Radiographic detection scanning path adaptive planning method and system for large complex structural parts

Abstract

This application discloses an adaptive planning method and system for scanning paths in ray inspection of large and complex structural components, relating to the field of nondestructive testing technology. The method includes: acquiring the 3D model parameters of the structural component under test and the parameters of the ray inspection system; performing orthogonal projection on the structural component and meshing it, filtering blank mesh areas to generate a globally covered base pose set; fusing geometric features and chord length / thickness assessments to select complex shape detection regions requiring multi-angle imaging; constructing and solving a multi-objective imaging quality function for the complex region to determine the optimal region detection pose; integrating the base pose and the region detection pose, and under the transformation cost assessment of introducing a collision risk penalty constant, planning a collision-free movement path connecting all poses as the scanning path. This application effectively integrates global base mapping and local adaptive compensation, eliminating detection blind spots and ensuring optimal ray imaging quality and equipment operational safety from the underlying physical algorithm.

Description

Technical Field

[0001] This invention relates to the field of industrial nondestructive testing, specifically to an adaptive planning method and system for X-ray inspection scanning paths of large and complex structural components. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the requirements for lightweight vehicle bodies and integrated structures are increasing. Large integrated structural components such as integrated subframes, rear floor panels, front engine compartments, and battery pack housings are widely used. These structural components are typically manufactured from lightweight materials such as aluminum alloys and magnesium alloys through casting and welding processes. During the forming process, internal defects such as porosity, shrinkage cavities, inclusions, and cracks are prone to occur. To ensure the safety and reliability of the structural components, rigorous non-destructive quality testing is essential.

[0003] Radiographic testing (RT) has become one of the core methods for inspecting the quality of structural components because it can visually display the two-dimensional or three-dimensional morphology of internal defects. Its basic principle is: an X-ray or gamma-ray source emits X-rays or gamma rays, which penetrate the structural component. A flat panel detector receives the attenuated X-ray signal and forms a digital image. By analyzing the image, it is possible to determine whether defects exist inside the structural component.

[0004] However, traditional X-ray inspection methods face severe challenges when dealing with large structural components in new energy vehicles. Firstly, the sheer size presents a significant challenge in achieving full coverage. The cone-angle radiation range of the X-ray source and the physical size of the flat panel detector limit the coverage of a single image to a limited area. The macroscopic dimensions of these components far exceed the effective detection area of ​​conventional X-ray imaging equipment. Therefore, current technology cannot achieve a complete characterization of the entire structure with a single X-ray image; multiple imaging scans at different locations are necessary for full coverage inspection. Secondly, to meet the demands of lightweight design and multifunctional integration, these components have extremely complex geometric topologies. Their surfaces not only contain large areas of thin walls but also densely distributed features such as various functional reinforcing ribs, mounting lugs, connecting bosses, and embedded grooves. This complex three-dimensional geometry introduces new inspection bottlenecks: for thick lugs, bosses, and other features extending at large angles (especially near-vertical) to the main wall surface, vertically incident X-ray beams may cause significant decreases in imaging contrast, or even complete undetectability, at their roots or at defects with specific orientations (such as delamination or cold shuts), due to projection overlap or insufficient penetration thickness.

[0005] Currently, in industrial practice, radiographic inspection of such complex structural components still heavily relies on the personal experience and technical skills of the operators. Inspectors need to repeatedly review two-dimensional drawings or three-dimensional digital models, combining their understanding of the forming process, defect mechanisms, and the physics of radiographic imaging to manually set a series of inspection points on the inspection equipment. The inspection process is heavily dependent on personal experience; different operators develop significantly different plans, leading to unstable inspection quality, and the inspection results for the same structural component under different conditions are difficult to reproduce. Furthermore, the efficiency is extremely low; for a completely new and complex structural component, complete manual path planning can take several hours to several days, which cannot meet the rapid response requirements of modern production lines.

[0006] Therefore, when dealing with large and complex structural components, the existing X-ray inspection path planning technology, which relies on human experience, not only has high labor costs, but also low efficiency in inspection path planning and low stability in the inspection process. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of existing technologies, such as low efficiency, poor coverage, inability to guarantee imaging quality for complex structures, and lack of systematic verification, which are caused by relying on manual experience for X-ray inspection path planning. This invention provides an adaptive X-ray inspection scanning path planning method and system for large and complex structural components.

[0008] To address the aforementioned technical problems, firstly, this application provides an adaptive planning method for the scanning path of ray inspection of large and complex structural components, comprising: Obtain the three-dimensional model parameters of the structural component to be tested, as well as the detection system parameters of the X-ray inspection system, wherein the detection system parameters include the effective imaging size of the X-ray inspection system; Determine the reference projection direction, which is the direction perpendicular to the plane of the tray supporting the structure to be measured; A two-dimensional projection image is obtained by orthogonally projecting the structure under test along the reference projection direction. A rectangular bounding box that completely surrounds the two-dimensional projection image is constructed. The rectangular bounding box is divided into grids to obtain multiple grid regions. The grid region size is the same as the effective imaging size. Each of the grid regions is compared with the two-dimensional projection image to obtain the grid region containing the two-dimensional projection image as the detection grid region. The basic detection pose corresponding to each detection grid region is determined based on the center point of each detection grid region and the reference projection direction. The curvature and surface normal vector of the structure to be tested are extracted. The surface patches whose curvature is greater than a set curvature threshold and whose angle between the surface normal vector and the principal plane normal vector is greater than a set angle threshold are extracted as feature surface patches. At least two feature surface patches whose spatial distance is less than a set distance threshold are aggregated using a region aggregation algorithm to obtain a potential complex region. Multiple sampling points are set in the potentially complex area, and rays are emitted along the normal direction of each sampling point. The chord length between the first intersection point and the second intersection point generated by the rays passing through the inside of the structure under test is simulated and calculated. Calculate the average or maximum value of the chord length of all sampling points within the potentially complex region as the region thickness; The potentially complex regions whose thickness is greater than a set thickness threshold are identified as complex shape detection regions; For any of the complex shape detection regions, with the center point of the complex shape detection region as the center of a sphere, any vector passing through the center of the sphere is defined as the optimization vector; An imaging quality function is constructed with the optimized vector as the independent variable and the imaging quality score as the dependent variable. The imaging quality function is solved to obtain the optimized vector that maximizes the imaging quality score, which is then used as the scanning vector of the complex shape detection region. The region detection pose corresponding to the complex shape detection region is determined based on the scanning vector and the detection system parameters. A path planning algorithm is used to generate a collision-free movement path that connects all the basic detection poses and the region detection poses, which serves as the final scanning path.

