coating peeling boundary region is identified and located. Finally, the spatial coordinate information of the above four types of defects (bulge defects, abnormal thickness, micro-crack defects, and peeling boundary) is bound to the aluminum rail spatial coordinate system, making all defect data have spatial consistency, and generating a complete aluminum rail defect distribution topology, providing accurate defect position information and characteristic data for subsequent adhesion evaluation, quality analysis, and production optimization.
[0028] Furthermore, the curvature mutation threshold is set based on the local topography standard deviation of the 3D topology map of the aluminum rail surface.
[0029] When conducting curvature mutation detection on the characteristic region of the oil coating, to ensure the accurate identification of bulge defects, the curvature mutation threshold is set based on the local topography standard deviation of the 3D topology map of the aluminum rail surface. Specifically, first, the normal vector distribution of the local region is extracted from the 3D topology data of the aluminum rail surface, and the Gaussian curvature and mean curvature of each detection point are calculated to characterize the surface geometric features. Subsequently, the sliding window method or the adaptive grid division strategy is adopted to calculate the statistical features of the local curvature within different scale ranges, including the mean value, standard deviation, and skewness coefficient, etc., to evaluate the natural topography change trend of the aluminum rail surface in different regions. Based on these statistical features, the curvature mutation threshold is adaptively set to have a higher sensitivity in the smooth region and be able to effectively distinguish normal structural changes from abnormal bulge defects in the complex geometric region. Finally, during the detection process, when the curvature change of a certain detection point exceeds the set local topography standard deviation threshold, it is determined that there is a coating bulge defect at this point, and its spatial position is bound to the aluminum rail spatial coordinate system, thereby improving the accuracy and robustness of defect identification while reducing the false detection rate and missed detection rate.
[0030] With reference to the topological distribution of aluminum rail defects, perform a grid-based oil coating adhesion level detection on the aluminum rail of the skylight to be detected, and output a multi-level adhesion defect distribution.
[0031] Based on the reference of the topological distribution of aluminum rail defects, perform a grid-based oil coating adhesion level detection on the aluminum rail of the skylight to be detected, so as to accurately identify the adhesion defects in different regions and output a multi-level adhesion defect distribution.
[0032] Furthermore, with reference to the topological distribution of the aluminum rail defects, perform a grid-based oil coating adhesion level detection on the aluminum rail of the skylight, and output a multi-level adhesion defect distribution. The method includes: According to the structural characteristics of the aluminum rail and the topological distribution of aluminum rail defects, perform a detection grid level division to obtain a multi-level detection grid distribution; configure a multi-level coating detection strategy for the multi-level detection grid distribution; use the multi-level coating detection strategy as a detection method constraint, traverse the multi-level detection grid distribution to perform oil coating adhesion detection, and obtain the multi-level adhesion defect distribution.
[0033] Specifically, first, according to the structural characteristics and defect distribution of the aluminum rail, perform a grid level division on the detection area to generate a multi-level detection grid distribution. This process sets a reference grid scale based on the geometric characteristics of the aluminum rail (such as bolt hole spacing, track curvature radius, etc.), and combines the defect density characteristics. Through an adaptive grid refinement method, the local grid size is adjusted, so that a refined grid is formed in the high defect density area, while a larger scale grid is maintained in the low defect density area, to optimize the detection efficiency and accuracy.
[0034] After the grid level division is completed, configure a multi-level coating detection strategy for different detection grids to achieve accurate adhesion evaluation. Coarse grid (≥10mm): Use the thermal detuning analysis method to quickly evaluate the adhesion (detect the vibration frequency shift of the coating excited by a thermal pulse); Medium grid (5 - 10mm): Enable ultrasonic surface wave attenuation detection (measure the acoustic wave energy dissipation rate at the coating-substrate interface); Fine grid (≤5mm): Perform a micro-area scratch test (nano-scratch with a diamond indenter + real-time acoustic emission monitoring).
[0035] During the detection process, use the configured multi-level coating detection strategy as a method constraint, traverse all detection grids, and select corresponding detection means according to the grid level to comprehensively test the oil coating adhesion. After the detection data is normalized, construct a grid-level adhesion feature data set, and fuse the detection results of different levels to generate a complete multi-level adhesion defect distribution. Finally, bind this distribution data to the aluminum rail space coordinate system to form an accurate adhesion evaluation model, providing a basis for subsequent quality optimization, production control, and maintenance decision-making.
