Automatic production line digital model hole filling method and device

Abstract

According to the automatic production line digital model hole filling method and device provided by the embodiment of the invention, the to-be-processed area is accurately identified through multi-level grid analysis and curvature field calculation. A patch generation mechanism based on density clustering and a neural network is innovatively designed, and intelligent filling optimization is realized in combination with local geometric features. According to the system, non-uniform rational basis spline surface reconstruction and a Laplacian grid smoothing algorithm are adopted, continuity and fairness of a filling area are ensured, and the system is closely combined with motion control and path planning of an automatic production line. According to the method, the limitation of traditional model repair is broken through, and an efficient and reliable solution is provided for integrity reconstruction of the industrial digital model.

Description

Technical Field

[0001] This application relates to the field of data processing, and particularly to a method and device for filling holes in a digital model of an automated production line. Background Art

[0002] Traditional methods for filling holes in digital models mainly rely on simple geometric interpolation or surface fitting, making it difficult to handle complex topological structures and maintain the continuity of the model. Existing technologies lack a systematic method for extracting hole boundary features and surface reconstruction, and it is difficult to achieve adaptive filling optimization.

[0003] At the same time, existing systems have obvious deficiencies in curvature analysis and patch generation. Traditional methods often adopt a unified filling strategy, failing to fully consider the influence of local geometric features and topological constraints, and lacking an intelligent evaluation mechanism for filling quality. The system is also relatively simple in handling transition regions and mesh smoothing, and fails to achieve seamless model repair.

[0004] In addition, existing technologies also have limitations in industrial application integration. There is a lack of in-depth integration with the motion control and path planning of automated production lines, resulting in insufficient usability of the repaired model in practical applications. Solving these problems is of great significance for improving the integrity of digital models and the effects of industrial applications. Summary of the Invention

[0005] In view of the problems in the prior art, this application provides a method and device for filling holes in a digital model of an automated production line, which can break through the limitations of traditional model repair and provide an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0006] To solve at least one of the above problems, this application provides the following technical solutions:

[0007] In a first aspect, this application provides a method for filling holes in a digital model of an automated production line, including:

[0008] Decompose the digital model of the automated production line into a multi-level mesh representation, calculate the local curvature field distribution at each mesh level, project the curvature field to the finest-level mesh, calculate the curvature gradient change based on the finest-level mesh, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0009] Perform density clustering analysis on the boundary feature descriptor, map the density clustering result to a three-dimensional space to form a filling patch, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary point, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patch;

[0010] Input the patch to be optimized into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function considering the surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, and performs local adjustment in the transition region using the Laplacian mesh smoothing algorithm to generate a final filling model. Import the filling model into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection.

[0011] Further, decomposing the digital model of the automated production line into grid representations at multiple levels, calculating the local curvature field distribution at each grid level, projecting the curvature field to the finest level grid, calculating the curvature gradient change based on the finest level grid, and using an adaptive threshold segmentation algorithm to mark the region to be processed, including:

[0012] Use the octree decomposition algorithm to divide the digital model of the automated production line into grid levels of different scales, construct a vertex adjacency relationship graph for each grid level, calculate the Gaussian curvature and mean curvature of the grid vertices based on the adjacency relationship graph, and combine the Gaussian curvature and mean curvature to form the local curvature field distribution;

[0013] Use the interpolation mapping algorithm between grid levels to project the curvature fields of different levels to the finest level grid, construct a curvature gradient tensor field, calculate the principal direction and eigenvalues of the curvature gradient tensor field, construct an adaptive threshold segmentation function based on the eigenvalues, and mark the output of the segmentation function as the region to be processed.

[0014] Further, extracting boundary points from the region to be processed, calculating the normal vector field of the boundary points, and combining the position information of the boundary points and the normal vector field to construct a boundary feature descriptor, including:

[0015] Construct a boundary tracing algorithm for the region to be processed, search for the connected boundary of the region to be processed based on the grid topological connection relationship, extract the grid vertices on the connected boundary as a boundary point set, and calculate the normal vector field of the boundary point set using the local surface fitting method;

[0016] Perform a spatial coordinate transformation on the set of boundary points, establish a local coordinate system centered on the boundary points, orthogonally decompose the three-dimensional coordinate information of the boundary points and the normal vector field in the local coordinate system, and construct a boundary feature descriptor that includes position coordinate components and normal vector components.

[0017] Furthermore, perform density clustering analysis on the boundary feature descriptor, map the density clustering result to three-dimensional space to form a filling patch, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary point, and construct a neural network prediction model based on local geometric features, including:

[0018] Construct a similarity metric function for the boundary feature descriptor, calculate the distance matrix between feature descriptors based on the similarity metric function, use a density-based spatial clustering algorithm to divide the boundary feature descriptors into multiple clustering clusters, calculate the core points and boundary points for each clustering cluster, and project the spatial coordinates of the core points and boundary points back to three-dimensional space to form a filling patch;

[0019] Calculate the patch normal vector based on the grid vertices of the filling patch, establish an angle evaluation model between the filling patch normal vector and the boundary point normal vector, use the angle evaluation model as the input feature of the neural network, construct a multi-layer perceptron network structure, and optimize the neural network using the local geometric features of the boundary points as training samples.

[0020] Furthermore, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patch, including:

[0021] Apply the trained neural network model to the local geometric features of the filling patch, output the optimal angle threshold between the patch normal vector and the boundary point normal vector, screen the filling patches based on the optimal angle threshold, construct an index set of the patches to be optimized, calculate the principal curvature direction and curvature value of the boundary of the patches to be optimized, and establish an adjacency relationship graph between the patches;

[0022] Convert the boundary curvature information of the patches to be optimized into a discrete sampling point sequence, establish a parametric representation of the patch boundary based on the sampling point sequence, use the topological connection relationship to construct transition constraint conditions between the patches, and use the transition constraint conditions as constraint terms of the control instructions to generate a motion control instruction sequence that meets the continuity requirements.

[0023] Further, input the patch to be optimized into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline (NURBS) surface based on the boundary curvature information, applies continuity constraints to the NURBS surface using the topological connection relationship, and constructs an objective function considering the surface smoothness, including:

[0024] Convert the boundary curvature information of the patch to be optimized into a control point sequence, construct the knot vector of the NURBS surface based on the control point sequence, calculate the values of the NURBS basis functions on the knot vector, optimize the weights of the basis functions to obtain the mathematical expression of the non-uniform rational basis spline surface, and establish the mapping relationship between the surface parameter domain and the three-dimensional space;

[0025] Construct the position continuity and tangential continuity constraint conditions between adjacent patches according to the topological connection relationship of the patch to be optimized, combine the continuity constraint conditions with the mean curvature and Gaussian curvature of the surface to construct a smoothness evaluation function, and use the smoothness evaluation function as the optimization objective function.

[0026] Further, iteratively optimize the objective function to obtain an initial filling surface, calculate the transition region between the initial filling surface and the boundary of the original model, apply the Laplacian mesh smoothing algorithm in the transition region for local adjustment, generate a final filling model, and import the filling model into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection, including:

[0027] Numerically optimize the objective function using the gradient descent method, update the control point positions and weight parameters of the NURBS surface through iterative calculation until the objective function converges to obtain an initial filling surface that satisfies the continuity constraints, construct the distance field function between the initial filling surface and the boundary of the original model, and determine the range of the transition region based on the isocontours of the distance field function;

[0028] Construct the Laplacian operator of the grid vertices in the transition region, calculate the displacement vector of the grid vertices based on the Laplacian operator, iteratively update the spatial positions of the grid vertices to achieve local smoothing, merge the smoothed grid with the original model to form a complete filling model, and establish the spatial index structure of the filling model for path planning and collision detection.

[0029] In a second aspect, the present application provides an automated production line digital model hole filling device, including:

[0030] A boundary determination module, which is used to decompose the digital model of the automated production line into grid representations at multiple levels, calculate the local curvature field distribution at each grid level, project the curvature field onto the finest-level grid, calculate the change in curvature gradient based on the finest-level grid, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0031] A filling calculation module, which is used to perform density clustering analysis on the boundary feature descriptor, map the density clustering result to three-dimensional space to form filling patches, calculate the angle between the normal vector of the filling patches and the normal vector of the boundary points, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patches;

[0032] A surface filling module, which is used to input the patches to be optimized into a surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function considering surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, and performs local adjustment in the transition region using the Laplace grid smoothing algorithm to generate a final filling model, and imports the filling model into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection.

[0033] In a third aspect, the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the method for filling holes in the digital model of the automated production line are implemented.

[0034] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for filling holes in the digital model of the automated production line are implemented.

[0035] In a fifth aspect, the present application provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by a processor, the steps of the method for filling holes in the digital model of the automated production line are implemented.