[0009] In one possible embodiment, constructing the imaging quality function with the optimization vector as the independent variable and the imaging quality score as the dependent variable includes: Multiple imaging quality sub-functions are provided, with the optimization vector as the independent variable, and the dependent variable of each imaging quality sub-function is an imaging quality influence parameter; The imaging quality function is obtained by weighting all of the multiple imaging quality sub-functions.

[0010] In one possible embodiment, the imaging quality sub-function includes a defect visibility function and an image distortion function; The defect visibility function is: ,in, For the optimized vector, For defect visibility parameters, For the first The complex shape detection region. For complex shape detection areas The first A surface, for The unit normal vector; The image distortion function is: , ,in, For complex shape detection areas Principal plane normal vector, These are image distortion parameters; In one possible embodiment, the imaging quality sub-function further includes a geometric occlusion function; The geometric occlusion function includes: ,in, For geometric occlusion parameters, The total number of sampled rays from the ray source, along the optimized vector direction, directed toward the three-dimensional bounding box of the complex shape detection region. The number of unblocked sampling rays that were not blocked by the structure under test.

[0011] In one possible embodiment, generating a collision-free movement path connecting all poses using a path planning algorithm as the scanning path includes: Calculate the pose transformation cost of the ray detection system from the first detection pose to the second detection pose. The transformation cost includes position movement cost, attitude transformation cost and collision risk cost. The first detection pose and the second detection pose are any two detection poses from all the integrated poses. The scanning path is selected from the movement paths that traverse all the detected poses and whose sum of pose transformation costs satisfies the set conditions.

[0012] In one possible embodiment, calculating the pose transformation cost of the ray detection system from a first detection pose to a second detection pose includes: Construct several pose transformation paths from the first detection pose to the second detection pose; For any of the pose transformation paths, calculate whether a collision will occur between the structure under test and the ray detection system when the ray detection system performs pose transformation along the pose transformation path, based on the three-dimensional model parameters and the detection system parameters; If yes, the collision risk cost is determined to be a set penalty constant; if no, the collision risk cost is determined to be zero. The penalty constant is greater than the maximum value of the sum of the position movement cost and the attitude transformation cost, so that the pose transformation path with collision risk is eliminated from the planning of the scanning path.

[0013] During the planning phase of the global scanning path sequence, this application comprehensively evaluates the transformation costs across multiple dimensions, including position movement, attitude change, and collision risk. The cost associated with collision risk is set as a specific penalty constant, which is limited to the maximum value greater than the sum of all other costs. This mechanism directly eliminates high-risk pose transformation paths at the underlying mathematical calculation level of path solving, thereby ensuring the safety and industrial feasibility of the detection path execution in actual automated detection equipment or robotic arms.

[0014] In one possible embodiment, it further includes: The structure under test is simulated and scanned along the scanning path to obtain the simulated scanning results. The system automatically feeds back and adjusts the corresponding detection pose based on the simulated scan results until the simulated scan results meet the set verification conditions.

[0015] By simulating the detection pose and generating simulation scan results before issuing the final control command, and automatically performing pose feedback and iterative adjustments based on the simulation scan results, a systematic verification of the path plan is achieved. This pre-verification closed loop significantly reduces the trial-and-error costs and scrap rate on the actual physical production line, and realizes high-precision, adaptive, and intelligent path planning.

[0016] In one possible embodiment, the simulated scanning of the structure under test along the scanning path includes: For each detection pose, the coordinates of the three-dimensional model of the structure under test are rotated and translated to the camera coordinate system of the detector based on perspective projection transformation, and corrected based on the lens distortion model to obtain the pixel coordinates; For each pixel coordinate, based on the ray attenuation law, combined with the ray's path length within the structure under test, the initial ray intensity, and the material attenuation coefficient, the theoretical grayscale value of the pixel coordinate is calculated. A simulated ray image is generated by traversing all pixel coordinates as the simulated scanning result.