[0036] Furthermore, according to the aluminum rail structure characteristics and the topology of the aluminum rail defect distribution, a detection grid level division is performed to obtain a multi-level detection grid distribution. The method includes: Extract the bolt hole spacing from the aluminum rail structure characteristics as the reference grid scale; perform an initial grid division on the aluminum rail of the skylight to be detected based on a preset grid vector to obtain an initial detection grid distribution that conforms to the reference grid scale; extract the aluminum rail defect density field from the aluminum rail defect distribution topology; map the aluminum rail defect density field to the initial detection grid distribution to obtain a defect density grid distribution; perform a secondary grid division on the defect density grid distribution based on the defect density characteristics and the defect type characteristics to obtain the multi-level detection grid distribution.
[0037] In the process of performing the detection grid level division according to the aluminum rail structure characteristics and the topology of the aluminum rail defect distribution, first, extract the bolt hole spacing from the aluminum rail structure characteristics as the reference grid scale to ensure the uniformity and rationality of the grid division. Subsequently, perform an initial grid division on the aluminum rail of the skylight to be detected based on a preset grid vector to generate an initial detection grid distribution that conforms to the reference grid scale, thereby establishing a basic detection area. After the initial grid division is completed, extract the aluminum rail defect density field from the aluminum rail defect distribution topology to describe the distribution intensity of defects in different regions, and map the defect density field to the initial detection grid distribution to obtain a defect density grid distribution associated with the defect distribution. For this defect density grid distribution, further perform a secondary grid division based on the defect density characteristics and the defect type characteristics, adaptively refine the grid in areas with higher defect density, while maintaining the original grid scale in areas with lower defect density to form a multi-level detection grid distribution.
[0038] When calculating the defect type characteristics, first determine the single defect density, that is, for a specific type of defect (such as bulge, thickness anomaly, microcrack, etc.), calculate the defect distribution intensity per unit area in the detection area. Then, calculate the grid defect density, that is, the comprehensive distribution density of all types of defects in the current detection grid area. Next, based on the ratio of the single defect density to the corresponding grid defect density, as the calculation basis for the defect type characteristics. Through different manifestations of this ratio, defects can be divided into three type characteristics, including dominant defects, mixed defects, and minor defects. Among them, when the ratio of the single defect density to the grid defect density is close to 1, it is a dominant defect; when the ratio is in the middle range (for example, 0.3 - 0.7), it is a mixed defect; when the ratio is small (for example, less than 0.3), it is a minor defect.
[0039] Fusion analysis of the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect characteristics.
[0040] By comprehensively analyzing the topology of aluminum rail defects distribution and the multi-level adhesion defects distribution, correlating the data at different detection levels, extracting key feature information, and thus generating M comprehensive adhesion defect features, which provide data support for subsequent production traceability and process optimization.
[0041] Furthermore, by fusing and analyzing the topology of the aluminum rail defects distribution and the multi-level adhesion defects distribution to obtain M comprehensive adhesion defect features, the method includes: According to the aluminum rail structure characteristics, perform symmetric feature permutation update on the topology of the aluminum rail defects distribution to obtain W local aluminum rail defect features; according to the aluminum rail structure characteristics, perform grid-level symmetric feature permutation update on the multi-level adhesion defects distribution to obtain F local adhesion defect features; perform multi-modal feature fusion on the W local aluminum rail defect features and the F local adhesion defect features, and output the M comprehensive adhesion defect features, where W and F are positive integers, and M ≥ F + W.
[0042] Specifically, based on the aluminum rail structure characteristics, perform symmetric feature permutation update on the topology of the aluminum rail defects distribution, that is, when there are defects in a certain area, similar or related defects may also exist in its symmetric area; through symmetric feature permutation update, that is, select the feature of the area with more significant defects to represent the feature of its symmetric area (if there are also defects in the symmetric area, but with a lower degree), thus avoiding repeated calculations and highlighting more critical defect information, and finally obtaining W non-repetitive and representative local aluminum rail defect features. Secondly, according to the aluminum rail structure characteristics, perform grid-level symmetric feature permutation update on the multi-level adhesion defects distribution; since the adhesion defects are obtained through grid detection, therefore, the adhesion performance in each grid may be different; in the process of symmetric feature permutation update, select the feature with the worst adhesion performance (i.e., the most serious defect) in the grid to represent the feature of the entire grid (if the adhesion performance in the symmetric grid is better, no replacement is performed), and then obtain F local adhesion defect features that can accurately reflect the adhesion defects at the grid level. Finally, perform multi-modal feature fusion on the W local aluminum rail defect features and the F local adhesion defect features, and obtain M comprehensive adhesion defect features by superposition, where M ≥ F + W.