[0036] As can be seen from the above technical solutions, this application provides a method and device for filling holes in the digital model of an automated production line. By means of multi-level grid analysis and curvature field calculation, the area to be processed can be accurately identified. An innovative patch generation mechanism based on density clustering and neural network is designed, and intelligent filling optimization is realized by combining local geometric features. The system adopts non-uniform rational basis spline surface reconstruction and Laplacian grid smoothing algorithms to ensure the continuity and smoothness of the filled area, and is closely combined with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0038] Figure 1 It is one of the schematic flowcharts of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0039] Figure 2 It is the second schematic flowchart of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0040] Figure 3 It is the third schematic flowchart of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0041] Figure 4 It is the fourth schematic flowchart of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0042] Figure 5 It is the fifth schematic flowchart of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0043] Figure 6 It is the sixth schematic flowchart of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0044] Figure 7 It is the seventh schematic flowchart of the method for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0045] Figure 8 It is the structural diagram of the device for filling holes in the digital model of the automated production line in the embodiment of the present application;

[0046] Figure 9Schematic diagram of the structure of the electronic device in the embodiments of the present application.

[0047] Reference numerals:

[0048] Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver program storage unit 9144, antenna 9111, speaker 9131, microphone 9132. Detailed implementation manners

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0050] In the technical solutions of the present application, the acquisition, storage, use, processing, etc. of data all comply with the relevant provisions of national laws and regulations.

[0051] Considering the problems existing in the prior art, the present application provides an automated production line digital model hole filling method and device. Through multi-level grid analysis and curvature field calculation, the area to be processed is accurately identified. An innovative patch generation mechanism based on density clustering and neural network is designed, and intelligent filling optimization is realized by combining local geometric features. The system adopts non-uniform rational basis spline surface reconstruction and Laplacian grid smoothing algorithm to ensure the continuity and smoothness of the filled area, and is closely combined with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0052] In order to be able to break through the limitations of traditional model repair and provide an efficient and reliable solution for the integrity reconstruction of industrial digital models, the present application provides an embodiment of an automated production line digital model hole filling method. Refer to Figure 1 The automated production line digital model hole filling method specifically includes the following contents:

[0053] Step S101: decomposing the digital model of the automated production line into multiple levels of grid representation, calculating the local curvature field distribution at each grid level, projecting the curvature field to the finest level grid, calculating the curvature gradient change based on the finest level grid, using an adaptive threshold segmentation algorithm to mark the area to be processed, extracting boundary points from the area to be processed, calculating the normal vector field of the boundary points, and combining the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0054] Optionally, in the process of mesh decomposition, this embodiment uses an adaptive octree structure for multi-level division. During the division process, a bounding box is first established for the entire automated production line model, and then the subspace division is dynamically adjusted according to the local geometric complexity. For complex areas such as the turns of the conveyor system, the joint structure of the robot arm, and the equipment connection surface, a recursive subdivision strategy is adopted until the grid size meets the preset accuracy requirements. For example, near the end effector of the robot arm, the grid size can reach the millimeter level due to the need for precise path planning and collision detection; while in simple structural areas such as large support frames, the centimeter-level grid size is maintained to achieve optimal allocation of computing resources.

[0055] In the curvature field calculation link, this embodiment implements an adaptive curvature estimation algorithm based on local fitting. For each mesh vertex, its topological neighborhood range is first determined, and the K nearest neighbor search method is used to select sample points involved in the fitting. Then, a local parameterized plane is constructed by the moving least squares method, and a quadratic surface fitting model is established. In the special-shaped surface areas of the production line, such as the transition fillet of the robot base, the guide groove of the conveyor belt, etc., the fitting window size is dynamically adjusted by analyzing the distribution characteristics of the local point cloud. For areas with drastic changes in curvature, the fitting window is reduced to improve accuracy; for gradual transition areas, the fitting window is expanded to enhance stability.

[0056] This embodiment develops a feature-preserving hierarchical projection algorithm during the curvature field projection process. First, the spatial correspondence between grid levels is established, and the projection mapping is determined using the inclusion relationship of the octree nodes. During the projection calculation, a weight function based on geodesic distance is introduced to ensure the accurate transmission of feature information between different resolutions. For example, at the installation interface of the control cabinet, a local feature decomposition strategy is used to accurately project the curvature information of the high-level grid to the finest level, avoiding the loss of important features during the multi-resolution conversion process.

[0057] In this embodiment, when calculating the curvature gradient, a gradient estimation method with a hybrid difference scheme is adopted. By constructing a difference operator on an irregular grid, the gradient calculation of the curvature field is realized. When dealing with the feature boundaries of the production line model, such as the mating surfaces between devices and the dividing lines of functional units, etc., a one-sided difference scheme is adopted to avoid numerical oscillations across the feature boundaries. For smooth transition regions, the central difference scheme is used to improve the calculation accuracy. At the same time, an adaptive step size control mechanism is introduced to automatically reduce the calculation step size at locations where the gradient changes violently, ensuring numerical stability.

[0058] In this embodiment, during the threshold segmentation process, an adaptive segmentation algorithm for multi-scale feature analysis is realized. First, a statistical analysis of the curvature gradient field is performed to construct a local histogram, and the initial threshold is determined through peak detection. Then, a region growing strategy is introduced to dynamically adjust the threshold according to the spatial continuity of the gradient. In the precision assembly areas of the production line, such as bearing seats and guide rails, etc., a smaller threshold interval is adopted to capture fine features; in the areas of large structural parts, a larger threshold interval is used to improve the processing efficiency.

[0059] In this embodiment, in the boundary point extraction step, a boundary tracking algorithm based on topological manifolds is developed. By analyzing the connection relationship of grid cells, a half-edge data structure is constructed to realize the automatic extraction of boundary loops. When dealing with the feature transition regions of the production line model, such as the connection between the conveyor belt and the bracket, the contact surface between the electrical control box and the bottom plate, etc., the curvature continuity constraint is adopted to ensure the smoothness of the boundary. For complex nodes with multiple connections, a branch processing mechanism is introduced to ensure the integrity and topological correctness of the boundary.

[0060] In this embodiment, during the normal vector calculation process, a robust estimation method based on tensor voting is realized. First, a principal direction analysis of the local neighborhood of the boundary points is performed to construct a covariance matrix. Then, the initial direction of the normal vector is determined through eigenvalue decomposition, and the tensor voting mechanism is used for direction optimization. In the key positioning surfaces of the production line, such as the installation reference surfaces of robotic arms and the installation surfaces of sensors, etc., flatness constraints are added to improve the accuracy of the normal vector. For the surface transition regions, the smooth change of the normal vector is ensured by adjusting the curvature weight.

[0061] In this embodiment, when constructing the feature descriptor, a multi-level feature encoding strategy is adopted. Geometric features such as the position information of boundary points, the normal vector field, and the principal curvature direction are organized into a hierarchical description structure. At the local feature encoding level, a feature aggregation method based on geodesic distance is adopted to ensure the sensitivity of the descriptor to local geometric changes. At the global feature level, long-range dependencies are captured through a feature propagation network, enabling the descriptor to express the functional associations between production line components.

[0062] Through the above implementation solution, a complete production line digital model feature analysis system is constructed in this embodiment. This solution can accurately identify and characterize various feature structures in the model, providing a reliable feature basis for subsequent hole filling. Practice shows that when dealing with complex automated production line digital models, this solution shows good adaptability and stability, and can effectively support the digital transformation and optimization and upgrading of the production line.

[0063] Step S102: Perform density clustering analysis on the boundary feature descriptor, map the density clustering result to a three-dimensional space to form a filling patch, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary point, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patch;

[0064] Optionally, in this embodiment, a clustering algorithm with an adaptive density threshold is developed in the density clustering analysis. For different types of defects in the automated production line, such as mesh fractures at the joints of robotic arms and model missing in the conveyor belt system, etc., the density threshold is dynamically adjusted. The algorithm first calculates the local density of each boundary feature descriptor, and determines the clustering core points by analyzing the gradient field of the density distribution. In the precision fitting areas of the production line, such as the contact surfaces of bearing seats and the sliding surfaces of guide rails, etc., a higher density threshold is adopted to ensure the accurate division of features.

[0065] In this embodiment, during the mapping process of the clustering result, a space reconstruction method based on feature preservation is realized. By establishing the mapping relationship between the feature descriptor space and the three-dimensional Euclidean space, the clustering boundary is converted into an actual geometric patch. During the mapping process, a local coordinate system transformation mechanism is introduced to ensure that the direction of the patch is consistent with the surface of the original model. For example, when dealing with the cylindrical hole of the robot base, by analyzing the local geometric features, the tangential continuity between the reconstructed patch and the cylindrical surface is ensured.

[0066] In this embodiment, in the calculation link of the normal vector angle, a weighted geodesic distance evaluation method is adopted. For each filling patch, a geodesic path with the boundary point is constructed, and the gradual change relationship of the normal vector is calculated along the path. When dealing with the transition areas of production line equipment, such as the connection between the control cabinet and the bracket, the sensor installation groove, etc., by introducing the curvature weight, the smoothness of the normal vector transition is ensured.

[0067] In the construction of the neural network model in this embodiment, a multi-level feature extraction structure is designed. The input layer receives local geometric features, including geometric quantities such as principal curvature values, Gaussian curvature, and mean curvature, as well as the topological relationships of adjacent regions. The hidden layer performs feature transformation through a multi-layer perceptron to gradually extract high-level semantic features. During the training process, typical samples from different parts of the production line, such as plane contact, arc transition, and special-shaped surfaces, are used to ensure the generalization ability of the model.