[0017] Secondly, this application provides an adaptive planning system for the scanning path of ray inspection of large and complex structural components, comprising: The data acquisition module is used to acquire the three-dimensional model parameters of the structural component under test, as well as the detection system parameters of the X-ray inspection system, the detection system parameters including the effective imaging size of the X-ray inspection system; A global scanning pose generation module is used to determine a reference projection direction, which is perpendicular to the plane of the tray supporting the structure under test. The structure under test is orthogonally projected along the reference projection direction to obtain a two-dimensional projection image. A rectangular bounding box completely surrounds the two-dimensional projection image. The rectangular bounding box is divided into multiple grid regions, the size of which is the same as the effective imaging size. Each grid region is compared with the two-dimensional projection image to obtain the grid region containing the two-dimensional projection image as the detection grid region. The basic detection pose corresponding to each detection grid region is determined based on the center point of each detection grid region and the reference projection direction. A complex region identification module is used to extract the curvature and surface normal vectors of the surface of the structure under test. Surfaces with curvature greater than a set curvature threshold and an angle between the surface normal vector and the principal plane normal vector greater than a set angle threshold are extracted as feature surfaces. A region aggregation algorithm is used to aggregate at least two feature surfaces with a spatial distance less than a set distance threshold to obtain a potential complex region. Multiple sampling points are set within the potential complex region, and rays are emitted along the normal direction of each sampling point. The chord length between the first and second intersection points generated by the rays passing through the interior of the structure under test is simulated and calculated. The average or maximum value of the chord lengths of all sampling points within the potential complex region is calculated as the region thickness. Potential complex regions with a region thickness greater than a set thickness threshold are identified as complex shape detection regions. The optimal pose determination module, for any of the complex shape detection regions, is used to define any vector passing through the center of the complex shape detection region as an optimization vector, with the center point of the complex shape detection region as the center of a sphere; construct an imaging quality function with the optimization vector as the independent variable and the imaging quality score as the dependent variable; solve the imaging quality function to obtain the optimization vector that maximizes the imaging quality score, which is then used as the scanning vector of the complex shape detection region; and determine the region detection pose corresponding to the complex shape detection region based on the scanning vector and the detection system parameters. The scanning path planning module is used to generate a collision-free movement path that connects all the basic detection poses and the region detection poses through a path planning algorithm, which serves as the final scanning path.

[0018] Compared with related technologies, the adaptive planning method and system for X-ray inspection scanning paths of large and complex structural components provided in this application have the following significant technological advancements and beneficial effects: This application combines two-dimensional orthogonal projection with a meshed bounding box to accurately eliminate blank and redundant regions that do not contain the projected image, thereby efficiently generating the global basic pose corresponding to the detected mesh region. Building upon this, a region filtering mechanism based on chord length and thickness threshold evaluation is further introduced to accurately locate complex shape detection regions requiring multi-angle compensation. This strategy effectively combines global basic scanning with local adaptive compensation, significantly improving detection coverage while avoiding invalid calculations for thin-walled structures, effectively overcoming the technical defect of easily missing local features in complex structural components.

[0019] This application also transforms the ray detection pose planning into a multi-objective mathematical optimization problem, overcoming the limitation of traditional detection methods that rely excessively on human experience. By constructing a comprehensive imaging quality function with the optimization vector as the independent variable, quantitative calculations and function maxima solutions are performed on defect visibility, image distortion, and geometric occlusion factors. This application ensures optimal ray imaging quality for complex-shaped detection areas from the perspective of physical geometric features, significantly improving the stability and reliability of the detection results. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall process of the adaptive planning method for X-ray inspection scanning path of large and complex structural components provided in the embodiments of this application. Figure 2 This is a schematic diagram of the simulation model of the structural component and the ray inspection system in the adaptive planning method for the scanning path of ray inspection of large and complex structural components provided in the embodiments of this application. Figure 3 This is a schematic diagram of a two-dimensional projection image in the adaptive planning method for the scanning path of ray inspection of large and complex structural components provided in the embodiments of this application. Figure 4 This is a schematic diagram illustrating the filtering of blank grid regions completely outside the two-dimensional projection image in the adaptive planning method for ray inspection scanning paths of large and complex structural components provided in the embodiments of this application. Figure 5 This is a flowchart illustrating the process of determining the region detection pose (multi-objective optimization) in the adaptive planning method for the scanning path of ray inspection of large and complex structural components provided in the embodiments of this application. Figure 6 This is a schematic diagram of the region detection pose and the complex shape detection region in the adaptive planning method for the scanning path of ray inspection of large and complex structural components provided in the embodiments of this application. Figure 7 This is a schematic diagram of the global scanning path in the adaptive planning method for ray inspection scanning path of large and complex structural components provided in the embodiments of this application; Figure 8This is a schematic diagram of the adaptive planning system for the scanning path of a large and complex structural component for X-ray inspection provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0022] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0023] The terms "first," "second," etc., used in the embodiments of this application are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of those features.

[0024] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0025] This application provides an adaptive planning method and system for the scanning path of ray inspection of large and complex structural components, which will be described in detail below.

[0026] Please refer to Figure 1 The adaptive planning method for X-ray inspection scanning paths of large and complex structural components provided in this application includes: Step S101: Obtain the three-dimensional model parameters of the structural component to be tested, as well as the detection system parameters of the X-ray inspection system.

[0027] In this application, the three-dimensional model parameters of the structural component under test are, for example, three-dimensional CAD models in formats such as STEP, IGES, or STL. Specifically, any model parameters that can indicate the geometric topology of the structural component under test are acceptable, and there is no limitation on the specific data format.

[0028] In this application, the X-ray detection system specifically includes a X-ray source for emitting scanning X-rays and a detector for receiving scanning X-ray imaging. Based on this, the detection system parameters specifically include parameters such as the focal spot size and radiation cone angle of the X-ray source, the effective imaging area and pixel resolution of the detector, the equivalent focal length of the system, and the limits of the mechanical movement range of the detection equipment.

[0029] Further, please refer to Figure 2 In this embodiment, a simulation model of the structural component and the X-ray inspection system is also constructed based on the three-dimensional model parameters and the detection system parameters. That is, the three-dimensional structural simulation of the structural component and the scanning process simulation of the X-ray inspection system are realized in the same simulation model.

[0030] In this step, the simulation of the scanning process of the X-ray inspection system specifically includes: S1: Establish a coordinate mapping model using a pinhole camera model and define the ray source. Define the probe's pose (including the position of its center in the world coordinate system) as the perspective center. and the rotation matrix describing its spatial orientation. For any sampling point on the surface of the structural component. Perform the following transformation to simulate the pixel coordinates projected onto the detector imaging plane. : 1. Perform coordinate transformation and calculate. , sampling points Transformation from world coordinate system to detector coordinate system .

[0031] 2. Perform perspective projection transformation to project the point in the camera coordinate system onto the detector, obtaining its projected position coordinates on the normalized image plane. ,in, This is the equivalent focal length of the pinhole camera model.