[0043] Based on the M defect regions of the M comprehensive adhesion defect features, perform production traceability to locate the production deviation node sequence.
[0044] First, based on M comprehensive adhesion defect features, extract the defect regions in the spatial coordinate system of the aluminum rail and construct a defect distribution matrix; then, map the defect distribution matrix to the production process flow of the aluminum rail, and combine production batch information, processing parameter records, and process control data to establish a defect traceability path; next, based on the spatial distribution characteristics of the defect regions, analyze the corresponding fluctuations in production process parameters, calculate the correlation between the defect regions and changes in production process parameters, and identify the key production links that may cause adhesion defects; further, combine production equipment operation data, historical records of process parameters, and quality inspection data, and use statistical analysis or machine learning methods to extract abnormal processing parameters and locate the production deviation node sequence that affects defect formation.
[0045] Perform processing operation and maintenance intervention in the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence.
[0046] Specifically, based on the production deviation node sequence, establish a production deviation impact model to analyze the impact weights of each deviation node on the adhesion defect of the aluminum rail oil coating; then, according to the impact weights, adjust production process parameters, including but not limited to spraying pressure, coating thickness, baking temperature, curing time, etc., to optimize process stability; at the same time, combine the results of defect backtracking analysis to monitor the status of relevant processing equipment, identify operation deviations of key equipment, and implement equipment calibration, maintenance, or replacement strategies; further, in the subsequent batch production process, establish an online monitoring mechanism and use means such as optical detection, thermal imaging, and fluorescence imaging to perform real-time quality inspection on the aluminum rail oil coating to ensure the stability of the coating adhesion; finally, through closed-loop control of production quality, continuously compare the detection results with the optimized process parameters, dynamically adjust the processing operation and maintenance strategy, improve the overall manufacturing quality of the skylight aluminum rail, reduce the incidence of adhesion defects, and ensure product consistency and reliability.
[0047] In summary, the embodiments of the present application have at least the following technical effects: First, start the confocal laser scanner according to the structural characteristics of the aluminum rail, and comprehensively scan the skylight aluminum rail to be detected that is directionally clamped to generate a 3D topology map of the aluminum rail surface. Next, perform multimodal optical detection on the skylight aluminum rail to be detected to obtain the original optical detection data. After structurally aligning the 3D topology map of the aluminum rail surface and the original optical detection data based on feature point matching, obtain the aluminum rail defect distribution topology through defect identification and marking. Next, refer to the aluminum rail defect distribution topology, perform a grid-based oil coating adhesion level detection on the skylight aluminum rail to be detected, and output a multi-level adhesion defect distribution; fuse and analyze the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect features. Then, perform production traceback according to the M defect regions of the M comprehensive adhesion defect features to locate the production deviation node sequence. Finally, perform processing and operation and maintenance intervention on the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence. This solves the technical problems in the prior art that the detection efficiency of the oil coating adhesion of the skylight aluminum rail is low and it is difficult to trace the deviation link, and by identifying adhesion defects at different levels to locate the production deviation link, the technical effect of improving the detection efficiency is achieved.
[0048] Embodiment 2, based on the same inventive concept as the intelligent detection method for the oil coating adhesion of a skylight aluminum rail in the foregoing embodiment, as Figure 2 shown, the present application provides an intelligent detection system for the oil coating adhesion of a skylight aluminum rail, wherein the system includes: A scanning module 11, configured to start a confocal laser scanner according to the structural characteristics of the aluminum rail, and comprehensively scan the skylight aluminum rail to be detected that is directionally clamped to generate a 3D topology map of the aluminum rail surface; a first detection module 12, configured to obtain the original optical detection data by performing multimodal optical detection on the skylight aluminum rail to be detected; a defect identification module 13, configured to, after structurally aligning the 3D topology map of the aluminum rail surface and the original optical detection data based on feature point matching, obtain the aluminum rail defect distribution topology through defect identification and marking; a second detection module 14, configured to perform a grid-based oil coating adhesion level detection on the skylight aluminum rail to be detected with reference to the aluminum rail defect distribution topology, and output a multi-level adhesion defect distribution; an analysis module 15, configured to fuse and analyze the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect features; a traceback module 16, configured to perform production traceback according to the M defect regions of the M comprehensive adhesion defect features to locate the production deviation node sequence; an operation and maintenance module 17, configured to perform processing and operation and maintenance intervention on the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence.