[0068] In the prediction of the optimal angle threshold in this embodiment, feature aggregation based on the attention mechanism is realized. The model dynamically adjusts the feature weights by analyzing the importance distribution of local geometric features. For key functional surfaces in the production line, such as positioning reference surfaces and assembly contact surfaces, the model assigns higher feature weights, thereby generating more stringent angle threshold constraints.

[0069] In the process of marking the patches to be optimized in this embodiment, a multi-level screening mechanism is developed. First, a preliminary screening is performed based on the predicted angle threshold, and then a refined screening is carried out in combination with local curvature continuity and topological consistency. When dealing with precision mating components of the production line, such as the cylindrical surface of the bearing seat and the sliding surface of the guide rail, more stringent screening criteria are adopted to ensure the geometric accuracy of the filled patches.

[0070] In the link of boundary curvature calculation in this embodiment, an adaptive grid subdivision strategy is adopted. By analyzing the gradient of curvature change, local subdivision is performed in the region of rapid curvature change to improve the accuracy of curvature estimation. For the feature transition regions in the production line, such as fillet transitions and chamfers, the continuity of curvature information is ensured through curvature tensor analysis.

[0071] In the construction of the topological connection relationship in this embodiment, an intelligent connection mechanism based on feature recognition is realized. By analyzing the positional relationship and geometric features between the patches to be optimized, the patch combinations related to functions are automatically recognized. When dealing with the model defects of the conveyor belt system, the movement direction of the conveyor belt and the functional associations of adjacent components can be accurately recognized.

[0072] In the process of generating motion control instructions in this embodiment, a trajectory planning algorithm based on geometric constraints is developed. By analyzing the geometric features and topological relationships of the filled patches, a motion trajectory that meets the continuity requirements is generated. For operations that require precise positioning, such as the grasping points of the robotic arm and the detection positions of vision sensors, the accuracy of the control instructions is ensured through feature point extraction.

[0073] Through the above implementation solutions, this embodiment constructs a complete model repair and control instruction generation system. This solution can effectively handle various defects in the digital model of the automated production line and generate accurate motion control instructions. Practice shows that this solution exhibits good adaptability and reliability in the actual operation and maintenance of the production line, and can effectively support the automated operation and precise control of the production line.

[0074] Step S103: Input the patch to be optimized into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational B-spline surface based on the boundary curvature information, applies continuity constraints to the B-spline surface using the topological connection relationship, constructs an objective function considering the surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, applies the Laplace mesh smoothing algorithm for local adjustment in the transition region to generate a final filling model, and imports the filling model into the 3D visualization system of the automated production line for material handling path planning and equipment collision detection.

[0075] Optionally, in the process of constructing the non-uniform rational B-spline surface in this embodiment, an adaptive knot vector generation algorithm is developed. For different types of damaged areas in the automated production line, such as the surface transition at the robotic arm joint and the guiding groove of the conveyor belt system, the distribution of control points is dynamically adjusted. By analyzing the boundary curvature distribution characteristics, the control point density is increased at the locations where the curvature changes drastically to ensure the accurate expression of local geometric features by the reconstructed surface.

[0076] In the control point weight optimization step of this embodiment, a weight adjustment mechanism based on geometric features is realized. By analyzing the principal curvature direction and curvature value of the damaged area, an initial weight is assigned to each control point. At critical positions on the precision assembly surfaces of the production line, such as the mating surface of the bearing seat and the sliding surface of the guide rail, a larger weight value is used to improve the local control accuracy of the surface. A smaller weight value is used for the transition region to achieve smooth transition.

[0077] In the construction of continuity constraints in this embodiment, a hierarchical constraint system is adopted. First, a position continuity constraint is established to ensure the accurate connection between the reconstructed surface and the boundary of the original model. Then, a tangential continuity constraint is introduced to ensure the smoothness of the surface transition. When dealing with the functional surfaces of production line equipment, such as the installation surface of the vision sensor and the grasping surface of the fixture, a normal continuity constraint also needs to be satisfied to ensure the accuracy of subsequent operations.

[0078] In the design of the smoothness objective function in this embodiment, a multi-objective trade-off mechanism is realized. The objective function includes a geometric error term, a surface energy term, and a topological consistency term. The geometric error term ensures the fitting accuracy between the reconstructed surface and the boundary points, the surface energy term controls the smoothness of the surface, and the topological consistency term guarantees the continuous transition between adjacent patches. During the optimization process, the weights of each term are dynamically adjusted according to the functional requirements of different parts of the production line.

[0079] In this embodiment, during the iterative optimization process, an adaptive step size control strategy is developed. By analyzing the gradient information of the objective function, the optimization step size is dynamically adjusted. A larger step size is used at the initial stage of optimization to quickly approach the solution space, and the step size is gradually reduced as the optimization process progresses to improve the accuracy. For precision mating components in the production line, more stringent convergence criteria are adopted to ensure the geometric accuracy of the reconstructed surface.

[0080] In this embodiment, during the calculation of the transition region, a distance field analysis method is adopted. By constructing a distance field function between the initial filling surface and the boundary of the original model, the range of the transition region is determined based on contour line analysis. When determining the transition region, the functional characteristics of the production line equipment are considered, such as the clearance requirements of moving parts and the accuracy requirements of assembly benchmarks, etc., and the range of the transition region is reasonably set.

[0081] In this embodiment, during the mesh smoothing process, a Laplacian operator that preserves features is implemented. By analyzing the local geometric features of mesh vertices, a weighted Laplacian operator is constructed. During the smoothing process, a smaller smoothing intensity is used for feature edges and feature points that are functionally important, while a larger smoothing intensity can be used for general transition regions to achieve selective smoothing.

[0082] In this embodiment, when generating the final filling model, a mesh quality optimization algorithm is developed. By analyzing the shape quality indicators of mesh elements, such as aspect ratio, included angle, etc., local optimization and adjustment of the mesh are performed. In key functional areas of the production line, such as the movement trajectory area of the robotic arm and the detection area of the sensor, etc., ensure that the mesh quality meets the requirements of subsequent simulation analysis.

[0083] In this embodiment, during the integration of the 3D visualization system, a spatial index structure based on octree is implemented. By establishing a hierarchical spatial partition of the filling model, the computational efficiency of collision detection and path planning is improved. When dealing with the motion planning of the robotic arm, potential collision areas can be quickly located and the optimal motion trajectory can be generated.

[0084] Through the above implementation scheme, this embodiment constructs a complete model reconstruction and optimization system. This scheme is particularly suitable for the repair and optimization of digital models of automated production lines, can accurately reconstruct various defect areas, and ensure the geometric accuracy and functional integrity of the repaired model. Practice shows that this scheme reliably supports the virtual commissioning and motion planning of the production line, providing strong digital support for intelligent manufacturing.

[0085] As can be seen from the above description, the automated production line digital model hole filling method provided by the embodiments of the present application can accurately identify the area to be processed through multi-level grid analysis and curvature field calculation. An innovative patch generation mechanism based on density clustering and neural network is designed to achieve intelligent filling optimization in combination with local geometric features. The system adopts non-uniform rational basis spline surface reconstruction and Laplacian grid smoothing algorithms to ensure the continuity and smoothness of the filled area and is closely integrated with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0086] In an embodiment of the automated production line digital model hole filling method of the present application, referring to Figure 2 , it may specifically include the following content:

[0087] Step S201: Use the octree decomposition algorithm to divide the automated production line digital model into grid levels of different scales, construct a vertex adjacency relationship graph for each grid level, calculate the Gaussian curvature and mean curvature of the grid vertices based on the adjacency relationship graph, and combine the Gaussian curvature and mean curvature to form a local curvature field distribution;

[0088] Step S202: Use the interpolation mapping algorithm between grid levels to project the curvature fields of different levels onto the finest level grid, construct a curvature gradient tensor field, calculate the main direction and eigenvalues of the curvature gradient tensor field, construct an adaptive threshold segmentation function based on the eigenvalues, and mark the output of the segmentation function as the area to be processed.

[0089] Optionally, in the octree decomposition process of this embodiment, an adaptive partitioning strategy based on geometric features is implemented. For different functional units in the automated production line, such as the joint structure of the robotic arm, the guiding device of the conveyor belt, the installation surface of the control cabinet, etc., the grid division accuracy is dynamically adjusted. In the precise operation areas of the production line, such as the grasping position of the robot fixture, the detection area of the vision sensor, etc., a finer grid division is adopted to ensure the accurate capture of geometric features.

[0090] In the construction of grid levels in this embodiment, a multi-resolution topology maintenance mechanism is developed. By establishing the parent-child relationship between octree nodes, an orderly organization between grid levels is achieved. For complex transition areas in the production line, such as equipment connections, interfaces of functional components, etc., geometric details are maintained through recursive subdivision, and at the same time, the accuracy and efficiency are balanced within the allowable range of computing resources.

[0091] In the process of constructing the vertex adjacency relationship graph in this embodiment, a dynamically updated data structure is adopted. For each grid level, by analyzing the topological connection relationship between vertices, an efficient adjacency query mechanism is established. When dealing with the complex curved surfaces of production line equipment, such as the shell structure of the robotic arm and the guiding groove of the conveyor belt, the integrity and correctness of the adjacency relationship are ensured.

