Method and system for intelligently detecting structural parameters of special-shaped aluminum plate based on spectral analysis

Abstract

The invention relates to the technical field of metal structure detection, in particular to an intelligent special-shaped aluminum plate structure parameter detection method and system based on spectral analysis, and solves the problems of multi-parameter splitting, poor complex structure adaptability and inaccurate risk assessment in traditional detection. According to the method, data are synchronously collected through multi-spectral imaging and line laser scanning, surface component parameters such as the thickness gradient of an oxide film are inverted, the local curvature is calculated based on three-dimensional point cloud, a spiral encryption or grid sparse scanning track is dynamically switched, differential detection of a complex structure and a plane area is achieved, and the detection precision is improved. And then a space correlation model of the oxidation film thickness and the hole deformation quantity is constructed, a structure failure risk area is accurately recognized by combining a three-level risk judgment rule and cross-regional influence analysis, and the system effectively improves the accuracy and detection efficiency of defect recognition of the special-shaped aluminum plate through cooperation of dynamic scanning control, component deformation correlation and a composite structure decision unit.

Description

Technical Field

[0001] This invention relates to the field of metal structure inspection technology, and more specifically, to an intelligent detection method and system for structural parameters of irregularly shaped aluminum plates based on spectral analysis. Background Technology

[0002] Metal structure inspection technology is an important technology. In high-end manufacturing fields such as aerospace and rail transportation, the inspection of structural parameters of irregular aluminum plates is a core link to ensure the safety and reliability of components by obtaining key parameters such as oxide film thickness and deformation defects. With the widespread application of irregular aluminum plates in complex load environments, accurate identification of their surface defects and internal structural damage is crucial to preventing catastrophic failures.

[0003] However, traditional detection technologies suffer from core problems such as fragmented multi-parameter detection and insufficient adaptability to complex structures. Existing solutions rely solely on spectral analysis or 3D scanning without establishing a spatiotemporal correlation model between spectral reflectance and 3D point clouds. When the oxide film thickness gradient exceeds a preset gradient threshold, it cannot simultaneously capture the deformation of the hole edge, leading to defect location errors exceeding the preset error threshold. Using a fixed scanning trajectory results in insufficient sampling point density in angular areas with local curvature greater than 0.8 radians per millimeter, increasing the missed crack rate. Redundant scanning in planar areas reduces detection efficiency. Furthermore, traditional methods do not construct a correlation model between oxide layer thickness and structural deformation. When irregularly shaped aluminum plates simultaneously suffer from oxidation corrosion and mechanical damage, the risk level cannot be quantified, reducing the accuracy of structural failure early warning. This lack of multi-parameter detection separation and dynamic adaptability ultimately leads to low overall detection efficiency of traditional solutions under complex working conditions, making it difficult to meet the quality control requirements of high-end manufacturing. To address this technical problem, we provide an intelligent detection method and system for structural parameters of irregularly shaped aluminum plates based on spectral analysis. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent detection method and system for structural parameters of irregularly shaped aluminum plates based on spectral analysis, so as to solve the problems mentioned in the background art.

[0005] 1. Because traditional methods perform multi-parameter detection independently without establishing a correlation model between the spectrum and point cloud, the defect localization deviation is large. Therefore, this case study uses multiple sensors to collect data synchronously and build a spatiotemporal correlation model to achieve multi-parameter fusion analysis and improve localization accuracy.

[0006] 2. Since fixed scanning trajectories cannot adapt to complex structures, and sampling in corner areas is insufficient while planar areas are redundant, this case dynamically switches scanning trajectories based on curvature, densifying sampling in complex areas and scanning in sparse planar areas to improve detection efficiency and defect recognition rate.

[0007] To achieve the above objectives, one of the objectives of this invention is to provide an intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis, comprising the following steps:

[0008] S1. A multispectral imaging system and a line laser scanner are synchronously moved along the surface of the irregular aluminum plate to acquire spectral reflectance data and three-dimensional point cloud data. Based on the spectral reflectance data, the surface composition parameters of the irregular aluminum plate, including the oxide film thickness gradient distribution and impurity concentration distribution, are inverted in real time.

[0009] S2. Calculate the local curvature value based on the three-dimensional point cloud data of the current irregular aluminum plate surface. When the local curvature value is greater than 0.8 radians per millimeter, control the scanning device to perform encrypted sampling of the corner and hole areas with a spiral trajectory. When the local curvature value is less than 0.2 radians per millimeter, switch to a grid trajectory to perform sparse sampling of the planar area.

[0010] S3. Map the oxide film thickness gradient distribution on the surface of the irregular aluminum plate to a three-dimensional point cloud space constructed from three-dimensional point cloud data, establish a linear relationship model between the hole edge deformation and the surrounding oxide layer thickness gradient. When the surrounding oxide layer thickness gradient is greater than 15% per millimeter and the hole edge deformation is greater than 0.3 millimeters, the area is determined to be a structural failure risk zone, and a defect coordinate report is generated.

[0011] As a further improvement to this technical solution, the method for calculating the local curvature value in step S2 includes the following sub-steps:

[0012] Based on 3D point cloud data, a normal vector is generated for each measurement point on the surface of the irregular aluminum plate. The normal vector is calculated by fitting the neighborhood plane using the least squares method. A spherical neighborhood space with a radius of 1 mm is divided with the current measurement point as the center, and all point cloud coordinate data in this space are extracted.

[0013] Calculate the spatial angle between the normal vector of each measurement point in the neighborhood plane and the normal vector of the current point. When the standard deviation of the angle exceeds 15 degrees, the region is determined to be a curvature change zone, and the direction of the maximum angle deviation is recorded.