[0049] Further, the second detection module 14 is configured to execute the following method: According to the structural characteristics of the aluminum rail and the topological distribution of aluminum rail defects, perform hierarchical division of the detection grid to obtain a multi-level detection grid distribution; configure multi-level coating detection strategies for the multi-level detection grid distribution; use the multi-level coating detection strategies as detection method constraints, and traverse the multi-level detection grid distribution to perform oil coating adhesion detection to obtain the multi-level adhesion defect distribution.
[0050] Further, the first detection module 12 is used to execute the following method: After selecting the wavelength band of the near-infrared camera based on the detection accuracy, call the target near-infrared camera that meets the target wavelength band setting; aim the target near-infrared camera at the oil coating surface of the skylight aluminum rail to be detected for photographing and scanning, and output the first raw detection data; after adjusting the light source intensity, operate the ultraviolet fluorescence imaging device to irradiate and scan the oil coating surface of the skylight aluminum rail to be detected, and output the second raw detection data; perform spatial alignment and storage on the first raw detection data and the second raw detection data as the optical raw detection data.
[0051] Further, the first detection module 12 is used to execute the following method: Preset the heating distance interval scale; use the heating distance interval scale as a constraint, randomly select multiple heating nodes on the skylight aluminum rail to be detected; during the multi-node local heating of the skylight aluminum rail to be detected according to the multiple heating nodes, record the heat diffusion through a thermal imager to obtain the third raw detection data; after performing spatial registration of the third raw detection data according to the multiple heating nodes, use the third raw detection data to update the optical raw detection data.
[0052] Further, the defect recognition module 13 is used to execute the following method: Establish an aluminum rail space coordinate system based on the 3D topology map of the aluminum rail surface; use the aluminum rail space coordinate system to perform feature point registration iteration on the optical raw detection data and the 3D topology map of the aluminum rail surface to generate a unified data pool of spatial benchmarks; perform multi-modal image segmentation processing on the unified data pool of spatial benchmarks to separate the oil coating feature regions; perform multi-dimensional defect feature recognition on the oil coating feature regions, and bind the defect position identifiers to the aluminum rail space coordinate system to generate the topological distribution of aluminum rail defects.
[0053] Further, the defect recognition module 13 is used to execute the following method: Using a preset curvature mutation threshold, perform curvature mutation detection on the characteristic region of the oil coating to identify the distribution of coating bulge defects; traverse the characteristic region of the oil coating with a preset absorption rate mutation threshold to identify the distribution of thickness anomaly defects; detect and locate the distribution of microcrack defects in the characteristic region of the oil coating according to the extreme value of the second derivative of the fluorescence intensity gradient data; perform coating heat conduction anomaly analysis on the characteristic region of the oil coating based on the variation of the heat conduction path to locate the distribution of the coating peeling boundary; based on the distribution of the coating bulge defects, thickness anomaly defects, microcrack defects and coating peeling boundary, bind the defect position identifiers to the aluminum rail space coordinate system to generate the aluminum rail defect distribution topology.
[0054] Further, the defect identification module 13 is used to execute the following method: The curvature mutation threshold is set based on the local topography standard deviation of the 3D topology map of the aluminum rail surface.
[0055] Further, the second detection module 14 is used to execute the following method: Extract the bolt hole spacing from the aluminum rail structure features as the reference grid scale; perform initial meshing on the to-be-detected skylight aluminum rail based on a preset meshing vector to obtain an initial detection grid distribution that conforms to the reference grid scale; extract the aluminum rail defect density field from the aluminum rail defect distribution topology; map the aluminum rail defect density field to the initial detection grid distribution to obtain the defect density grid distribution; based on the defect density features and defect type features, perform secondary meshing on the defect density grid distribution to obtain the multi-level detection grid distribution.