[0092] In the curvature calculation link of this embodiment, a curvature estimation algorithm based on local fitting is realized. By analyzing the neighborhood structure of each grid vertex, the moving least squares method is used to fit the local surface. In the characteristic transition regions of the production line, such as fillet transitions and chamfers, by adaptively adjusting the size of the fitting window, the accuracy of curvature calculation is ensured.

[0093] When constructing the local curvature field in this embodiment, a curvature synthesis method that preserves features is developed. By combining the Gaussian curvature and the mean curvature, a complete curvature descriptor is constructed. When dealing with the special-shaped surfaces of the production line, such as the mounting surface of the robot base and the heat dissipation holes of the control cabinet, the change law of local geometric features can be accurately expressed.

[0094] In the process of curvature field projection in this embodiment, a hierarchical mapping algorithm that preserves features is realized. By constructing the spatial correspondence relationship between grids with different resolutions and adopting an interpolation strategy based on distance weights, the accurate transmission of curvature information during the hierarchical conversion process is ensured. For the precision fitting structures in the production line, such as the mating surface of the bearing seat and the sliding surface of the guide rail, special attention is paid to maintaining the continuity of geometric features.

[0095] When calculating the curvature gradient tensor field in this embodiment, a robust numerical differentiation method is adopted. By constructing a local coordinate system and analyzing the change rate of curvature in different directions, a complete gradient tensor description is formed. When dealing with the characteristic boundaries of production line equipment, such as the motion limit of the robotic arm and the guiding edge of the conveyor belt, an adaptive step size strategy is adopted to ensure the accuracy of gradient calculation.

[0096] In the process of principal direction analysis in this embodiment, a direction extraction algorithm based on eigenvalue decomposition is developed. By analyzing the eigenstructure of the gradient tensor, the main direction of local geometric change is determined. When dealing with the functional surfaces of the production line, such as the operating surface of the robot tool end and the detection surface of the vision sensor, accurately identifying the characteristic direction is of great significance for subsequent operation planning.

[0097] When constructing the threshold segmentation function in this embodiment, a multi-scale feature analysis mechanism is realized. By comprehensively considering the eigenvalue distribution of the curvature gradient, the segmentation threshold is dynamically adjusted. When dealing with different types of defects in the production line, such as model fractures and mesh missing, an adaptive threshold strategy is adopted to ensure the rationality of the segmentation result.

[0098] This embodiment establishes a complete geometric feature analysis system through the above implementation scheme. The scheme can accurately identify abnormal areas in the digital model of the automated production line and provide a reliable feature basis for subsequent model repair. Practice has shown that the scheme has good adaptability and stability when dealing with complex industrial scenarios and can effectively support the digital optimization and transformation of production lines.

[0099] In practical applications, the solution of this embodiment successfully handles various types of production line model defects, such as mesh breakage at the joints of the robotic arm and missing models of the conveyor belt system. It provides a reliable geometric basis for the virtual debugging and motion planning of the production line, and effectively improves the digitalization level of the automated production line.

[0100] In one embodiment of the method for filling holes in the digital model of the automated production line of the present application, see Figure 3 , and can also include the following:

[0101] Step S301: constructing a boundary tracking algorithm for the area to be processed, searching for connected boundaries of the area to be processed based on the mesh topological connection relationship, extracting mesh vertices on the connected boundaries as a boundary point set, and using a local surface fitting method to calculate the normal vector field of the boundary point set;

[0102] Step S302: perform spatial coordinate transformation on the boundary point set, establish a local coordinate system centered on the boundary point, orthogonally decompose the three-dimensional coordinate information of the boundary point and the normal vector field in the local coordinate system, and construct a boundary feature descriptor including position coordinate components and normal vector components.

[0103] Optionally, in the process of constructing the boundary tracking algorithm, this embodiment implements a boundary recognition mechanism based on topological manifolds. An adaptive boundary search strategy is developed for different types of defective areas in the automated production line, such as mesh fractures at the joints of the robot arm, missing models of the conveyor belt system, and deformation of the mounting surface of the control cabinet. By analyzing the connection relationship between grid cells and constructing a half-edge data structure, fast positioning and tracking of the boundary can be achieved.

[0104] This embodiment adopts a hierarchical and progressive search strategy in the process of searching for connected boundaries. First, the starting point of the boundary is identified based on the topological connection relationship of the grid, and then it is gradually expanded along the boundary direction until the construction of the closed boundary is completed. When dealing with complex boundaries of production line equipment, such as the special-shaped surface of the robot tool end and the internal contour of the sensor installation slot, the accuracy of boundary tracking is ensured by directional continuity constraints.

[0105] In the boundary point extraction stage of this embodiment, a sampling algorithm that preserves features is developed. By analyzing the boundary curvature distribution, the sampling density is increased at locations with drastic curvature changes, while the sampling frequency is appropriately reduced in flat regions. For precision mating components in the production line, such as the cylindrical surface of a bearing housing and the sliding surface of a guide rail, a denser sampling strategy is adopted to ensure the complete retention of geometric features.

[0106] In the process of calculating the normal vector in this embodiment, a robust estimation method based on local fitting is implemented. For each boundary point, a local neighborhood range is constructed, and the local surface is fitted by the moving least squares method. When dealing with feature transition regions in the production line, such as fillet transitions and chamfers, the accuracy of normal vector calculation is ensured by adaptively adjusting the size of the fitting window.

[0107] In the process of establishing the local coordinate system in this embodiment, a construction method based on principal direction analysis is adopted. By analyzing the spatial distribution characteristics of the boundary point set, the main direction of change is determined as the coordinate axis. When dealing with functional surfaces in the production line, such as the motion reference surface of a robotic arm and the detection surface of a vision sensor, a reasonable selection of the coordinate system direction has an important impact on subsequent feature description.

[0108] In the implementation of coordinate transformation in this embodiment, a transformation algorithm that preserves features is developed. By constructing a rigid body transformation matrix, the boundary point information in the global coordinate system is mapped to the local coordinate system. During the transformation process, special attention is paid to maintaining the invariance of geometric features to ensure that the boundary features can be accurately expressed in the local coordinate system.

[0109] In the orthogonal decomposition stage of this embodiment, a multi-scale feature decomposition mechanism is implemented. The position information and normal vector information of the boundary points are projected onto the orthogonal basis of the local coordinate system to form a complete feature representation. When dealing with different types of features in the production line, such as planes, cylindrical surfaces, and free-form surfaces, the key geometric features are highlighted through reasonable component weight configuration.

[0110] In the construction of the feature descriptor in this embodiment, a hierarchical feature coding strategy is adopted. The position components and normal vector components obtained from the orthogonal decomposition are organized into a structured description vector. For key functional components in the production line, such as positioning references and assembly interfaces, their importance is highlighted by increasing the feature weights.

[0111] In the process of normalizing the descriptor in this embodiment, an adaptive normalization method is developed. By analyzing the statistical distribution of the feature components, the normalization parameters are dynamically adjusted. When dealing with production line components of different scales, such as large support frames and precision connectors, a hierarchical normalization strategy is adopted to ensure the balance of feature representation.

[0112] Through the above implementation solution, a complete boundary feature extraction and description system is established in this embodiment. This solution can accurately capture the boundary features in the digital model of the automated production line, providing a reliable feature basis for subsequent model repair and optimization. Practice shows that this solution exhibits good adaptability and stability when dealing with complex industrial scenarios.

[0113] In practical applications, the solution of this embodiment has successfully addressed the problem of extracting boundary features of various types of production line model defects, such as complex curved surfaces at the joints of robotic arms and guiding grooves of conveyor belt systems, providing reliable geometric feature support for the digital optimization and transformation of production lines, and effectively improving the digital twin level of automated production lines.

[0114] In an embodiment of the method for filling holes in the digital model of an automated production line of this application, refer to Figure 4 , and it may specifically include the following content:

[0115] Step S401: Construct a similarity metric function for boundary feature descriptors, calculate the distance matrix between feature descriptors based on the similarity metric function, use a density-based spatial clustering algorithm to divide the boundary feature descriptors into multiple clustering clusters, calculate the core points and boundary points for each clustering cluster, and project the spatial coordinates of the core points and boundary points back to the three-dimensional space to form filling patches;

[0116] Step S402: Calculate the patch normal vector based on the mesh vertices of the filling patch, establish an angle evaluation model between the filling patch normal vector and the boundary point normal vector, use the angle evaluation model as the input feature of the neural network, construct a multi-layer perceptron network structure, and optimize the neural network with the local geometric features of the boundary points as training samples.

[0117] Optionally, in the process of constructing the similarity metric function in this embodiment, a distance calculation mechanism for multi-feature fusion is developed. For different types of boundary features in the automated production line, such as the joint structure of the robotic arm, the guiding groove of the conveyor belt, and the installation surface of the control cabinet, a weighted Euclidean distance metric is constructed by comprehensively considering the differences in position coordinate components and normal vector components. When dealing with precision mating components of the production line, such as the cylindrical surface of the bearing seat and the sliding surface of the guide rail, the influence of key geometric features is highlighted by dynamically adjusting the feature weights.