[0014] The standard deviation of the included angle is converted into a curvature value proportionally. The conversion factor is set to increase the curvature value by 1 radian per millimeter for every 10 degrees of the standard deviation of the included angle. An additional curvature compensation factor is superimposed on the edge area of ​​the hole, and the compensation value is 0.2 times the reciprocal of the hole diameter.

[0015] As a further improvement to this technical solution, in step S2, when the local curvature value is greater than 0.8 radians per millimeter, the execution rules for encrypted sampling include:

[0016] Within a 0.5mm range on both sides of the bend line, an Archimedean spiral is used for trajectory planning. The starting point of the spiral is located at the point of maximum curvature, the spiral ring spacing is fixed at 0.1mm, and 36 sampling points with equal angle distribution are set in each ring.

[0017] For the perforated area, the spiral line converges and scans from the outer edge to the center. The convergence step size is dynamically adjusted according to the aperture size. The step size calculation formula is 1 / 50 of the aperture size value and does not exceed 0.3 mm.

[0018] In the range of curvature values ​​of 0.8 to 1.2 radians per millimeter, the scanning speed is linearly reduced to 30% of the reference speed. When the reflectivity difference between adjacent sampling points exceeds 40%, supplementary sampling points are automatically inserted, with the spacing between supplementary points being 1 / 3 of the original trajectory spacing.

[0019] If three consecutive sampling points of single-point spectral reflectance data are invalid, immediately start a local concentric circle trajectory rescan with a diameter of 0.5 mm centered on the current point, with no less than three rescan rings.

[0020] As a further improvement to this technical solution, in step S2, when the local curvature value is less than 0.2 radians per millimeter, the execution rules for sparse sampling include:

[0021] Use a zigzag grid path in the planar area, with the row spacing set to 1 / 20 of the width of the planar area and not less than 1mm. Add an arc transition at the path turning point, with a transition arc radius of 0.2mm.

[0022] When the reflectance fluctuation of 5 consecutive sampling points is less than 5%, the fast skip scan mode is activated, and the skip scan step size is increased to 3 times the normal line spacing. If the difference between adjacent points is greater than 15% during the skip scan, the scan is immediately reversed to the last valid point to resume normal scanning.

[0023] A bidirectional cross-scan is used for the transition area between the plane and the curved surface. The first scan is along the normal direction of the curved surface, and the second scan is along the tangent direction. The data from the two scans are weighted and fused, with the normal direction accounting for 70% of the weight. The curvature of the transition area is 0.1 to 0.2 radians per millimeter.

[0024] As a further improvement to this technical solution, when the curvature value is between 0.2 and 0.8 radians per millimeter, a strategy is implemented by subdividing the curvature value:

[0025] When the curvature value is between 0.2 and 0.5 radians per millimeter, a compressed grid mode is used, the line spacing is reduced to 0.5 mm, and the scanning speed is increased to 150% of the reference value;

[0026] When the radius is 0.5 to 0.8 radians per millimeter, the extended spiral mode is adopted, the spiral ring spacing is increased to 0.3 mm, and the number of sampling points per ring is reduced to 24.

[0027] When the same area simultaneously meets the characteristics of holes and corners, the hole spiral scan is performed first, and the corner area is processed later. If the processing time exceeds 500ms, the corner area is started in fast point scan mode, and the sampling point density is reduced to 50% of the normal value.

[0028] As a further improvement to this technical solution, the method for constructing the linear relationship model in step S3 includes:

[0029] The oxide film thickness gradient data is spatially registered with the three-dimensional point cloud coordinates, and the deformation of the hole edge is normalized, i.e., deformation = measured deformation value - theoretical deformation value, where the theoretical deformation value is obtained by looking up a table based on the ratio of plate thickness to hole diameter.

[0030] Eight detection sites were selected at equal intervals along the circumference of the hole. The average gradient value of each site was taken within a radius of 1 mm. When establishing the linear relationship between deformation and gradient, cubic spline interpolation was performed on abnormal points with gradient values ​​greater than 20% / mm.

[0031] For areas identified as failure risk zones, the deformation growth rate from the three most recent detection data is automatically retrieved. If the growth rate exceeds 5% per month, a warning label is added to the risk report.

[0032] As a further improvement to this technical solution, the determination criteria for the failure risk zone in step S3 are expanded to include:

[0033] Set three levels of judgment conditions:

[0034] Level 1 judgment: When the gradient is greater than 15% / mm and the deformation is greater than 0.3mm, it is directly marked as a red high-risk area;

[0035] Secondary judgment: When the gradient is greater than 20% / mm and the deformation is greater than 0.2mm, it is marked as a yellow warning zone and the manual review process is triggered;

[0036] Level 3 judgment: When the gradient is greater than 25% / mm and the deformation is greater than 0.1mm, it is marked as a blue observation area and monthly tracking detection is started;

[0037] If there are two or more warning zones within 5mm of a high-risk zone, the warning zone will be automatically upgraded to a high-risk zone. An ultrasonic flaw detection verification step will be added to isolated warning zones.

[0038] As a further improvement to this technical solution, the failure determination method for the intersection area of ​​the folding angle and the hole area in step S3 includes:

[0039] When the edge of a hole overlaps with the bend line, hole failure analysis is performed first, and the bend area is processed later. The scanning energy allocation does not exceed 30% of the total. The oxide film gradient is taken as the maximum value at the intersection of the hole center and the bend line. The deformation is calculated by superimposing the weights of 60% for the hole edge and 40% for the bend line. When the gradient attenuation along the bend line direction is greater than 15% / mm, the latent crack risk marker is triggered.

[0040] As a further improvement to this technical solution, the cross-regional association method for the failure risk zone includes:

[0041] The radius of influence of the high-risk area is R = 0.5 × (deformation / 0.3). 2 mm, the warning zone within the influence radius is automatically upgraded to a high-risk zone. When the hole spacing is less than twice the minimum hole diameter, three additional measurement points are inserted on the connecting line. When the gradient of the additional points is greater than 25% of the average value on both sides, it is marked as the hole fatigue zone.