[0032] 3. Perform distortion correction and pixel conversion: First, apply the lens distortion model (including radial and tangential distortion) to the ideal coordinates. Perform correction to obtain the corrected coordinates. Subsequently, the detector's intrinsic parameter matrix (including pixel size, principal point coordinates, etc.) is used to convert the corrected coordinates into projected pixel coordinates. .

[0033] S2: According to the Beer-Lambert Law Construct a ray attenuation imaging simulation model, adjust the ray attenuation according to the following steps, and generate the simulation grayscale value corresponding to each pixel: 1. Perform ray tracing and path length calculation: For each pixel on the detector Based on the mapping relationship established by S1, the corresponding three-dimensional ray direction is determined in reverse. From the ray source... A virtual ray is emitted along this direction. The intersection point of the virtual ray and the 3D structural model of the component is determined, and the total path length of the ray traversing the interior of the component's solid material is calculated. .

[0034] 2. Based on the material attenuation coefficient of the structural component under test. Initial radiation intensity of the radiation detection system Calculate the pixel point Simulated grayscale values .

[0035] S3: Traverse all pixels within the detector's imaging area, repeating S1 and S2 above to generate the corresponding pixel coordinates and simulated grayscale values, thus obtaining a complete simulated ray image of the structural component corresponding to its current pose.

[0036] Step S102: Determine the reference projection direction and generate a global coverage base pose set.

[0037] Please refer to Figure 3 and Figure 4 This step aims to generate a global baseline scan pose through two-dimensional orthogonal projection and meshed Boolean comparison, specifically including the following steps: First, determine the reference projection direction of the X-ray inspection system, such as... Figure 2 As shown, the reference projection direction is defined as the direction perpendicular to the plane of the tray supporting the structure under test, i.e. Figure 2 The Z-axis direction is used. In the simulation model of the structural component and the ray inspection system constructed in the preceding steps, the structural component is orthogonally projected along the reference projection direction to obtain a two-dimensional projection image. .

[0038] Then, construct a fully enclosed two-dimensional projected image. The rectangular bounding box. Specifically, this includes: for two-dimensional projected images. Based on the formula Calculate the bounding box of a rectangle ,in, , These represent the minimum and maximum values ​​of the two-dimensional contour points in the X direction, respectively. , These are the minimum and maximum values ​​in the Y direction, respectively.

[0039] Furthermore, in some embodiments of this application, to compensate for positioning errors and ray beam edge effects, and to ensure that the edges of structural components are completely covered, a redundancy distance can be preset. (Specifically set to a fixed value or calculated proportionally based on the size of the two-dimensional projected image), based on the formula For rectangular bounding boxes The boundary is expanded to obtain the expanded bounding box. .

[0040] Finally, the expanded bounding box The detector is divided into grids, resulting in multiple grid regions with the same size as the effective imaging size. Specifically, this includes: assuming the effective imaging size of the detector is... Set the overlap rate of adjacent scan areas. Based on the formula Calculate the scan step size. Using the expanded bounding box... The bottom left corner ( Using ) as the starting reference point, Using a step size, generate a regular rectangular mesh within the expanded bounding box. Each grid area The coordinates of the lower left corner are: ;in, , , and These represent the number of points required to cover the expanded bounding box in the X and Y directions, respectively.

[0041] For further details, please refer to Figure 4 In some embodiments of this application, blank grid regions that are completely outside the two-dimensional projected image are filtered. For each grid region... Calculate its relationship with the two-dimensional projected image. overlapping area If the overlapping area If the value is zero, it is determined to be a completely empty grid region on the outside, and it is removed from the grid lattice. The remaining mesh regions are then filtered out. Based on the center point and reference projection direction of the remaining mesh regions, the basic detection pose is determined, ultimately resulting in the first set of filtered detection poses. .

[0042] Step S103: Identify the complex regions that need to be further detected, and obtain the complex shape detection regions that require multi-angle imaging.

[0043] Based on the generated global baseline pose, this step further identifies regions that are easily missed due to their large thickness or complex geometry, and performs secondary selection. Specifically, by extracting the curvature and patch normal vectors of the surface of the structure under test, patches with curvature greater than a set curvature threshold and whose angle between the patch normal vector and the principal plane normal vector is greater than a set angle threshold are extracted as feature patches. A region aggregation algorithm is then used to aggregate at least two feature patches whose spatial distance is less than a set distance threshold to obtain potentially complex regions. This includes the following steps: 1. Extraction of surface geometric features and preliminary screening: Extract the geometric features of the surface of the structural component under test, including but not limited to Gaussian curvature (K), mean curvature (H), and unit normal vector (H). (etc.). Feature patches are selected by combining curvature thresholds, and geometric semantic rules are constructed and applied based on the definition of typical engineering structural features to perform preliminary classification and selection of complex structural regions. The defined geometric semantic rules include, but are not limited to: (a) Ear / protrusion recognition conditions: Recognize the area adjacent to a large area with low curvature, and the normal vector of its own surface. Normal vector of the principal plane The set of surfaces with an included angle greater than a set threshold .

[0044] (b) Rib identification criteria: Identify a set of slender, strip-shaped surfaces whose normal vectors on both sides along their extension direction are approximately opposite. . (c) Recess / hole identification conditions: Identify the set of recessed regions that are surrounded by the principal plane and whose internal curved surface normal vectors are opposite in direction to or form a large obtuse angle with the principal plane normal vector. . Region aggregation algorithms (such as DBSCAN) are used to group regions that are spatially close and have similar characteristics, forming a set of potentially complex regions. (include , , ).