[0118] In this embodiment, in the distance matrix calculation link, an efficient calculation strategy based on parallel computing is implemented. By constructing a spatial index structure of feature descriptors, the search efficiency of distance calculation is optimized. When dealing with large-scale production line models, such as complex assembly line layouts and multi-robot collaborative workstations, it can efficiently handle the distance relationships between a large number of feature points.

[0119] In the implementation of the density clustering algorithm in this embodiment, a clustering strategy with an adaptive density threshold is adopted. By analyzing the point density distribution in the feature space, the clustering parameters are dynamically determined. When processing the feature regions of production line equipment, such as the irregular surfaces at the tool ends of robots and the internal contours of sensor mounting grooves, it is possible to accurately identify the region combinations with similar geometric features.

[0120] In the core point extraction step of this embodiment, an evaluation method based on topological importance is developed. By analyzing the density distribution and connection relationships within the clustering clusters, representative core feature points are identified. When processing the key functional surfaces of the production line, such as the positioning reference surface and the assembly contact surface, special attention is paid to maintaining the representativeness of geometric features.

[0121] In the spatial projection process of this embodiment, a mapping algorithm that preserves features is implemented. By constructing the mapping relationship between the feature space and the three-dimensional Euclidean space, it is ensured that the projection results meet the requirements of geometric continuity. When processing the feature transition regions in the production line, such as fillet transitions and chamfers, the smoothness of the projected patches is ensured through curvature constraints.

[0122] In the calculation of the patch normal vector in this embodiment, a robust normal vector estimation method is adopted. By analyzing the local distribution characteristics of the patch vertices, a weighted least squares fitting plane is constructed to extract reliable normal vector information. When processing the precision surfaces of production line equipment, such as the detection surface of vision sensors and the grasping surface of fixtures, special attention is paid to the accuracy of normal vector calculation.

[0123] In the construction process of the angle evaluation model in this embodiment, a multi-scale feature analysis mechanism is developed. By comprehensively considering multiple evaluation indicators such as the normal vector angle, curvature similarity, and position continuity, a complete evaluation system is constructed. When processing the complex curved surfaces of the production line, such as the housing structure of the robotic arm and the guiding device of the conveyor belt, it is possible to accurately evaluate the geometric rationality of the filled patches.

[0124] In the neural network construction step of this embodiment, a deep learning model based on local geometric features is implemented. The network input layer receives multi-dimensional features such as the normal vector angle, curvature features, and topological relationships, and performs feature transformation and fusion through a multi-layer perceptron. During the training process, typical samples from different parts of the production line, such as plane contact, arc transition, and irregular surfaces, are used to ensure the generalization ability of the model.

[0125] In the network optimization process of this embodiment, a dynamic learning rate adjustment strategy is adopted. By analyzing the feature distribution of the training samples, the optimization parameters are adaptively adjusted. For the key functional components in the production line, such as the positioning reference and the assembly interface, the sensitivity of the model to key features is improved by increasing the sample weights.

[0126] Through the above implementation solution, a complete feature clustering and evaluation system is established in this embodiment. This solution can accurately identify and evaluate the filling areas in the digital model of the automated production line, providing a reliable quality evaluation basis for subsequent model optimization. Practice shows that this solution shows good adaptability and stability when dealing with complex industrial scenarios.

[0127] In practical applications, the solution of this embodiment has successfully handled various types of filling quality evaluation problems of production line models, such as complex curved surfaces at the joints of robotic arms and guiding grooves of conveyor belt systems, providing reliable quality assurance for the digital optimization and transformation of production lines, and effectively improving the digital twin accuracy of automated production lines.

[0128] In an embodiment of the method for filling holes in the digital model of the automated production line of the present application, refer to Figure 5 , and it may specifically include the following content:

[0129] Step S501: Apply the trained neural network model to the local geometric features of the filling patch, output the optimal angle threshold between the patch normal vector and the boundary point normal vector, screen the filling patches based on the optimal angle threshold, construct an index set of the patches to be optimized, calculate the principal curvature direction and curvature value of the boundary of the patches to be optimized, and establish an adjacency relationship graph between the patches;

[0130] Step S502: Convert the boundary curvature information of the patches to be optimized into a discrete sampling point sequence, establish a parametric representation of the patch boundary based on the sampling point sequence, construct transition constraint conditions between the patches using topological connection relationships, and use the transition constraint conditions as constraint terms of the control command to generate a motion control command sequence that meets the continuity requirements.

[0131] Optionally, in the process of predicting the optimal angle threshold in this embodiment, the trained neural network is used for inference calculation. The network inputs include the local geometric features of the filling patch, such as principal curvature value, Gaussian curvature, average curvature, etc., and the position relationship features between the patch and the boundary point. When dealing with different functional components of the automated production line, such as the joint structure of the robotic arm and the guiding groove of the conveyor belt, the model can dynamically adjust the threshold prediction strategy according to the specific geometric features.

[0132] In the patch screening link of this embodiment, a multi-level screening mechanism is implemented. First, a preliminary screening is performed based on the predicted angle threshold, and then a refined screening is performed in combination with local curvature continuity and topological consistency. When dealing with precision mating components of the production line, such as the mating surface of the bearing seat and the sliding surface of the guide rail, more stringent screening criteria are adopted to ensure the geometric accuracy of the filling patches.

[0133] In this embodiment, a robust curvature estimation algorithm is developed during the main curvature calculation process. By constructing a local quadratic surface fitting, the curvature changes of the surface in different directions are analyzed. When dealing with the feature transition regions of production line equipment, such as fillet transitions, chamfers, etc., the fitting window size is adaptively adjusted to ensure the accuracy of curvature calculation.

[0134] In this embodiment, an intelligent connection mechanism based on feature recognition is adopted during the construction of the adjacency relationship graph. By analyzing the positional relationship and geometric features between the patches to be optimized, the patch combinations related to functions are automatically recognized. When dealing with the model defects of the conveyor belt system, the movement direction of the conveyor belt and the functional associations of adjacent components can be accurately recognized.

[0135] In this embodiment, an adaptive sampling strategy is implemented during the discretization process of curvature information. By analyzing the gradient of curvature changes, the sampling density is increased in the regions of drastic curvature changes, while the sampling frequency is appropriately reduced in the flat regions. For the feature transition regions in the production line, such as the special-shaped surfaces at the tool ends of robots, the installation grooves of control cabinets, etc., it is ensured that the sampling point sequence can accurately represent the geometric features.

[0136] In this embodiment, a feature-preserving parameterization method is developed during the construction of the parameterization representation. By analyzing the spatial distribution characteristics of the sampling point sequence, a suitable parameterization basis function is constructed. When dealing with the complex surfaces of the production line, such as the housing structure of the robotic arm, the installation surface of the sensor, etc., the geometric accuracy of the parameterization representation is ensured through curvature constraints.

[0137] In this embodiment, a hierarchical constraint system is adopted during the construction of the transition constraint conditions. First, the position continuity constraint is established to ensure the precise connection between adjacent patches. Then, the tangential continuity constraint is introduced to ensure the smoothness of the transition. When dealing with the functional surfaces of production line equipment, such as the detection surface of the vision sensor, the grasping surface of the fixture, etc., the normal continuity constraint also needs to be satisfied.

[0138] In this embodiment, a trajectory planning algorithm based on geometric constraints is implemented during the generation of motion control instructions. By analyzing the geometric features and topological relationships of the patches, a motion trajectory that meets the continuity requirements is generated. For operations that require precise positioning, such as the grasping points of the robotic arm, the detection positions of the vision sensor, etc., the accuracy of the control instructions is ensured through feature point extraction.

[0139] In this embodiment, an optimization strategy considering dynamic constraints is developed during the optimization of the instruction sequence. By analyzing the acceleration and speed limits of the moving parts, the control instructions are dynamically adjusted. When dealing with the motion planning of robots, special attention is paid to maintaining the smoothness and continuity of the motion.

[0140] Through the above implementation solution, this embodiment constructs a complete geometric optimization and motion control system. This solution can accurately identify and optimize problem areas in the digital model of the automated production line and generate reliable motion control instructions. Practice shows that this solution shows good adaptability and stability when dealing with complex industrial scenarios.

[0141] In practical applications, the solution of this embodiment has successfully handled various types of production line motion control problems, such as the precise positioning of the robotic arm, the speed control of the conveyor belt, etc., providing a reliable control basis for the automated operation of the production line and effectively improving the operation efficiency and accuracy of the automated production line.

[0142] In an embodiment of the method for filling holes in the digital model of the automated production line of this application, refer to Figure 6 , it may specifically include the following content:

[0143] Step S601: Convert the boundary curvature information of the patch to be optimized into a control point sequence, construct a knot vector of the NURBS surface based on the control point sequence, calculate the values of the NURBS basis functions on the knot vector, optimize the weights of the basis functions to obtain the mathematical expression of the non-uniform rational B-spline surface, and establish the mapping relationship between the surface parameter domain and the three-dimensional space;

[0144] Step S602: Construct the position continuity and tangential continuity constraint conditions between adjacent patches according to the topological connection relationship of the patch to be optimized, combine the continuity constraint conditions with the mean curvature and Gaussian curvature of the surface to construct a fairness evaluation function, and use the fairness evaluation function as the optimization objective function.