[0042] The second objective of this invention is to provide a system for realizing an intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis, comprising:

[0043] The dynamic scanning control unit integrates the multispectral imaging system and the line laser scanner's collaborative motion module. It drives the scanning trajectory to switch dynamically through the real-time curvature calculation module. When the local curvature is greater than 0.8 rad / mm, it activates spiral encrypted sampling, and when the curvature is less than 0.2 rad / mm, it switches to grid sparse sampling.

[0044] The component deformation correlation analysis unit inverts the oxide film thickness gradient in real time based on spectral reflectance data, maps the gradient distribution to a three-dimensional point cloud through a spatial registration engine, and uses a linear relationship modeling module to simultaneously analyze the deformation of the pore edge and the oxide layer gradient, and outputs a structural failure risk coordinate report.

[0045] The composite structure collaborative decision-making unit configures a priority arbitrator to automatically allocate detection resources to the intersection area of ​​holes and bends, identifies hidden cracks through the stress transmission analysis module, and finally activates the cross-regional correlation engine to conduct collaborative early warning of radiation effects in high-risk areas and hole group failure.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] This invention achieves fusion analysis of spectral reflectance and three-dimensional point cloud by simultaneously acquiring data through multispectral imaging and line laser scanning. It can invert oxide film thickness gradient and impurity concentration distribution in real time, providing multidimensional data support for defect analysis. Based on curvature dynamic adjustment of the scanning trajectory, a spiral trajectory is used for encrypted sampling in corners and hole areas with local curvature greater than 0.8 radians per millimeter to ensure detection accuracy in complex structures. In planar areas, a grid trajectory is switched to sparse sampling to improve detection efficiency. The oxide film thickness gradient is mapped to three-dimensional point cloud space to establish a linear relationship model between hole edge deformation and oxide layer gradient, which can accurately determine structural failure risk areas and generate coordinate reports. The three-level judgment conditions combined with cross-regional correlation analysis can quantify the risk level and warn of potential failure areas, effectively improving the accuracy and reliability of the detection of structural parameters of irregular aluminum plates. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the overall workflow of the present invention;

[0049] Figure 2 This is a schematic diagram of the overall structure of the present invention;

[0050] The meanings of the labels in the diagram are as follows:

[0051] 1. Dynamic scanning control unit; 2. Component deformation correlation analysis unit; 3. Composite structure collaborative decision-making unit. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Please see Figures 1-2 As shown, one of the objectives of this embodiment is to provide an intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis, including the following steps:

[0054] S1. A multispectral imaging system and a line laser scanner are synchronously moved along the surface of the irregular aluminum plate to acquire spectral reflectance data and three-dimensional point cloud data. Based on the spectral reflectance data, the surface composition parameters of the irregular aluminum plate, including the oxide film thickness gradient distribution and impurity concentration distribution, are inverted in real time.

[0055] To achieve spatiotemporal alignment between spectral and 3D data, a multispectral imaging system and a line laser scanner are synchronously moved along the aluminum plate surface via a robotic arm. The system employs a timestamp synchronization mechanism, triggering the laser scanner to acquire one line point cloud data point after each set of spectral data is acquired, ensuring a time deviation of less than 0.5ms. This lays the foundation for subsequent composition inversion. The multispectral imaging system receives reflected light from the aluminum plate surface via a fiber optic spectrometer, acquiring spectral reflectance data in real time. During acquisition, the system automatically subtracts ambient light interference, filters the original spectrum to remove noise, and then uses a baseline correction algorithm to eliminate drift, providing reliable input for oxide film thickness inversion. The processed spectral reflectance data is used to invert the oxide film thickness gradient and impurity concentration using a partial least squares regression model. The model first establishes a spectral composition mapping relationship using standard aluminum plate samples. During inversion, the system inputs real-time spectral data into the model and outputs the oxide film thickness and impurity concentration, generating a gradient distribution cloud map according to spatial location. A line laser scanner emits a laser beam and acquires three-dimensional point cloud data of the aluminum plate surface through triangulation. During acquisition, the system performs statistical outlier filtering on the original point cloud to remove random noise points during the scanning process, and then reduces the amount of data through voxel mesh downsampling while retaining detailed features to ensure the efficiency and accuracy of subsequent curvature calculations.

[0056] S2. Calculate the local curvature value based on the three-dimensional point cloud data of the current irregular aluminum plate surface. When the local curvature value is greater than 0.8 radians per millimeter, control the scanning device to perform encrypted sampling of the corner and hole areas with a spiral trajectory. When the local curvature value is less than 0.2 radians per millimeter, switch to a grid trajectory to perform sparse sampling of the planar area.

[0057] To accurately calculate the local curvature value of the irregular aluminum plate surface and achieve dynamic adjustment of the scanning trajectory, it is necessary to perform normal vector calculation and curvature transformation on the 3D point cloud data. The specific implementation method is as follows:

[0058] The calculation method for local curvature values ​​in step S2 includes the following sub-steps:

[0059] To obtain the normal vector of each measurement point, a spherical neighborhood space with a radius of 1 mm is defined centered on the current measurement point. The coordinate data of all point clouds within this space are extracted. Assume there are n points in the neighborhood with coordinates (x...). i y i , z iFor each point i = 1 to n, the least squares method is used to fit the neighborhood plane. The plane equation is Ax + By + Cz + D = 0. By solving the system of equations to minimize the sum of the squared distances from each point to the plane, the plane coefficients (A, B, C, D) are obtained. The normal vector of the current point is then (A, B, C) and normalized to a unit vector. This method can effectively reflect the orientation of the local surface and provide a reliable basis for subsequent curvature analysis. The spatial angle between the normal vectors of all measurement points in the neighborhood and the normal vector of the current point is calculated, and the standard deviation of all angles is calculated. When the standard deviation of the angle exceeds 15 degrees, the area is determined to be a curvature change zone, and the direction of the maximum angle deviation is recorded, i.e., the direction with the largest angle with the current normal vector. This determination method can accurately identify areas with drastic curvature changes such as bends and holes. The standard deviation of the angle is converted into a curvature value k proportionally. The conversion coefficient is set to increase the curvature value by 1 radian per millimeter for every 10 degrees of standard deviation of the angle. σ represents the standard deviation of the included angle. Additionally, a curvature compensation coefficient is superimposed on the edge region of the hole. The compensation value is 0.2 times the reciprocal of the hole diameter, i.e., the compensation coefficient Δk = 0.2 / r. The final curvature value is k. final =k+Δk, this compensation mechanism can enhance the curvature characteristics of the hole edge, improve the sensitivity of curvature calculation to the defect area, and avoid insufficient scanning density due to insufficient curvature calculation of the hole edge.

[0060] After calculating the local curvature values, for complex areas with curvature values ​​greater than 0.8 radians per millimeter, a denser sampling strategy needs to be implemented to ensure the accuracy of defect detection. The specific implementation method is as follows:

[0061] In step S2, when the local curvature value is greater than 0.8 radians per millimeter, the execution rules for encrypted sampling include:

[0062] Within a 0.5 mm range on both sides of the bend line, an Archimedean spiral is used for trajectory planning. The starting point of the spiral is located at the point of maximum curvature to ensure that the most defect-prone location is scanned first. The spiral ring spacing is fixed at 0.1 mm to ensure the density of sampling points. 36 sampling points are set in each ring at equal angles to fully cover the surface features of the bend area. This scanning method improves the sampling point density and scanning rate in the bend area, and reduces the crack missed rate. For the hole area, the spiral converges from the outer edge to the center. The convergence step size is dynamically adjusted according to the hole diameter. The step size calculation formula is 1 / 50 of the hole diameter value and does not exceed 0.3 mm. This dynamic adjustment mechanism can adapt to holes of different sizes and ensure the sampling accuracy of the hole center area. During scanning, the spiral starts from the edge of the hole and converges towards the center according to the set step size, so that every position on the hole surface can be scanned uniformly. In the curvature range of 0.8 to 1.2 radians per millimeter, the scanning speed is [not specified]. The speed is linearly reduced to 30% of the baseline speed to ensure data acquisition stability. The baseline speed is usually set to 100 mm / s, and the reduced speed is 30 mm / s, allowing the sensor sufficient time to acquire accurate spectral and point cloud data. Simultaneously, when a reflectance difference exceeding 40% is detected between adjacent sampling points, supplementary sampling points are automatically inserted. The spacing between supplementary points is 1 / 3 of the original trajectory spacing, facilitating the effective identification of areas with sudden changes in oxide film thickness and improving the detection accuracy of areas with abnormal reflectance. If three consecutive sampling points of single-point spectral reflectance data are invalid (invalidity refers to reflectance values ​​exceeding the reasonable range of 0-1 or fluctuations exceeding 50%), a local concentric circle trajectory rescan with a diameter of 0.5 mm is immediately initiated, centered on the current point. The rescan rings are no less than three layers, with concentric circles of radii of 0.25 mm, 0.5 mm, and 0.75 mm respectively. Each layer has 36 equiangular sampling points to avoid misjudgments of defects due to invalid data.

[0063] After calculating the local curvature values, for planar regions with curvature values ​​less than 0.2 radians per millimeter, a sparse sampling strategy needs to be implemented to improve detection efficiency while maintaining detection accuracy. The specific implementation method is as follows:

[0064] In step S2, when the local curvature value is less than 0.2 radians per millimeter, the sparse sampling execution rules include:

[0065] To reduce redundant sampling while ensuring coverage of the planar area, a zigzag grid path is used for scanning. The row spacing is set to 1 / 20 of the width of the planar area and not less than 1 mm. For example, when the width of the planar area is 20 mm, the row spacing is 1 mm; when the width is 50 mm, the row spacing is 2.5 mm. A 0.2 mm radius arc is added at the path turning points to avoid speed fluctuations caused by abrupt changes in the scanning trajectory. For example, in a 100×100 mm planar area, a zigzag scan is performed with a row spacing of 5 mm, and the turning points are smoothly transitioned with a 0.2 mm arc to keep the scanning speed stable. This scanning method improves the detection efficiency of sampling point density in the planar area. When the reflectance fluctuation of 5 consecutive sampling points is less than 5%, the surface composition of the area is determined to be uniform, and a fast skip scan mode is activated, increasing the skip scan step size to 3 times the normal row spacing to further improve scanning efficiency. For example, when the normal row spacing is 1 mm, the skip scan step size is 3 mm. If a reflectance difference greater than 15% is detected between adjacent points during the skip scan, the scan immediately reverts to the last point. A valid point is identified and the regular scan resumes. If the reflectance of five consecutive points in a certain scan ranges from 50.5 ± 0.025, the skip scan mode is activated, and the step size becomes 3 mm. When the reflectance of a skip scan point suddenly drops to 0.7, the scan returns to the previous valid point and continues with a 1 mm line spacing. This mechanism improves the scanning speed of uniform planar areas and ensures that areas with abnormal composition are not missed. For the transition area between a plane and a curved surface with a curvature of 0.1 to 0.2 radians per millimeter, a bidirectional cross-scanning strategy is adopted to balance the detection efficiency of the plane and the detail capture of the curved surface. The first scan is performed along the normal direction of the curved surface, and the second scan is performed along the tangent direction. The data from the two scans are fused according to weights. For example, if the normal direction of a point in the transition area is perpendicular to the curved surface and points outward, and the tangent direction is the direction of the edge extension of the curved surface, the first scan is performed along the normal direction to obtain depth information, and the second scan is performed along the tangent direction to obtain contour information. During fusion, the normal direction data accounts for 70% to highlight the surface undulation features, effectively avoiding the problem of feature loss in the transition area caused by traditional single-direction scanning.