[0045] 2. Local thickness assessment based on chord length (secondary refinement): For each set of potential complex regions obtained from the above preliminary screening Local thickness assessment is performed to filter out redundant areas (such as thin stiffeners) that, despite their complex geometry, have thin walls and do not require multi-angle imaging. Specifically, this includes: 1. Calculate local thickness: in the region Multiple sampling points are selected evenly inside. For each sampling point Along its normal vector Rays are emitted in both positive and negative directions and intersected with the three-dimensional mesh model of the structure under test. The distance between the first and second intersection points (i.e., the chord length) of the rays as they travel through the interior of the structure under test is calculated using simulation. ), which represents the local material thickness at that point.

[0046] 2. Perform region thickness characterization: Calculate the thickness (crossing chord length) of all sampling points within the region. (Average) or maximum value This serves as a representative thickness for the region.

[0047] 3. Perform threshold determination: Set a thickness threshold. (This threshold can be based on the overall average wall thickness of the structural component under test) (Multiplied by a safety factor to determine). If the area The representative thickness satisfies (or If the potentially complex region is identified as a complex shape detection region requiring multi-angle imaging, then it is denoted as... .

[0048] Step S104: Determine the region detection pose corresponding to the complex shape detection region.

[0049] Please refer to Figure 5 In this step, for each complex shape detection region identified in the preceding steps, an optimal imaging pose, i.e., the region detection pose, is determined by constructing and solving a multi-objective optimization problem. Specifically, this includes the following steps: Step S501: For any complex shape detection region, take the center point of the complex shape detection region as the center of the sphere, and define any unit vector passing through the center of the sphere as the optimization vector. This vector points positively to the center of the detector and negatively to the X-ray source.

[0050] In this step, the detection areas for each complex shape are first determined based on the three-dimensional structure of the structural component. The center point of the detection area with this complex shape Construct an interference-free space at the center of the sphere that can be scanned and detected. For example, in the scanning area of ​​a flat plate structure, only the front and back sides can be subjected to ray detection; ray detection is usually not performed along its extension direction. Therefore, the corresponding space... That is, a hemispherical space. Then in space... Define any vector passing through the center of the sphere as the optimization vector. .

[0051] Step S502: Construct an imaging quality function with the optimization vector as the independent variable and the imaging quality score as the dependent variable.

[0052] In this step, the simulation model of the structural components and the X-ray inspection system constructed in the previous steps is used to optimize the vector. The X-ray inspection system simulates scanning of various complex-shaped inspection areas using its ray direction. It also provides multiple imaging quality sub-functions, each corresponding to an imaging quality influencing parameter; an optimization vector is calculated based on each imaging quality sub-function. The simulated scan results correspond to the scores of multiple influencing parameters; the optimization vector is obtained by weighting all the influencing parameter scores. Imaging quality score of simulated scan results .

[0053] The imaging quality sub-function specifically includes: Defect visibility function To ensure that the X-ray direction is advantageous for detecting surface or near-surface defects (such as cracks), the X-ray direction should be made as directional as possible. Perpendicular to the plane where the potential defect is located. Define the defect visibility function. for: ,in, To optimize the vector, For defect visibility parameters, For the first A complex shape detection area, For complex shape detection areas The first A surface, for The unit normal vector.

[0054] Image distortion function To reduce image distortion caused by perspective projection, the detector plane should be made as parallel as possible to the main surface of the area to be inspected. (Definition) Normal vector of the detector plane With the region Principal plane normal vector The cosine of the included angle (which can be calculated using PCA) is expressed by the following formula: The closer this value is to 1, the more parallel the two planes are, and the smaller the perspective distortion.

[0055] In addition to the defect visibility function and image distortion function mentioned above, in some embodiments of this application, the imaging quality sub-function may also include: Geometric occlusion minimization function To avoid obstruction of the detection path by other parts of the structure, the distance from the X-ray source to the area needs to be maximized. The visibility of the radiation path. Using the simulation model of the structural components and radiation detection system constructed in the preceding steps, the radiation path is observed from the radiation source to the area. A dense beam of sampling rays is emitted, and the proportion of unblocked rays is counted. ,definition: The larger the value, the less occlusion occurs.

[0056] Path connectivity efficiency function To improve overall detection efficiency, the selected pose should be easily integrated into the global scan sequence. This function evaluates candidate poses. With its adjacent poses in the global path , The conversion cost between them is used to reflect this: ,in, and These represent the Euclidean distance between positions and the angular change between attitudes, respectively. These are the weighting coefficients. The larger the value, the smoother the connection.

[0057] Collision risk avoidance function To ensure equipment safety, collisions between the detector / radiation source and structural components or the environment must be avoided. Rapid interference checks are performed using the equipment's kinematic model and the bounding box of the 3D scene. If in pose... If there is no risk of collision, then ;otherwise, .

[0058] Finally, the image quality score is calculated using a weighted sum method. for: ,in, It is a non-negative weighting coefficient, and its specific value can be adjusted according to the priority of imaging quality, efficiency and safety requirements in different detection tasks.

[0059] Step S503: Solve for the imaging quality function to obtain the optimized vector that maximizes the imaging quality score, which is then used as the scanning vector for the complex shape detection region.

[0060] In this step, the imaging quality function is constructed. Then, the imaging quality function is maximized using methods such as intelligent optimization algorithms (e.g., genetic algorithms, particle swarm optimization algorithms) or search strategies based on Pareto optimal solution sets to obtain the imaging quality score. highest optimization vector As a complex shape detection area scan vector Simultaneously, the scan vector is defined under the region detection pose. It is necessary to pass through the center of the radiation source and the detector, and scan the vector. normal vector of the detector plane Parallel, the distance from the radiation source to the detector is a fixed value. The ray cone angle needs to cover the entire complex-shaped detection area. Based on this, according to Fixed distance and regional center Calculate the specific coordinates of the ray source. and detector pose (including the detector's center coordinates) Corresponding rotation matrix ).