[0145] Optionally, in the process of generating the control point sequence in this embodiment, an adaptive sampling strategy based on curvature features is adopted. For different types of functional surfaces in the automated production line, such as the joint structure of the robotic arm, the guiding groove of the conveyor belt, the installation surface of the control cabinet, etc., the control point density is dynamically adjusted by analyzing the boundary curvature distribution. Increase the number of control points at the locations where the curvature changes violently to ensure accurate capture of local geometric features.

[0146] In the link of constructing the knot vector in this embodiment, a knot distribution algorithm based on geometric features is realized. By analyzing the spatial distribution characteristics of the control point sequence, the cumulative chord length parameterization method is used to determine the knot positions. When dealing with the precision mating parts of the production line, such as the cylindrical surface of the bearing seat, the sliding surface of the guide rail, etc., the local control accuracy of the surface is improved by refining the knot distribution.

[0147] In this embodiment, a numerically stable recursive algorithm is developed during the calculation of basis functions. By constructing the recurrence relation of the basis functions, the calculation stability is ensured when the multiplicity of the knot vector is relatively high. When dealing with the feature transition regions of production line equipment, such as fillet transitions, chamfers, etc., the order of the basis functions is adjusted to balance the accuracy and smoothness.

[0148] In the weight optimization process of this embodiment, an optimization strategy considering geometric features is adopted. By analyzing the local geometric features of the control points, an initial weight value is assigned to each control point. When dealing with key positions such as the precision operation surface of the robot tool end and the detection surface of the vision sensor, a larger weight value is adopted to improve the local control accuracy.

[0149] When constructing the mapping relationship in this embodiment, a parameterization method that preserves features is realized. By constructing a bi-directional mapping relationship between the parameter domain and the three-dimensional space, the uniformity and stability of surface parameterization are ensured. When dealing with complex surfaces in the production line, such as the housing structure of the robotic arm and the guiding device of the conveyor belt, the isometric mapping is used to maintain the uniform distribution of geometric features.

[0150] In the process of constructing the continuity constraint in this embodiment, a hierarchical constraint system is adopted. First, the position continuity constraint is established to ensure the precise connection between adjacent patches. Then, the tangential continuity constraint is introduced to ensure the smoothness of the transition. When dealing with the functional surfaces of the production line, such as the positioning reference surface and the assembly contact surface, the geometric continuity requirements are ensured through the constraint conditions.

[0151] In the mean curvature calculation process of this embodiment, a curvature estimation algorithm based on local fitting is developed. By constructing a local parametric surface, the principal curvature changes of the surface in different directions are analyzed. When dealing with the feature regions of production line equipment, such as the installation surface of the robot base and the heat dissipation holes of the control cabinet, the local geometric features of the surface are accurately calculated.

[0152] In the process of calculating the Gaussian curvature in this embodiment, a robust numerical calculation method is realized. By analyzing the basic shape features of the surface, a suitable calculation window size is constructed. When dealing with irregular surfaces in the production line, such as the installation groove of the sensor and the grasping surface of the fixture, the accuracy and stability of the curvature calculation are ensured.

[0153] When constructing the fairness evaluation function in this embodiment, a multi-objective trade-off evaluation mechanism is adopted. By comprehensively considering the continuity constraint, mean curvature, and Gaussian curvature, a complete evaluation system is constructed. During the optimization process, the weights of each item are dynamically adjusted according to the functional requirements of different parts of the production line to achieve the balanced optimization of geometric features.

[0154] Through the above implementation solutions, this embodiment has established a complete surface reconstruction and optimization system. This solution can accurately reconstruct the problem areas in the digital model of the automated production line and ensure the geometric quality of the reconstructed surface. Practice has shown that this solution exhibits good adaptability and stability when dealing with complex industrial scenarios.

[0155] In practical applications, the solution of this embodiment has successfully handled various types of production line model reconstruction problems, such as complex surfaces at the joints of robotic arms and guiding grooves of conveyor belt systems, providing a reliable geometric basis for the digital optimization and transformation of production lines and effectively improving the digital twin accuracy of automated production lines.

[0156] In an embodiment of the method for filling holes in the digital model of an automated production line of the present application, refer to Figure 7 , and it may specifically include the following content:

[0157] Step S701: Numerically optimize the objective function using the gradient descent method, update the control point positions and weight parameters of the NURBS surface through iterative calculations until the objective function converges to obtain an initial filling surface that satisfies the continuity constraint, construct a distance field function between the initial filling surface and the boundary of the original model, and determine the range of the transition region based on the isocontours of the distance field function;

[0158] Step S702: Construct the Laplace operator of the grid vertices in the transition region, calculate the displacement vectors of the grid vertices based on the Laplace operator, iteratively update the spatial positions of the grid vertices to achieve local smoothing, merge the smoothed grid with the original model to form a complete filling model, and establish a spatial index structure of the filling model for path planning and collision detection.

[0159] Optionally, in the gradient descent optimization process of this embodiment, an iterative strategy with an adaptive step size is adopted. For different types of functional surfaces in the automated production line, such as the joint structure of the robotic arm, the guiding groove of the conveyor belt, and the installation surface of the control cabinet, the optimization step size is dynamically adjusted by analyzing the gradient change of the objective function. Increase the step size in areas with slow convergence speed and decrease the step size in sensitive areas to ensure the stability of the optimization process.

[0160] In the control point update link of this embodiment, a constraint-based position adjustment mechanism is implemented. By analyzing the continuity constraint conditions, an update rule for the control point positions is constructed. When dealing with precision mating components of the production line, such as the cylindrical surface of the bearing seat and the sliding surface of the guide rail, special attention is paid to maintaining the continuity of geometric features.

[0161] In the process of optimizing the weight parameters in this embodiment, a weight adjustment strategy considering local features is developed. By analyzing the local geometric features of the surface, the weight values of the control points are dynamically adjusted. When dealing with the feature transition regions of production line equipment, such as fillet transitions, chamfers, etc., the smooth transition of the surface is ensured through weight optimization.

[0162] When constructing the distance field function in this embodiment, the fast marching algorithm is used to calculate the distance field. By constructing the distance mapping relationship between the initial filling surface and the boundary of the original model, the shortest distance from any point in space to the boundary is efficiently calculated. When dealing with key positions such as the precise operation surface of the robot tool end and the detection surface of the vision sensor, the accuracy of the distance field calculation is ensured.

[0163] In the link of determining the transition region in this embodiment, an adaptive region division strategy is implemented. By analyzing the contour line distribution of the distance field, the range of the transition region is dynamically determined. When dealing with complex surfaces in the production line, such as the housing structure of the robotic arm and the guiding device of the conveyor belt, the size of the transition region is adjusted according to the local geometric features.

[0164] When constructing the Laplace operator in this embodiment, a weight configuration method based on geometric features is adopted. By analyzing the local neighborhood structure of the mesh vertices, a Laplace matrix considering geometric features is constructed. When dealing with the functional surfaces of the production line, such as the positioning reference surface and the assembly contact surface, the key features are maintained through weight adjustment.

[0165] In the link of calculating the displacement vector in this embodiment, a smooth algorithm for maintaining features is developed. By analyzing the eigenstructure of the Laplace operator, the optimal displacement direction of the mesh vertices is calculated. When dealing with the feature regions of production line equipment, such as the installation surface of the robot base and the heat dissipation holes of the control cabinet, it is ensured that the original geometric features are not damaged during the smoothing process.

[0166] In the process of updating the mesh vertices in this embodiment, an iterative position optimization strategy is implemented. By gradually adjusting the spatial positions of the mesh vertices, a smooth transition of the local region is achieved. When dealing with irregular surfaces in the production line, such as the installation slots of sensors and the grasping surfaces of fixtures, the smooth effect is ensured through multiple iterations.

[0167] In the model merging link of this embodiment, a merging strategy with topological consistency is adopted. By analyzing the boundary relationship between the filling model and the original model, a reliable model connection is constructed. During the optimization process, special attention is paid to maintaining the topological integrity of the model to ensure the geometric continuity of the merged model.

[0168] In the construction of the spatial index in this embodiment, a multi-level index structure is developed. By establishing a spatial partition tree for the filling model, efficient spatial query operations are supported. When performing path planning and collision detection, such as robot motion trajectory planning and equipment layout optimization, potential collision areas can be quickly located.

[0169] Through the above implementation solution, this embodiment establishes a complete model optimization and index system. This solution can accurately optimize the problem areas in the digital model of the automated production line and provide efficient spatial retrieval support. Practice shows that this solution shows good adaptability and stability when dealing with complex industrial scenarios.

[0170] In practical applications, the solution of this embodiment has successfully handled various types of production line model optimization problems, such as complex curved surfaces at the joints of robotic arms and guide grooves of conveyor belt systems, providing a reliable geometric basis for the motion planning and layout optimization of the production line, and effectively improving the operation efficiency and safety of the automated production line.