[0066] After calculating the local curvature values, for the medium curvature region with curvature values ​​between 0.2 and 0.8 radians per millimeter, a differentiated scanning strategy needs to be implemented based on curvature subdivision to balance detection efficiency and accuracy. The specific implementation method is as follows:

[0067] When the curvature value is between 0.2 and 0.8 radians per millimeter, the strategy is implemented by subdividing the curvature value:

[0068] When the curvature value is between 0.2 and 0.5 radians per millimeter, it indicates that the surface of the region has some undulations but not sharp transitions. A compressed grid mode is used to improve scanning efficiency, with the line spacing reduced to 0.5 millimeters to ensure sampling density on undulating surfaces. Simultaneously, the scanning speed is increased to 150% of the baseline value. For example, on a wavy surface with a curvature of 0.3 radians per millimeter, a zigzag scan is performed with a line spacing of 0.5 millimeters at a speed of 150 millimeters per second, ensuring that defects on moderately undulating surfaces, such as localized oxide film thickening, can be effectively identified. For regions with a curvature of 0.5 to 0.8 radians per millimeter, where the surface curvature approaches a critical value for drastic changes, an extended spiral mode is used to optimize the scanning trajectory. The spiral loop spacing is increased to 0.3 millimeters, and the loop spacing in high-curvature areas is 0.1 millimeters. The number of sampling points per loop is reduced to 24 in high-curvature areas and 36 in high-curvature areas. This reduces redundant sampling while ensuring the capture of key features, achieving… To balance efficiency and accuracy, when both holes and corner features exist in the same area, spiral scanning of holes is prioritized because the edges of holes are more prone to structural failure. Corner areas are processed later. If the delay exceeds 500 milliseconds, a fast point scan mode is activated for the corner area, reducing the sampling point density to 50% of the normal value. For example, if an area contains both a 5 mm diameter hole and a 90-degree corner, the hole is first spiral scanned with a ring spacing of 0.1 mm and 36 points per ring. The corner is then processed. If the interval is greater than 500 milliseconds (600 milliseconds), point scanning is performed on a 0.5 mm range on both sides of the corner line with a point spacing of 0.2 mm and the sampling density is halved. This strategy prioritizes the detection of hole edges while ensuring that key features in corner areas are not missed, effectively solving the problem of a single detection strategy and difficulty in balancing efficiency and accuracy in traditional solutions for medium curvature areas.

[0069] S3. Map the oxide film thickness gradient distribution on the surface of the irregular aluminum plate to the three-dimensional point cloud space constructed from the three-dimensional point cloud data, establish a linear relationship model between the hole edge deformation and the surrounding oxide layer thickness gradient. When the surrounding oxide layer thickness gradient is greater than 15% per millimeter and the hole edge deformation is greater than 0.3 millimeters, the area is determined to be a structural failure risk area, and a defect coordinate report is generated.

[0070] After completing the scanning and curvature analysis, in order to accurately determine the structural failure risk zone, it is necessary to construct a linear relationship model between the deformation of the hole edge and the oxide layer thickness gradient. The specific implementation method is as follows:

[0071] The methods for constructing the linear relationship model in step S3 include:

[0072] To establish a multi-parameter spatial correlation, the oxide film thickness gradient data was first spatially registered with the 3D point cloud coordinates to ensure consistency between compositional parameters and structural positions. During registration, the spectral inversion of the oxide film gradient was mapped to the 3D point cloud space using a coordinate transformation matrix. Simultaneously, the deformation of the pore edges was normalized using the formula: Deformation = Measured Deformation Value - Theoretical Deformation Value. The theoretical deformation value was obtained from a table based on the ratio of plate thickness to pore diameter. For example, with a plate thickness of 5mm and a pore diameter of 10mm, the theoretical deformation value was 0.1mm. If the measured deformation value was 0.4mm, the normalized deformation value was 0.3mm. This process eliminated deformation differences between plates of different specifications, providing a unified benchmark for linear modeling. Eight detectors were selected at equal intervals along the circumference of the pores. Measurement points are set at 45-degree intervals, and the average gradient value is taken within a 1mm radius for each point to reduce local fluctuation interference. When establishing a linear relationship between deformation and gradient, cubic spline interpolation is used to smooth out abnormal points with gradient values ​​greater than 20% / mm to avoid interference from abrupt changes in the model. For areas identified as failure risk zones, the deformation growth rate in the three most recent detection data is automatically retrieved. Deformation growth rate = latest deformation value - earliest deformation value / earliest deformation value × 12 / detection interval in months × 100%. If the growth rate exceeds 5% / month, a warning label is marked in the risk report, indicating that urgent processing is required. This provides a quantitative decision-making basis for the structural health management of irregular aluminum plates and effectively solves the problems of multi-parameter fragmentation and delayed risk assessment in traditional solutions.