[0061] For example, in this embodiment, a genetic algorithm (GA) is used for optimization, and the specific steps include: Initialize the population: in the constrained space A set of optimization vectors is randomly generated within the inner circle. This constitutes the initial population.

[0062] Assessing population fitness: For each individual in the population (observation direction) Each of them calculates its corresponding imaging quality score based on the imaging quality function.

[0063] Iterative evolution: Genetic operations such as selection, crossover, and mutation are performed to generate a new generation of the population. In each generation, individuals with high imaging quality scores are retained, while those with low imaging quality scores are eliminated.

[0064] Termination and Output: Evolution stops when the maximum number of iterations is reached or the fitness converges. The optimization vector corresponding to the individual with the highest imaging quality score in the final population is selected. As a scan vector.

[0065] For each complex shape detection area Repeat the above steps, such as Figure 6 As shown, the region detection poses corresponding to each complex shape detection region are generated. .

[0066] Step S105: Generate a collision-free moving path that connects all basic detection poses and region detection poses using a path planning algorithm, which will serve as the final scanning path.

[0067] In this step, the pose conversion cost of the ray detection system from the first detection pose to the second detection pose is first calculated. The conversion cost includes position movement cost, attitude transformation cost and collision risk cost. The first detection pose and the second detection pose are any two detection poses in the integrated set of all poses. Then, the movement path that traverses all detection poses and whose sum of pose conversion costs meets the set conditions is selected as the scanning path.

[0068] In this embodiment, it is specifically based on the formula The computational X-ray inspection system starts from the first detection pose. Convert to second detection pose pose transformation cost ,in, For position parameters, The first detection pose Convert to second detection pose The Euclidean distance between them; , For posture parameters, In order to adopt a posture Adjust to posture The minimum time required is proportional to the generalized angular difference between the two attitudes. This is the collision indication function. If from... arrive If there is a collision risk in the linear motion or planned trajectory, a large penalty constant is returned; otherwise, it is 0. The penalty constant is set to be greater than the maximum sum of the position movement cost and the attitude transformation cost, so that pose transformation paths with collision risks are eliminated from the planning of the scan path. These are the weighting coefficients for each cost item, used to balance the contributions of travel distance, attitude adjustment time, and collision risk to the total cost.

[0069] In this embodiment, the scanning path is selected as the movement path that traverses all regions to detect poses and whose sum of pose transformation costs meets a set condition. Specifically, this includes the following steps: Multiple movement paths are generated based on a path planning algorithm, traversing all detection poses. Each movement path includes all detection poses and path transformation points during the pose adjustment process. The movement direction and / or spatial attitude of the ray detection system are adjusted at the path transformation points to obtain multiple candidate paths. For each candidate path The pose transformation cost between every two adjacent detection poses in the candidate path is calculated based on the above formula. Calculate candidate paths The sum of the corresponding pose transformation costs. Finally, the candidate path whose sum of pose transformation costs meets the set conditions (e.g., the sum of pose transformation costs is less than a set threshold, or the sum of pose transformation costs is the smallest among all known candidate paths) is selected as the candidate path. Figure 7 The scan path shown includes all detection poses and corresponding path transformation points.

[0070] Furthermore, in some embodiments of this application, after determining the scanning path, the structural component to be tested is simulated and scanned along the scanning path in the simulation model of the structural component and the X-ray inspection system constructed in the aforementioned steps to obtain the simulated scanning result; the corresponding detection pose is automatically adjusted based on the simulated scanning result until the simulated scanning result meets the set verification conditions.

[0071] The simulated scanning results include a set of simulated ray images at various detection poses. and the corresponding three-dimensional irradiation area set Based on this, full coverage and blind zone verification were performed on the simulated scanning results, including: irradiating the areas corresponding to all poses along the path. Perform a three-dimensional Boolean union operation to obtain the total coverage area under path planning. The specific calculation formula is as follows: Calculate the total area of ​​the coverage region. Total theoretical projected area of ​​structural components The ratio of the two values ​​gives the path coverage. The calculated path coverage Compared with the preset coverage threshold Compare. If Less than the coverage threshold If the full coverage verification fails, then the verification is deemed unsuccessful.

[0072] The imaging quality of the simulated scan results can also be verified, specifically including: for each complex shape detection area From the simulated ray image set Extract the corresponding simulated ray image In the image In the middle, the positioning area The projection area. Within the projected area, preset image quality metrics are evaluated, including but not limited to: effective penetration, contrast, and geometric sharpness. The evaluation results are compared with preset quality standards. If any key metric fails to meet the standard, the imaging quality verification for that complex area is deemed unsuccessful.

[0073] The simulation scan results can also be used to verify the rationality of the imaging angle, specifically including: for each complex shape detection area And its corresponding imaging pose, calculate the principal ray direction vector. .calculate The normal vector of each sampling point on the surface of this region The included angle Count all included angles. average Alternatively, calculate the proportion of the included angle within a preset ideal angle range. Determine if the statistical results are within the preset allowable deviation range. If they exceed the range, the verification of the imaging angle's rationality is deemed unsuccessful.

[0074] For any of the aforementioned verification methods, if the simulated scan result fails the verification, i.e., the simulated scan result does not meet the set verification conditions, the relevant parameters in the aforementioned steps can be automatically adjusted based on the verification result, and the aforementioned planning steps can be re-executed to obtain a new detection pose and scanning path, and then the verification can be performed again until the verification is successful.