[0171] In order to break through the limitations of traditional model repair and provide an efficient and reliable solution for the integrity reconstruction of industrial digital models, this application provides an embodiment of an automated production line digital model hole filling device for implementing all or part of the content of the automated production line digital model hole filling method. See Figure 8 , the automated production line digital model hole filling device specifically includes the following content:

[0172] The boundary determination module 10 is used to decompose the automated production line digital model into a multi-level grid representation, calculate the local curvature field distribution at each grid level, project the curvature field onto the finest level grid, calculate the curvature gradient change based on the finest level grid, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0173] The filling calculation module 20 is used to perform density clustering analysis on the boundary feature descriptor, map the density clustering result to three-dimensional space to form filling patches, calculate the angle between the normal vector of the filling patches and the normal vector of the boundary points, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patches;

[0174] The surface filling module 30 is used to input the to-be-optimized patch into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function considering the surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, and performs local adjustment in the transition region using the Laplacian mesh smoothing algorithm to generate a final filling model. The filling model is imported into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection.

[0175] As can be seen from the above description, the automated production line digital model hole filling device provided by the embodiments of the present application can accurately identify the to-be-processed area through multi-level grid analysis and curvature field calculation. An innovative patch generation mechanism based on density clustering and neural network is designed, and intelligent filling optimization is realized by combining local geometric features. The system adopts non-uniform rational basis spline surface reconstruction and Laplacian mesh smoothing algorithm to ensure the continuity and smoothness of the filling area, and is closely integrated with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0176] At the hardware level, in order to break through the limitations of traditional model repair and provide an efficient and reliable solution for the integrity reconstruction of industrial digital models, the present application provides an embodiment of an electronic device for implementing all or part of the content in the automated production line digital model hole filling method. The electronic device specifically includes the following:

[0177] A processor, a memory, a communication interface, and a bus; wherein, the processor, the memory, and the communication interface complete communication with each other through the bus; the communication interface is used to realize information transmission between the automated production line digital model hole filling device and related devices such as the core business system, the user terminal, and the relevant database. The logic controller can be a desktop computer, a tablet computer, a mobile terminal, etc., and this embodiment is not limited thereto. In this embodiment, the logic controller can be implemented with reference to the embodiments of the automated production line digital model hole filling method and the embodiments of the automated production line digital model hole filling device, and the content is incorporated herein, and the repeated parts will not be described again.

[0178] It can be understood that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, smart watches, smart bracelets, etc.

[0179] In practical applications, part of the method for filling holes in the digital model of an automated production line can be executed on the side of the electronic device as described above, or all operations can be completed in the client device. Specifically, it can be selected according to the processing capacity of the client device and the limitations of the user usage scenario, etc. This application does not make any limitations in this regard. If all operations are completed in the client device, the client device may further include a processor.

[0180] The above-mentioned client device may have a communication module (i.e., a communication unit), and can communicate with a remote server to realize data transmission with the server. The server may include a server on the task scheduling center side, and in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or may include a server cluster composed of multiple servers, or a server structure of a distributed device.

[0181] Figure 9 It is a schematic block diagram of the system composition of the electronic device 9600 according to an embodiment of the present application. As Figure 9 shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It should be noted that this Figure 9 is exemplary; other types of structures can also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0182] In one embodiment, the function of the method for filling holes in the digital model of an automated production line can be integrated into the central processing unit 9100. Among them, the central processing unit 9100 may be configured to perform the following controls:

[0183] Step S101: Decompose the digital model of the automated production line into a grid representation of multiple levels, calculate the local curvature field distribution at each grid level, project the curvature field to the finest level grid, calculate the curvature gradient change based on the finest level grid, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0184] Step S102: Perform density clustering analysis on the boundary feature descriptors, map the density clustering results to a three-dimensional space to form filling patches, calculate the angle between the normal vector of the filling patches and the normal vector of the boundary points, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patches;

[0185] Step S103: Input the patches to be optimized into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function considering surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, and performs local adjustment in the transition region using the Laplacian mesh smoothing algorithm to generate a final filling model. Import the filling model into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection.

[0186] As can be seen from the above description, the electronic device provided in the embodiments of the present application accurately identifies the area to be processed through multi-level grid analysis and curvature field calculation. Innovatively designs a patch generation mechanism based on density clustering and neural networks, and combines local geometric features to achieve intelligent filling optimization. The system uses non-uniform rational basis spline surface reconstruction and Laplacian mesh smoothing algorithm to ensure the continuity and smoothness of the filling area, and is closely combined with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0187] In another embodiment, the automated production line digital model hole filling device can be separately configured from the central processing unit 9100. For example, the automated production line digital model hole filling device can be configured as a chip connected to the central processing unit 9100, and the function of the automated production line digital model hole filling method can be realized through the control of the central processing unit.

[0188] As Figure 9 shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It should be noted that the electronic device 9600 does not necessarily have to include all the components shown in Figure 9 ; in addition, the electronic device 9600 may further include components not shown in Figure 9 , and reference can be made to the prior art.

[0189] As Figure 9 shown, the central processing unit 9100, sometimes also referred to as a controller or operation control, may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives inputs and controls the operation of various components of the electronic device 9600.

[0190] Among them, the memory 9140 can be, for example, one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices. It can store the above information related to failures, and can also store programs for executing relevant information. And the central processing unit 9100 can execute the programs stored in the memory 9140 to implement information storage or processing, etc.

[0191] The input unit 9120 provides inputs to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to supply power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display can be, for example, an LCD display, but is not limited thereto.

[0192] The memory 9140 can be a solid-state memory. For example, a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It can also be a memory that stores information even when powered off, can be selectively erased and has more data. Examples of such a memory are sometimes referred to as EPROMs, etc. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 can include an application / function storage section 9142, which is used to store application programs and function programs or the processes for operating the electronic device 9600 through the central processing unit 9100.

[0193] The memory 9140 can also include a data storage section 9143, which is used to store data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage section 9144 of the memory 9140 can include various drivers of the electronic device for communication functions and / or for performing other functions of the electronic device (such as a messaging application, an address book application, etc.).

[0194] The communication module 9110 is a transmitter / receiver that transmits and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processing unit 9100 to provide input signals and receive output signals, which can be the same as in the case of a conventional mobile communication terminal.

[0195] Based on different communication technologies, in the same electronic device, multiple communication modules 9110 can be provided, such as a cellular network module, a Bluetooth module, and / or a wireless local area network module, etc. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby implementing normal telecommunication functions. The audio processor 9130 may include any suitable buffers, decoders, amplifiers, etc. Additionally, the audio processor 9130 is also coupled to a central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0196] Embodiments of the present application also provide a computer-readable storage medium capable of implementing all steps in the automated production line digital model hole filling method where the execution subject in the above embodiments is a server or a client. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all steps in the automated production line digital model hole filling method where the execution subject in the above embodiments is a server or a client are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0197] Step S101: Decompose the automated production line digital model into a grid representation of multiple levels. Calculate the local curvature field distribution at each grid level, project the curvature field onto the finest level grid, calculate the curvature gradient change based on the finest level grid, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0198] Step S102: Perform density clustering analysis on the boundary feature descriptor, map the density clustering result to a three-dimensional space to form a filling patch, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary points, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches smaller than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate a motion control instruction for the automated production line based on the filling patch;

[0199] Step S103: Input the patch to be optimized into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function considering the surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, applies the Laplacian mesh smoothing algorithm for local adjustment in the transition region to generate a final filling model, and imports the filling model into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection.

[0200] As can be seen from the above description, the computer-readable storage medium provided by the embodiments of the present application accurately identifies the area to be processed through multi-level grid analysis and curvature field calculation. Innovatively designs a patch generation mechanism based on density clustering and neural networks, and combines local geometric features to achieve intelligent filling optimization. The system uses non-uniform rational basis spline surface reconstruction and Laplacian mesh smoothing algorithm to ensure the continuity and smoothness of the filling area, and is closely integrated with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0201] Embodiments of the present application also provide a computer program product capable of implementing all steps in the automated production line digital model hole filling method with the execution subject being a server or a client in the above embodiments. When the computer program / instructions are executed by a processor, the steps of the automated production line digital model hole filling method are implemented. For example, the computer program / instructions implement the following steps:

[0202] Step S101: Decompose the automated production line digital model into a multi-level grid representation, calculate the local curvature field distribution at each grid level, project the curvature field to the finest-level grid, calculate the curvature gradient change based on the finest-level grid, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor;

[0203] Step S102: Perform density clustering analysis on the boundary feature descriptor, map the density clustering result to three-dimensional space to form filling patches, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary points, construct a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patches with an angle less than the optimal angle threshold as patches to be optimized, calculate the boundary curvature information and topological connection relationship of the patches to be optimized, and generate motion control instructions for the automated production line based on the filling patches.

[0204] Step S103: Input the patch to be optimized into the surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function considering the surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition region between the initial filling surface and the boundary of the original model, applies the Laplace mesh smoothing algorithm for local adjustment in the transition region to generate a final filling model, and imports the filling model into the three-dimensional visualization system of the automated production line for material handling path planning and equipment collision detection.

[0205] As can be seen from the above description, the computer program product provided by the embodiments of the present application accurately identifies the area to be processed through multi-level grid analysis and curvature field calculation. Innovatively designs a patch generation mechanism based on density clustering and neural networks, and combines local geometric features to achieve intelligent filling optimization. The system uses non-uniform rational basis spline surface reconstruction and Laplace mesh smoothing algorithm to ensure the continuity and smoothness of the filling area, and is closely integrated with the motion control and path planning of the automated production line. This method breaks through the limitations of traditional model repair and provides an efficient and reliable solution for the integrity reconstruction of industrial digital models.