[0056] Further, the analysis module 15 is used to execute the following method: According to the aluminum rail structure features, perform symmetric feature permutation update on the aluminum rail defect distribution topology to obtain W local aluminum rail defect features; according to the aluminum rail structure features, perform grid-level symmetric feature permutation update on the multi-level adhesion defect distribution to obtain F local adhesion defect features; perform multi-modal feature fusion on the W local aluminum rail defect features and F local adhesion defect features to output the M comprehensive adhesion defect features, where W and F are positive integers and M ≥ F + W.
[0057] It should be noted that the above sequence of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.
[0058] The above are only the preferred embodiments of the present application, and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.
[0059] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications therein.
Claims
1. An intelligent detection method for the adhesion of oil coating on skylight aluminum rails, characterized in that: The method comprises: According to the structural characteristics of the aluminum rail, the confocal laser scanner is started to perform a comprehensive scan of the aluminum rail to be inspected in a directional clamping manner, and a 3D topological map of the aluminum rail surface is generated; Obtaining original optical detection data by performing multimodal optical detection on the aluminum rail of the skylight to be detected; After structural alignment of the aluminum rail surface 3D topology map and the optical original detection data based on feature point matching, the aluminum rail defect distribution topology is obtained through defect recognition identification; Referring to the aluminum rail defect distribution topology, performing gridded oil coating adhesion level detection on the sunroof aluminum rail to be detected, and outputting a multi-level adhesion defect distribution; The aluminum rail defect distribution topology and the multi-level adhesion defect distribution are integrated and analyzed to obtain M comprehensive adhesion defect features; Perform production backtracking according to the M defect areas of the M comprehensive adhesion defect characteristics to locate the production deviation node sequence; The processing and operation and maintenance intervention of the subsequent batch production process of the skylight aluminum rails is carried out according to the production deviation node sequence.
2. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 1, characterized in that: Referring to the aluminum rail defect distribution topology, performing gridded oil coating adhesion level detection on the skylight aluminum rail, and outputting a multi-level adhesion defect distribution, the method includes: According to the aluminum rail structural characteristics and the aluminum rail defect distribution topology, the detection grid is divided into levels to obtain a multi-level detection grid distribution; A multi-level coating detection strategy is configured to distribute the multi-level detection grid; Taking the multi-level coating detection strategy as the detection method constraint, the multi-level detection grid distribution is traversed to perform oil coating adhesion detection to obtain the multi-level adhesion defect distribution.
3. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 2, characterized in that: By performing multimodal optical inspection on the aluminum rail of the skylight to be inspected, optical original inspection data is obtained, and the method includes: After selecting the band of the near-infrared camera based on the detection accuracy, call the target near-infrared camera that meets the target band setting; Aim the target near-infrared camera at the oil coating surface of the skylight aluminum rail to be inspected to take a picture and scan, and output first original inspection data; After the intensity of the light source is adjusted, the ultraviolet fluorescence imaging device is operated to irradiate and scan the oil coating surface of the aluminum rail of the skylight to be inspected, and the second original inspection data is output; The first original detection data and the second original detection data are spatially aligned and stored as the optical original detection data.
4. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 3, characterized in that: The method further comprises: Preset heating distance interval scale; Taking the heating distance interval scale as a constraint, a plurality of heating nodes are randomly selected on the skylight aluminum rail to be tested; During the process of performing multi-node local heating on the aluminum rail of the skylight to be inspected according to the multiple heating nodes, heat diffusion is recorded by a thermal imager to obtain third original inspection data; After spatial registration of the third original detection data is performed according to the multiple heating nodes, the optical original detection data is updated using the third original detection data.
5. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 4, characterized in that: After structural alignment of the aluminum rail surface 3D topology map and the optical original detection data based on feature point matching, the aluminum rail defect distribution topology is obtained through defect recognition identification. The method includes: Establishing an aluminum rail spatial coordinate system based on the aluminum rail surface 3D topological map; Using the aluminum rail spatial coordinate system to iterate feature point registration of the optical original detection data and the aluminum rail surface 3D topological map to generate a spatial reference unified data pool; Performing multimodal image segmentation processing on the spatial reference unified data pool to separate the oil coating feature area; The defect distribution topology of the aluminum rail is generated by performing multi-dimensional defect feature recognition on the characteristic area of the oil coating and binding the defect position identifier to the aluminum rail space coordinate system.
6. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 5, characterized in that: By performing multi-dimensional defect feature recognition on the oil coating feature area and binding the defect position identifier to the aluminum rail space coordinate system, the aluminum rail defect distribution topology is generated, and the method includes: Using a preset curvature mutation threshold, performing curvature mutation detection on the characteristic area of the oil coating to identify the distribution of coating bulge defects; Using a preset absorption rate mutation threshold to traverse the characteristic area of the oil coating to identify abnormal thickness defect distribution; Detecting and locating microcrack defect distribution in the characteristic area of the oil coating according to the second derivative extreme value of the fluorescence intensity gradient data; Based on the variation of the heat conduction path, an analysis of the thermal conduction anomaly of the characteristic area of the oil coating is performed to locate the distribution of the coating peeling boundary; According to the coating bulge defect distribution, thickness abnormality defect distribution, microcrack defect distribution and coating peeling boundary distribution, the defect position identifier is bound to the aluminum rail space coordinate system to generate the aluminum rail defect distribution topology.
7. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 6, characterized in that: The curvature mutation threshold is set based on the local morphology standard deviation of the 3D topological map of the aluminum rail surface.
8. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 6, characterized in that: According to the aluminum rail structural features and the aluminum rail defect distribution topology, the detection grid is divided into levels to obtain a multi-level detection grid distribution. The method includes: Extracting bolt hole spacing from the aluminum rail structural features as a reference grid scale; Performing initial gridding of the aluminum rail of the skylight to be inspected based on a preset gridding vector to obtain an initial inspection grid distribution that meets the reference grid scale; Extracting the aluminum rail defect density field from the aluminum rail defect distribution topology; Mapping the aluminum rail defect density field to the initial detection grid distribution, defect density grid distribution; Based on the defect density characteristics and defect type characteristics, the defect density grid distribution is subjected to secondary gridding to obtain the multi-level detection grid distribution.
9. The intelligent detection method for the adhesion of oil coating on skylight aluminum rail according to claim 8, characterized in that: The aluminum rail defect distribution topology and the multi-level adhesion defect distribution are integrated and analyzed to obtain M comprehensive adhesion defect features. The method includes: According to the aluminum rail structural characteristics, the aluminum rail defect distribution topology is symmetrically replaced and updated to obtain W local aluminum rail defect characteristics; According to the aluminum rail structural features, the multi-level adhesion defect distribution is updated by grid-level symmetric feature replacement to obtain F local adhesion defect features; Multimodal feature fusion is performed on the W local aluminum rail defect features and the F local adhesion defect features, and the M comprehensive adhesion defect features are output, wherein W and F are positive integers, and M≥F+W.
10. An intelligent detection system for the adhesion of oil coating on skylight aluminum rails, characterized in that: A method for intelligently detecting the adhesion of oil coating on a skylight aluminum rail according to any one of claims 1 to 9, the system comprising: The scanning module is used to start the confocal laser scanner according to the structural characteristics of the aluminum rail, conduct a comprehensive scan of the sunroof aluminum rail to be inspected in a directional clamping manner, and generate a 3D topological map of the aluminum rail surface; A first detection module is used to obtain optical original detection data by performing multimodal optical detection on the aluminum rail of the skylight to be detected; A defect recognition module is used to obtain the defect distribution topology of the aluminum rail through defect recognition identification after structural alignment of the aluminum rail surface 3D topological map and the optical original detection data based on feature point matching; The second detection module is used to refer to the aluminum rail defect distribution topology, perform gridded oil coating adhesion level detection on the sunroof aluminum rail to be detected, and output a multi-level adhesion defect distribution; An analysis module, used for integrating and analyzing the aluminum rail defect distribution topology and the multi-level adhesion defect distribution to obtain M comprehensive adhesion defect features; A backtracking module, used for performing production backtracking according to the M defect areas of the M comprehensive adhesion defect characteristics, and locating a production deviation node sequence; The operation and maintenance module is used to perform processing operation and maintenance intervention in the subsequent batch production process of the skylight aluminum rail according to the production deviation node sequence.
Citation Information
Cited By
Intelligent detection method for metal surface coating defects
CN120314317A
An intelligent detection method for metal surface coating defects
CN120314317B
Metal surface quality detection method and system
CN120668669A
Automatic detection method and equipment for coating test panel
CN120984572A
Automobile part coating quality monitoring system and method
CN121190481A