[0075] Compared with related technologies, the adaptive planning method and system for ray inspection scanning paths of large and complex structural components provided in this application combines two-dimensional orthogonal projection with meshed bounding boxes to accurately eliminate blank and redundant areas that do not contain the projected image, thereby efficiently generating the global basic pose corresponding to the detection mesh area. Based on this, a region filtering mechanism based on chord length and thickness threshold evaluation is further introduced to accurately locate complex shape detection areas requiring multi-angle compensation. This strategy effectively combines global basic scanning with local adaptive compensation, significantly improving detection coverage while avoiding invalid calculations for thin-walled structures, effectively overcoming the technical defect of easily missing local features of complex structural components. It also transforms ray inspection pose planning into a multi-objective mathematical optimization problem, overcoming the limitation of traditional detection methods relying excessively on human experience. By constructing a comprehensive imaging quality function with optimization vectors as independent variables, quantitative calculations and function maximization solutions are performed on defect visibility, image distortion, and geometric occlusion factors. This application ensures optimal ray imaging quality for complex shape detection areas from the perspective of physical geometric features, significantly improving the stability and reliability of the detection results.

[0076] To better implement the adaptive planning method for ray inspection scanning paths of large and complex structural components in the embodiments of this application, based on the adaptive planning method for ray inspection scanning paths of large and complex structural components, the corresponding method is as follows: Figure 8 As shown in the illustration, this application also provides an adaptive planning system for the scanning path of X-ray inspection of large and complex structural components. The adaptive planning system for the scanning path of X-ray inspection of large and complex structural components includes: The data acquisition module 801 is used to acquire the three-dimensional model parameters of the structural component to be tested, as well as the detection system parameters of the X-ray inspection system, including the effective imaging size of the X-ray inspection system. The global scanning pose generation module 802 is used to determine the reference projection direction, which is perpendicular to the plane of the tray supporting the structure to be tested. A two-dimensional projection image is obtained by orthogonally projecting the structure to be tested along the reference projection direction. A rectangular bounding box that completely surrounds the two-dimensional projection image is constructed. The rectangular bounding box is then divided into multiple grid regions, with the grid region size being the same as the effective imaging size. Each grid region is compared with the two-dimensional projection image, and the grid region containing the two-dimensional projection image is obtained as the detection grid region. The basic detection pose corresponding to each detection grid region is determined based on the center point of each detection grid region and the reference projection direction. The complex region identification module 803 is used to extract the curvature and surface normal vectors of the surface of the structure to be tested. Surfaces with curvature greater than a set curvature threshold and whose surface normal vectors are at an angle greater than a set angle threshold are extracted as feature surfaces. A region aggregation algorithm is used to aggregate at least two feature surfaces with a spatial distance less than a set distance threshold to obtain a potential complex region. Multiple sampling points are set within the potential complex region, and rays are emitted along the normal direction of each sampling point. The chord length between the first and second intersection points generated by the rays passing through the interior of the structure to be tested is simulated and calculated. The average or maximum value of the chord length of all sampling points within the potential complex region is calculated as the region thickness. Potential complex regions with a region thickness greater than a set thickness threshold are identified as complex shape detection regions. The optimal pose determination module 804 is used to determine the optimal pose for any complex shape detection region. The optimal pose determination module 804 is used to define any vector passing through the center of the complex shape detection region as the center of the sphere as the optimization vector; construct an imaging quality function with the optimization vector as the independent variable and the imaging quality score as the dependent variable; solve the imaging quality function to obtain the optimization vector that makes the imaging quality score take the maximum value as the scanning vector of the complex shape detection region; and determine the region detection pose corresponding to the complex shape detection region based on the scanning vector and the detection system parameters. The scanning path planning module 805 is used to generate a collision-free moving path that connects all basic detection poses and region detection poses through a path planning algorithm, which serves as the final scanning path.

[0077] The adaptive planning system for X-ray inspection scanning path of large and complex structural components provided in the above embodiments can realize the technical solutions described in the embodiments of the adaptive planning method for X-ray inspection scanning path of large and complex structural components. The specific implementation principles of each module or unit can be found in the corresponding content in the embodiments of the adaptive planning method for X-ray inspection scanning path of large and complex structural components, which will not be repeated here.

[0078] The above provides a detailed description of the adaptive planning method and system for X-ray inspection scanning paths of large and complex structural components provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An adaptive planning method for the scanning path of ray inspection of large and complex structural components, characterized in that, include: Obtain the three-dimensional model parameters of the structural component to be tested, as well as the detection system parameters of the X-ray inspection system, wherein the detection system parameters include the effective imaging size of the X-ray inspection system; Determine the reference projection direction, which is the direction perpendicular to the plane of the tray supporting the structure to be measured; A two-dimensional projection image is obtained by orthogonally projecting the structure under test along the reference projection direction. A rectangular bounding box that completely surrounds the two-dimensional projection image is constructed. The rectangular bounding box is divided into grids to obtain multiple grid regions. The grid region size is the same as the effective imaging size. Each of the grid regions is compared with the two-dimensional projection image to obtain the grid region containing the two-dimensional projection image as the detection grid region. The basic detection pose corresponding to each detection grid region is determined based on the center point of each detection grid region and the reference projection direction. The curvature and surface normal vector of the structure to be tested are extracted. The surface patches whose curvature is greater than a set curvature threshold and whose angle between the surface normal vector and the principal plane normal vector is greater than a set angle threshold are extracted as feature surface patches. At least two feature surface patches whose spatial distance is less than a set distance threshold are aggregated using a region aggregation algorithm to obtain a potential complex region. Multiple sampling points are set in the potentially complex area, and rays are emitted along the normal direction of each sampling point. The chord length between the first intersection point and the second intersection point generated by the rays passing through the inside of the structure under test is simulated and calculated. Calculate the average or maximum value of the chord length of all sampling points within the potentially complex region as the region thickness; The potentially complex regions whose thickness is greater than a set thickness threshold are identified as complex shape detection regions; For any of the complex shape detection regions, with the center point of the complex shape detection region as the center of a sphere, any vector passing through the center of the sphere is defined as the optimization vector; An imaging quality function is constructed with the optimized vector as the independent variable and the imaging quality score as the dependent variable. The imaging quality function is solved to obtain the optimized vector that maximizes the imaging quality score, which is then used as the scanning vector of the complex shape detection region. The region detection pose corresponding to the complex shape detection region is determined based on the scanning vector and the detection system parameters. A path planning algorithm is used to generate a collision-free movement path that connects all the basic detection poses and the region detection poses, which serves as the final scanning path.

2. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 1, characterized in that, The construction of the imaging quality function, with the optimized vector as the independent variable and the imaging quality score as the dependent variable, includes: Multiple imaging quality sub-functions are provided, with the optimization vector as the independent variable, and the dependent variable of each imaging quality sub-function is an imaging quality influence parameter; The imaging quality function is obtained by weighting all of the multiple imaging quality sub-functions.

3. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 2, characterized in that, The imaging quality sub-function includes a defect visibility function and an image distortion function; The defect visibility function is: ,in, For the optimized vector, For defect visibility parameters, For the first The complex shape detection region. For complex shape detection areas The first A surface, for The unit normal vector; The image distortion function is: , ,in, For complex shape detection areas Principal plane normal vector, These are the image distortion parameters.

4. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 2, characterized in that, The imaging quality sub-function also includes a geometric occlusion function; The geometric occlusion function includes: ,in, For geometric occlusion parameters, The total number of sampled rays from the ray source, along the optimized vector direction, directed toward the three-dimensional bounding box of the complex shape detection region. The number of unblocked sampling rays that were not blocked by the structure under test.

5. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 1, characterized in that, The step of generating a collision-free movement path connecting all poses using a path planning algorithm as the scanning path includes: Calculate the pose transformation cost of the ray detection system from the first detection pose to the second detection pose. The transformation cost includes position movement cost, attitude transformation cost and collision risk cost. The first detection pose and the second detection pose are any two detection poses from all the integrated poses. The scanning path is selected from the movement paths that traverse all the detected poses and whose sum of pose transformation costs satisfies the set conditions.

6. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 5, characterized in that, The calculation of the pose transformation cost of the X-ray detection system from the first detection pose to the second detection pose includes: Construct several pose transformation paths from the first detection pose to the second detection pose; For any of the pose transformation paths, calculate whether a collision will occur between the structure under test and the ray detection system when the ray detection system performs pose transformation along the pose transformation path, based on the three-dimensional model parameters and the detection system parameters; If yes, the collision risk cost is determined to be a set penalty constant; if no, the collision risk cost is determined to be zero. The penalty constant is greater than the maximum value of the sum of the position movement cost and the attitude transformation cost, so that the pose transformation path with collision risk is eliminated from the planning of the scanning path.

7. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 1, characterized in that, Also includes: The structure under test is simulated and scanned along the scanning path to obtain the simulated scanning results. The system automatically feeds back and adjusts the corresponding detection pose based on the simulated scan results until the simulated scan results meet the set verification conditions.

8. The adaptive planning method for X-ray inspection scanning path of large and complex structural components according to claim 7, characterized in that, The simulated scanning of the structure under test along the scanning path includes: For each detection pose, the coordinates of the three-dimensional model of the structure under test are rotated and translated to the camera coordinate system of the detector based on perspective projection transformation, and corrected based on the lens distortion model to obtain the pixel coordinates; For each pixel coordinate, based on the ray attenuation law, combined with the ray's path length within the structure under test, the initial ray intensity, and the material attenuation coefficient, the theoretical grayscale value of the pixel coordinate is calculated. A simulated ray image is generated by traversing all pixel coordinates as the simulated scanning result.

9. An adaptive planning system for X-ray inspection scanning paths of large and complex structural components, characterized in that, include: The data acquisition module is used to acquire the three-dimensional model parameters of the structural component under test, as well as the detection system parameters of the X-ray inspection system, the detection system parameters including the effective imaging size of the X-ray inspection system; A global scanning pose generation module is used to determine a reference projection direction, which is perpendicular to the plane of the tray supporting the structure under test. The structure under test is orthogonally projected along the reference projection direction to obtain a two-dimensional projection image. A rectangular bounding box completely surrounds the two-dimensional projection image. The rectangular bounding box is divided into multiple grid regions, the size of which is the same as the effective imaging size. Each grid region is compared with the two-dimensional projection image to obtain the grid region containing the two-dimensional projection image as the detection grid region. The basic detection pose corresponding to each detection grid region is determined based on the center point of each detection grid region and the reference projection direction. A complex region identification module is used to extract the curvature and surface normal vectors of the surface of the structure under test. Surfaces with curvature greater than a set curvature threshold and whose surface normal vectors form an angle greater than a set angle threshold are extracted as feature surfaces. A region aggregation algorithm is used to aggregate at least two feature surfaces whose spatial distance is less than a set distance threshold to obtain a potential complex region. Multiple sampling points are set within the potential complex region, and rays are emitted along the normal direction of each sampling point. The chord length between the first and second intersection points generated by the rays passing through the interior of the structure under test is simulated and calculated. Calculate the average or maximum value of the chord length of all sampling points within the potential complex region as the region thickness; determine the potential complex region whose region thickness is greater than a set thickness threshold as a complex shape detection region; The optimal pose determination module, for any of the complex shape detection regions, is used to define any vector passing through the center of the complex shape detection region as an optimization vector, with the center point of the complex shape detection region as the center of a sphere; construct an imaging quality function with the optimization vector as the independent variable and the imaging quality score as the dependent variable; solve the imaging quality function to obtain the optimization vector that maximizes the imaging quality score, which is then used as the scanning vector of the complex shape detection region; and determine the region detection pose corresponding to the complex shape detection region based on the scanning vector and the detection system parameters. The scanning path planning module is used to generate a collision-free movement path that connects all the basic detection poses and the region detection poses through a path planning algorithm, which serves as the final scanning path.

Images