[0073] After constructing the linear relationship model, in order to achieve refined hierarchical management of failure risks, it is necessary to expand the judgment criteria for failure risk zones. The specific implementation method is as follows:

[0074] The expanded criteria for determining the failure risk zone in step S3 include:

[0075] To accommodate different levels of defect risk, three levels of judgment criteria are set up and corresponding handling procedures are matched:

[0076] Level 1 judgment: When the oxide film thickness gradient is greater than 15% / mm and the deformation of the pore edge is greater than 0.3mm, it is directly marked as a red high-risk area, triggering the immediate maintenance process;

[0077] Secondary judgment: When the oxide film thickness gradient is greater than 20% / mm and the deformation of the hole edge is greater than 0.2mm, it is marked as a yellow warning zone and the manual review process is triggered, such as dispatching inspection personnel to confirm the defect status on site;

[0078] Level 3 Judgment: When the oxide film thickness gradient is greater than 25% / mm and the deformation of the hole edge is greater than 0.1mm, it is marked as a blue observation area and monthly tracking detection is initiated to monitor the development of defects by comparing historical data;

[0079] Considering the spatial transmission characteristics of defects, if there are two or more warning zones within 5 mm of a high-risk area, the warning zone will be automatically upgraded to a high-risk area. For example, if there are three yellow warning zones within 5 mm of a red high-risk area, the system will automatically mark these warning zones as red and simultaneously expand the detection range to 10 mm. This mechanism is based on stress transmission theory, which improves the integrity of regional risk assessment and avoids missed detections due to chain failures caused by isolated judgments. For isolated warning zones, i.e., those without other risk zones within 5 mm, an ultrasonic flaw detection verification step will be automatically added to improve the detection rate of deep damage in isolated defects.

[0080] Through the above-mentioned extended judgment conditions, the system realizes multi-dimensional classification and spatial correlation analysis of failure risk. The three-level judgment quantifies the severity of risk and matches differentiated handling procedures. The radiation effect upgrade mechanism takes into account the spatial transmission characteristics of defects. Ultrasonic verification enhances the detection depth of isolated defects, providing a scientific and quantitative basis for maintenance decisions of irregular aluminum plates. It effectively solves the problems of rough risk assessment and lack of spatial correlation analysis in traditional solutions.

[0081] After modeling the linear relationship between oxide film thickness gradient and pore deformation, a differentiated failure determination method needs to be implemented for the high-risk composite structure at the intersection of corners and pore regions. The specific implementation method is as follows:

[0082] The failure determination method for the intersection area of ​​the folding angle and the hole area in step S3 includes:

[0083] When the edge of a hole overlaps with a bend line, based on the characteristic that the hole edge is more prone to structural fracture, hole failure analysis is performed first, while bend area processing is delayed. To avoid missing detection in bend areas, no more than 30% of the total scanning energy is allocated for preliminary detection of bends. For example, if the total scanning energy for a certain intersection area is 100%, 70% of the energy is used for spiral scanning of the hole, and 30% is used for rapid point scanning of the bend line. This allocation strategy ensures the detection accuracy of high-risk hole areas while also taking into account the basic investigation of bend areas. In the failure determination of the intersection area, the oxide film gradient is taken as the highest value at the intersection of the hole center and the bend. The maximum value is used to reflect the most severe corrosion state. For example, the gradient at the center of the hole is 20% / mm, and the gradient at the intersection of the bends is 18% / mm. The maximum value of 20% / mm is taken as the criterion. The deformation is calculated by superimposing the deformation at the hole edge (60%) and the deformation at the bend (40%). The formula is: Superimposed deformation = 0.6 × Hole edge deformation + 0.4 × Bend deformation. If the hole edge deformation is 0.3mm and the bend deformation is 0.2mm, then the superimposed deformation is 0.6 × 0.3 + 0.4 × 0.2 = 0.26mm. This composite calculation method comprehensively considers the two types of defects. Synergistic effects can more accurately reflect the degree of structural damage in the intersection zone. When the oxide film gradient attenuation along the bend line direction is greater than 15% / mm, it is determined that there is a risk of latent cracks caused by stress concentration, triggering the marking mechanism. The gradient attenuation rate is calculated as follows: Gradient attenuation rate = |gradient at the bend start point - gradient at the bend end point| / bend line length. For example, if the bend line length is 10mm, the gradient at the start point is 25% / mm, and the gradient at the end point is 10% / mm, the attenuation rate is (25%~10%) / 10mm = 1.5% / mm = 15% / mm (1.5% / mm × 10 = 15% / mm). When the threshold is reached, the system automatically marks a latent crack warning in the risk report and suggests further inspection using penetrant testing. Through the above-mentioned failure determination method in the intersection zone, the system realizes priority detection of composite structural defects, multi-parameter fusion calculation, and latent risk warning. The priority strategy ensures that detection resources are tilted towards high-risk areas, the composite calculation rules quantify the defect synergy effect, and the gradient attenuation marking strengthens the early identification of latent cracks. It provides a precise composite defect analysis method for the structural safety assessment of irregular aluminum plates and effectively solves the problem of missing risk assessment of composite structures in traditional methods.

[0084] After determining the single failure risk zone, in order to comprehensively assess the structural safety of irregularly shaped aluminum plates, it is necessary to consider the cross-regional correlation effect of the risk zone. The cross-regional correlation methods for failure risk zones include:

[0085] To reflect the stress transmission effect of the high-risk area on the surrounding area, the influence radius of the high-risk area is defined as R = 0.5 × (deformation / 0.3). 2mm, where the deformation is the normalized hole edge deformation of the high-risk area. The warning area within the influence radius is automatically upgraded to a high-risk area. This formula is based on the stress attenuation law in material mechanics. By squared the deformation, the influence range of the high-deformation area is amplified, making the risk assessment more consistent with actual working conditions. The warning area upgrade mechanism within the influence radius can capture potential cascading failure risks and avoid missed detections due to isolated judgments. When the distance between adjacent holes is less than twice the minimum hole diameter, it is determined that there is a risk of stress concentration in the hole-to-hole area. Three additional measurement points need to be inserted on the line connecting the two holes. For example, if the minimum hole diameter is 5 mm and the hole distance is 8 mm, then three additional measurement points should be evenly inserted on the line connecting the two holes. One sampling point is placed at each end of the hole, 2 mm away, and another at the midpoint. The density of the sampling points is 1.5 times that of the conventional sampling points to enhance the detection of weak areas. For the additional measurement points inserted, if the oxide film thickness gradient is greater than 25% of the average gradient of the edges of the holes on both sides, it is marked as an inter-hole fatigue zone. The formula for calculating the average gradient is: average on both sides = gradient of the left hole edge + gradient of the right hole edge / 2. This marking mechanism combines the coupling effect of oxidation corrosion and mechanical fatigue, providing a key basis for structural life prediction. It effectively solves the problem of the lack of cross-regional risk correlation analysis in traditional schemes and provides a systematic solution for the safety management of metal structures in the high-end manufacturing field.