[0206] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, apparatus, or computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0207] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0208] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one process Figure 1 or more processes and / or blocks Figure 1 specified in one block or more blocks.

[0209] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process Figure 1 or more processes and / or blocks Figure 1 specified in one block or more blocks.

[0210] Specific embodiments are used in the present invention to illustrate the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for filling holes in a digital model of an automated production line, characterized in that: The method comprises: Decomposing the digital model of the automated production line into multiple levels of grid representation, calculating the local curvature field distribution at each grid level, projecting the curvature field to the finest level grid, calculating the curvature gradient change based on the finest level grid, using an adaptive threshold segmentation algorithm to mark the area to be processed, extracting boundary points from the area to be processed, calculating the normal vector field of the boundary points, and combining the position information of the boundary points with the normal vector field to construct a boundary feature descriptor; Perform density clustering analysis on the boundary feature descriptors, map the density clustering results to three-dimensional space to form a filling patch, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary point, build a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patch less than the optimal angle threshold as a patch to be optimized, calculate the boundary curvature information and topological connection relationship of the patch to be optimized, and generate motion control instructions for the automated production line based on the filling patch; The patch to be optimized is input into a surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, constructs an objective function that considers surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates a transition area between the initial filling surface and the boundary of the original model, applies a Laplace mesh smoothing algorithm to perform local adjustments in the transition area, generates a final filling model, and imports the filling model into a three-dimensional visualization system of an automated production line for material handling path planning and equipment collision detection.

2. The method for filling holes in a digital model of an automated production line according to claim 1, characterized in that: The method comprises: decomposing the digital model of the automated production line into multiple levels of grid representation, calculating the local curvature field distribution at each grid level, projecting the curvature field to the finest level grid, calculating the curvature gradient change based on the finest level grid, and using an adaptive threshold segmentation algorithm to mark the area to be processed, including: The digital model of the automated production line is divided into grid levels of different scales using an octree decomposition algorithm, a vertex adjacency graph is constructed for each grid level, the Gaussian curvature and the average curvature of the grid vertices are calculated based on the adjacency graph, and the Gaussian curvature and the average curvature are combined to form a local curvature field distribution; An interpolation mapping algorithm between grid levels is used to project the curvature fields of different levels onto the finest level grid, and a curvature gradient tensor field is constructed. The main direction and eigenvalue of the curvature gradient tensor field are calculated, and an adaptive threshold segmentation function is constructed based on the eigenvalue. The output of the segmentation function is marked as the area to be processed.

3. The method for filling holes in a digital model of an automated production line according to claim 1, characterized in that: The step of extracting boundary points from the area to be processed, calculating the normal vector field of the boundary points, and combining the position information of the boundary points with the normal vector field to construct a boundary feature descriptor includes: Construct a boundary tracking algorithm for the area to be processed, search for the connected boundary of the area to be processed based on the mesh topological connection relationship, extract the mesh vertices on the connected boundary as a boundary point set, and use the local surface fitting method to calculate the normal vector field of the boundary point set; The spatial coordinates of the boundary point set are transformed, a local coordinate system centered on the boundary point is established, the three-dimensional coordinate information of the boundary point and the normal vector field are orthogonally decomposed in the local coordinate system, and a boundary feature descriptor containing position coordinate components and normal vector components is constructed.

4. The method for filling holes in a digital model of an automated production line according to claim 1, characterized in that: The method of performing density clustering analysis on the boundary feature descriptor, mapping the density clustering result to a three-dimensional space to form a filling patch, calculating the angle between the normal vector of the filling patch and the normal vector of the boundary point, and constructing a neural network prediction model based on local geometric features includes: Constructing a similarity measurement function of the boundary feature descriptors, calculating the distance matrix between the feature descriptors based on the similarity measurement function, dividing the boundary feature descriptors into multiple clusters using a density-based spatial clustering algorithm, calculating the core points and boundary points for each cluster, and projecting the spatial coordinates of the core points and the boundary points back into the three-dimensional space to form a filling patch; The patch normal vector is calculated based on the mesh vertices of the padded patch, and an angle evaluation model between the padded patch normal vector and the boundary point normal vector is established. The angle evaluation model is used as the input feature of the neural network, a multi-layer perceptron network structure is constructed, and the local geometric features of the boundary points are used as training samples to optimize the neural network.

5. The method for filling holes in a digital model of an automated production line according to claim 1, characterized in that: The method of using the neural network prediction model to calculate the optimal angle threshold, marking the facets smaller than the optimal angle threshold as the facets to be optimized, calculating the boundary curvature information and the topological connection relationship of the facets to be optimized, and generating the motion control instructions of the automated production line based on the filled facets, includes: Applying the trained neural network model to the local geometric features of the filled patch, outputting the optimal angle threshold between the patch normal vector and the boundary point normal vector, screening the filled patch based on the optimal angle threshold, constructing an index set of the patch to be optimized, calculating the principal curvature direction and curvature value of the boundary of the patch to be optimized, and establishing an adjacency relationship graph between the patches; The boundary curvature information of the patch to be optimized is converted into a discrete sampling point sequence, a parameterized representation of the patch boundary is established based on the sampling point sequence, and transition constraints between patches are constructed using topological connection relationships. The transition constraints are used as constraints for control instructions to generate a motion control instruction sequence that meets continuity requirements.

6. The method for filling holes in a digital model of an automated production line according to claim 1, characterized in that: The step of inputting the patch to be optimized into a surface reconstruction module, wherein the surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, applies continuity constraints to the basis spline surface using the topological connection relationship, and constructs an objective function that considers surface smoothness, including: Convert the boundary curvature information of the patch to be optimized into a control point sequence, construct the node vector of the NURBS surface based on the control point sequence, calculate the value of the NURBS basis function on the node vector, perform weight optimization on the basis function to obtain the mathematical expression of the non-uniform rational basis spline surface, and establish a mapping relationship between the surface parameter domain and the three-dimensional space; According to the topological connection relationship of the facets to be optimized, the position continuity and tangential continuity constraints between adjacent facets are constructed, and the continuity constraints are combined with the mean curvature and Gaussian curvature of the surface to construct a smoothness evaluation function, which is used as the optimization objective function.

7. The method for filling holes in a digital model of an automated production line according to claim 1, characterized in that: The objective function is iteratively optimized to obtain an initial filling surface, a transition area between the initial filling surface and the boundary of the original model is calculated, a Laplace mesh smoothing algorithm is applied to the transition area for local adjustment to generate a final filling model, and the filling model is imported into a three-dimensional visualization system of an automated production line for material handling path planning and equipment collision detection, including: The objective function is numerically optimized using the gradient descent method. The control point positions and weight parameters of the NURBS surface are updated through iterative calculation until the objective function converges to obtain an initial filling surface that satisfies the continuity constraint. The distance field function between the initial filling surface and the original model boundary is constructed, and the transition area range is determined based on the contour lines of the distance field function. The Laplacian operator of the mesh vertices is constructed in the transition area, the displacement vectors of the mesh vertices are calculated based on the Laplacian operator, the spatial positions of the mesh vertices are iteratively updated to achieve local smoothing, the smoothed mesh is merged with the original model to form a complete filling model, and the spatial index structure of the filling model is established for path planning and collision detection.

8. A hole filling device for a digital model of an automated production line, characterized in that: The device comprises: A boundary determination module is used to decompose the digital model of the automated production line into multiple levels of grid representation, calculate the local curvature field distribution at each grid level, project the curvature field to the finest level grid, calculate the curvature gradient change based on the finest level grid, use an adaptive threshold segmentation algorithm to mark the area to be processed, extract boundary points from the area to be processed, calculate the normal vector field of the boundary points, and combine the position information of the boundary points with the normal vector field to construct a boundary feature descriptor; A filling calculation module is used to perform density clustering analysis on the boundary feature descriptors, map the density clustering results to three-dimensional space to form a filling patch, calculate the angle between the normal vector of the filling patch and the normal vector of the boundary point, build a neural network prediction model based on local geometric features, use the neural network prediction model to calculate the optimal angle threshold, mark the patch less than the optimal angle threshold as a patch to be optimized, calculate the boundary curvature information and topological connection relationship of the patch to be optimized, and generate motion control instructions for the automated production line based on the filling patch; A surface filling module is used to input the surface patch to be optimized into a surface reconstruction module. The surface reconstruction module first constructs a non-uniform rational basis spline surface based on the boundary curvature information, uses the topological connection relationship to impose continuity constraints on the basis spline surface, constructs an objective function that considers the surface smoothness, iteratively optimizes the objective function to obtain an initial filling surface, calculates the transition area between the initial filling surface and the original model boundary, applies the Laplace mesh smoothing algorithm to perform local adjustments in the transition area, generates a final filling model, and imports the filling model into a three-dimensional visualization system of an automated production line to perform material handling path planning and equipment collision detection.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for filling holes in the digital model of an automated production line described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for filling holes in a digital model of an automated production line as described in any one of claims 1 to 7 are implemented.

Images