[0086] The second objective of this invention is to provide a system for realizing an intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis, comprising:

[0087] The dynamic scanning control unit 1 integrates the multispectral imaging system and the line laser scanner's collaborative motion module. It drives the scanning trajectory to switch dynamically through the real-time curvature calculation module. When the local curvature is greater than 0.8 rad / mm, it activates spiral encrypted sampling, and when the curvature is less than 0.2 rad / mm, it switches to grid sparse sampling.

[0088] The component deformation correlation analysis unit 2 inverts the oxide film thickness gradient in real time based on spectral reflectance data, maps the gradient distribution to the three-dimensional point cloud through the spatial registration engine, and uses the linear relationship modeling module to simultaneously analyze the deformation of the pore edge and the oxide layer gradient, and outputs a structural failure risk coordinate report.

[0089] The composite structure collaborative decision-making unit 3 is configured with a priority arbitrator to automatically allocate detection resources to the intersection area of ​​holes and bends, and identifies hidden cracks through the stress transmission analysis module. Finally, the cross-regional correlation engine is activated to conduct collaborative early warning of radiation effects in high-risk areas and hole group failures.

[0090] This method simultaneously acquires data through multispectral imaging and line laser scanning, inverts surface composition parameters such as oxide film thickness gradient, and dynamically switches between spiral-density or sparse-grid scanning trajectories based on three-dimensional point cloud computing of local curvature to achieve differentiated detection of complex structures and planar areas. Furthermore, it constructs a spatial correlation model between oxide film thickness and pore deformation, and combines a three-level risk assessment rule with cross-regional impact analysis to accurately identify structural failure risk zones. Through dynamic scanning control, composition deformation correlation, and the collaboration of composite structure decision units, the system effectively improves the accuracy and efficiency of defect identification in irregularly shaped aluminum plates.

[0091] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent detection of structural parameters of irregularly shaped aluminum plates based on spectral analysis, characterized in that, Includes the following steps: S1. A multispectral imaging system and a line laser scanner are synchronously moved along the surface of the irregular aluminum plate to acquire spectral reflectance data and three-dimensional point cloud data. Based on the spectral reflectance data, the surface composition parameters of the irregular aluminum plate, including the oxide film thickness gradient distribution and impurity concentration distribution, are inverted in real time. S2. Calculate the local curvature value based on the three-dimensional point cloud data of the current irregular aluminum plate surface. When the local curvature value is greater than 0.8 radians per millimeter, control the scanning device to perform encrypted sampling of the corner and hole areas with a spiral trajectory. When the local curvature value is less than 0.2 radians per millimeter, switch to a grid trajectory to perform sparse sampling of the planar area. S3. Map the oxide film thickness gradient distribution on the surface of the irregular aluminum plate to a three-dimensional point cloud space constructed from three-dimensional point cloud data, establish a linear relationship model between the hole edge deformation and the surrounding oxide layer thickness gradient. When the surrounding oxide layer thickness gradient is greater than 15% per millimeter and the hole edge deformation is greater than 0.3 millimeters, the area is determined to be a structural failure risk zone, and a defect coordinate report is generated.

2. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 1, characterized in that, The method for calculating the local curvature value in step S2 includes the following sub-steps: Based on 3D point cloud data, a normal vector is generated for each measurement point on the surface of the irregular aluminum plate. The normal vector is calculated by fitting the neighborhood plane using the least squares method. A spherical neighborhood space with a radius of 1 mm is divided with the current measurement point as the center, and all point cloud coordinate data in this space are extracted. Calculate the spatial angle between the normal vector of each measurement point in the neighborhood plane and the normal vector of the current point. When the standard deviation of the angle exceeds 15 degrees, the region is determined to be a curvature change zone, and the direction of the maximum angle deviation is recorded. The standard deviation of the included angle is converted into a curvature value proportionally. The conversion factor is set to increase the curvature value by 1 radian per millimeter for every 10 degrees of the standard deviation of the included angle. An additional curvature compensation factor is superimposed on the edge area of ​​the hole, and the compensation value is 0.2 times the reciprocal of the hole diameter.

3. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 2, characterized in that, In step S2, when the local curvature value is greater than 0.8 radians per millimeter, the execution rules for encrypted sampling include: Within a 0.5mm range on both sides of the bend line, an Archimedean spiral is used for trajectory planning. The starting point of the spiral is located at the point of maximum curvature, the spiral ring spacing is fixed at 0.1mm, and 36 sampling points with equal angle distribution are set in each ring. For the perforated area, the spiral line converges and scans from the outer edge to the center. The convergence step size is dynamically adjusted according to the aperture size. The step size calculation formula is 1 / 50 of the aperture size value and does not exceed 0.3 mm. In the range of curvature values ​​of 0.8 to 1.2 radians per millimeter, the scanning speed is linearly reduced to 30% of the reference speed. When the reflectivity difference between adjacent sampling points exceeds 40%, supplementary sampling points are automatically inserted, with the spacing between supplementary points being 1 / 3 of the original trajectory spacing. If three consecutive sampling points of single-point spectral reflectance data are invalid, immediately start a local concentric circle trajectory rescan with a diameter of 0.5 mm centered on the current point, with no less than three rescan rings.

4. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 2, characterized in that, In step S2, when the local curvature value is less than 0.2 radians per millimeter, the sparse sampling execution rules include: Use a zigzag grid path in the planar area, with the row spacing set to 1 / 20 of the width of the planar area and not less than 1mm. Add an arc transition at the path turning point, with a transition arc radius of 0.2mm. When the reflectance fluctuation of 5 consecutive sampling points is less than 5%, the fast skip scan mode is activated, and the skip scan step size is increased to 3 times the normal line spacing. If the difference between adjacent points is greater than 15% during the skip scan, the scan is immediately reversed to the last valid point to resume normal scanning. A bidirectional cross-scan is used for the transition area between the plane and the curved surface. The first scan is along the normal direction of the curved surface, and the second scan is along the tangent direction. The data from the two scans are weighted and fused, with the normal direction accounting for 70% of the weight. The curvature of the transition area is 0.1 to 0.2 radians per millimeter.

5. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 4, characterized in that, When the curvature value is between 0.2 and 0.8 radians per millimeter, the strategy is implemented by subdividing the curvature value: When the curvature value is between 0.2 and 0.5 radians per millimeter, a compressed grid mode is used, the line spacing is reduced to 0.5 mm, and the scanning speed is increased to 150% of the reference value; When the radius is 0.5 to 0.8 radians per millimeter, the extended spiral mode is adopted, the spiral ring spacing is increased to 0.3 mm, and the number of sampling points per ring is reduced to 24. When the same area simultaneously meets the characteristics of holes and corners, the hole spiral scan is performed first, and the corner area is processed later. If the processing time exceeds 500ms, the corner area is started in fast point scan mode, and the sampling point density is reduced to 50% of the normal value.

6. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 1, characterized in that, The method for constructing the linear relationship model in step S3 includes: The oxide film thickness gradient data is spatially registered with the three-dimensional point cloud coordinates, and the deformation of the hole edge is normalized, i.e., deformation = measured deformation value - theoretical deformation value, where the theoretical deformation value is obtained by looking up a table based on the ratio of plate thickness to hole diameter. Eight detection sites were selected at equal intervals along the circumference of the hole. The average gradient value of each site was taken within a radius of 1 mm. When establishing the linear relationship between deformation and gradient, cubic spline interpolation was performed on abnormal points with gradient values ​​greater than 20% / mm. For areas identified as failure risk zones, the deformation growth rate from the three most recent detection data is automatically retrieved. If the growth rate exceeds 5% per month, a warning label is added to the risk report.

7. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 6, characterized in that, The expanded criteria for determining the failure risk zone in step S3 include: Set three levels of judgment conditions: Level 1 judgment: When the gradient is greater than 15% / mm and the deformation is greater than 0.3mm, it is directly marked as a red high-risk area; Secondary judgment: When the gradient is greater than 20% / mm and the deformation is greater than 0.2mm, it is marked as a yellow warning zone and the manual review process is triggered; Level 3 judgment: When the gradient is greater than 25% / mm and the deformation is greater than 0.1mm, it is marked as a blue observation area and monthly tracking detection is started; If there are two or more warning zones within 5mm of a high-risk zone, the warning zone will be automatically upgraded to a high-risk zone. An ultrasonic flaw detection verification step will be added to isolated warning zones.

8. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 7, characterized in that, The failure determination method for the intersection area of ​​the folding angle and the hole area in step S3 includes: When the edge of a hole overlaps with the bend line, hole failure analysis is performed first, and the bend area is processed later. The scanning energy allocation does not exceed 30% of the total. The oxide film gradient is taken as the maximum value at the intersection of the hole center and the bend line. The deformation is calculated by superimposing the weights of 60% for the hole edge and 40% for the bend line. When the gradient attenuation along the bend line direction is greater than 15% / mm, the latent crack risk marker is triggered.

9. The intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis according to claim 8, characterized in that, The cross-regional association method for the failure risk zone includes: The radius of influence of the high-risk area is R = 0.5 × (deformation / 0.3). 2 mm, the warning zone within the influence radius is automatically upgraded to a high-risk zone. When the hole spacing is less than twice the minimum hole diameter, three additional measurement points are inserted on the connecting line. When the gradient of the additional points is greater than 25% of the average value on both sides, it is marked as the hole fatigue zone.

10. A system for implementing the intelligent detection method for structural parameters of irregularly shaped aluminum plates based on spectral analysis as described in any one of claims 1-9, characterized in that, include: The dynamic scanning control unit (1) integrates the multispectral imaging system and the line laser scanner's collaborative motion module. It drives the scanning trajectory to switch dynamically through the curvature real-time calculation module. When the local curvature is greater than 0.8 rad / mm, it activates spiral encrypted sampling. When the curvature is less than 0.2 rad / mm, it switches to grid sparse sampling. The component deformation correlation analysis unit (2) inverts the oxide film thickness gradient in real time based on spectral reflectance data, maps the gradient distribution to the three-dimensional point cloud through the spatial registration engine, and uses the linear relationship modeling module to simultaneously analyze the hole edge deformation and oxide layer gradient, and outputs a structural failure risk coordinate report. The composite structure collaborative decision-making unit (3) configures a priority arbitrator to automatically allocate detection resources to the intersection area of ​​holes and bends, and identifies hidden cracks through the stress transmission analysis module. Finally, it starts the cross-regional correlation engine to conduct collaborative early warning of radiation effects in high-risk areas and hole group failure.

Images