Bridge large segment component digital assembly method fusing point cloud and mechanical simulation

Abstract

The application discloses a kind of fusion point cloud and bridge large section component digital assembly method and system of mechanical simulation, and the method comprises: obtaining the measured three-dimensional data of large component to be assembled, and simultaneously establishing mechanical analysis model.Extract the multi-dimensional key assembly features of large component in three-dimensional data in parallel, and the node position set, element boundary information and mechanical parameters of mechanical analysis model are simultaneously established, and the deep semantic association of both is simultaneously established.The component three-dimensional data model and the bound mechanical analysis model are virtually assembled to the specified position using a two-stage strategy.The node mechanical parameters are obtained by simulating calculation according to actual stress condition, the node displacement vector is superimposed to the key assembly feature based on the global correlation, the coordinate is updated to obtain the assembly feature position reflecting the real stress state, and then digital assembly is carried out again according to the updated assembly position.The method solves the assembly deviation problem caused by the difference between the scanning state and the assembly state of the component.

Description

Technical Field

[0001] This invention belongs to the field of digital assembly technology, and more specifically, relates to a digital assembly method and system for large-segment bridge components that integrates point cloud and mechanical simulation. Background Technology

[0002] Modern long-span bridges are developing towards "larger scale, prefabrication, and precision." Large-segment prefabrication and assembly technology is widely used due to its ability to improve construction efficiency and reduce the risks of working at heights. Before components leave the factory, they need to be pre-assembled to check the accuracy of the interfaces. Traditional physical pre-assembly occupies a large area, consumes a lot of manpower and resources, and is difficult to reproduce the design accuracy due to the influence of gravity deformation. Digital assembly based on 3D laser scanning and digital twin technology has become a new trend in the industry.

[0003] Current digital assembly technologies mostly acquire point cloud data using high-precision 3D laser scanners, and then use algorithms to virtually align the point cloud models of adjacent components to analyze manufacturing and assembly deviations. However, for large-segment bridge components, the state of the components during 3D scanning is often inconsistent with that during actual assembly (e.g., lying flat on the ground during scanning, but suspended in the air during assembly). Relying solely on geometric assembly cannot fully reflect the mechanical deviations in the assembly of large-segment components. Furthermore, large-segment components undergo non-rigid deformation under their own weight or temperature. Although existing digital assembly technologies incorporate finite element analysis, they mostly employ strategies of "sparse point correction" and "projection calculation deviation," considering only the characteristics of individual points and ignoring the overall continuous deformation and actual 3D deviations of the component, thus failing to accurately reflect digital assembly deviations under stress. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a digital assembly method that is compatible with various heterogeneous components such as welding and concrete connections. By establishing deep semantic binding between multi-dimensional features of points, lines, and surfaces and finite element nodes, it achieves precise correction of global deformation. An iterative optimization mechanism is introduced to simulate the real assembly process, ultimately obtaining the optimal assembly scheme that balances geometric accuracy and mechanical safety.

[0005] To address the aforementioned deficiencies or improvement needs of existing technologies, as a first aspect of this invention, the present invention provides a digital assembly method for large-segment bridge components that integrates point cloud and mechanical simulation, comprising:

[0006] S1. Obtain the measured three-dimensional data of the large component to be assembled and complete the preprocessing; at the same time, obtain the corresponding theoretical mechanical analysis model of the component;

[0007] S2. Perform two operations in parallel: identify and extract multi-dimensional key assembly features from the preprocessed 3D data, and simultaneously obtain the corresponding node location set, unit boundary information and mechanical parameters from the mechanical analysis model;

[0008] S3. Using spatial geometric mapping method and hierarchical binding strategy, establish deep semantic association between multi-dimensional key assembly features and mechanical analysis model nodes in three-dimensional data, and complete the global association between geometric features and mechanical nodes;

[0009] S4. A two-stage registration strategy of "coarse registration + fine registration" is adopted to virtually assemble the three-dimensional data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model;

[0010] S5. Based on the actual stress conditions of the component at the virtual positioning position, perform mechanical simulation calculations on the mechanical analysis model to obtain the mechanical parameters of each node in the model;

[0011] S6. Based on the global correlation between geometric features and mechanical nodes, the node displacement vectors obtained from mechanical simulation are superimposed onto the key assembly features of the three-dimensional data, and the feature coordinates are updated to obtain the assembly results that reflect the actual stress state.

[0012] Furthermore, the types of large components in S1 include bridge components such as bolted connections, welded or concrete structures including trusses, box girders, and arch ribs, as well as mixed irregular segments composed of the above structures;

[0013] The mechanical analysis model adopts the finite element analysis model, which is constructed by using beam elements, shell elements or solid elements, and its mesh density meets the mapping accuracy requirements of the assembly features.

[0014] Furthermore, the multi-dimensional key assembly features in S2 include:

[0015] Zero-dimensional point features include bolt hole center points, component corner points, measurement reference points, and temporary hoisting points;

[0016] One-dimensional line features include plate element edge lines, arch rib center axis, welds, box girder outline, and component edge lines;

[0017] Two-dimensional surface features include splicing contact surfaces, web planes, top and bottom plate planes, and curved skin areas.

[0018] Furthermore, the process of completing the global association between geometric features and mechanical nodes in S3 includes a hierarchical binding strategy for features of different dimensions:

[0019] For zero-dimensional point features: For discrete key points in the point cloud, K-nearest neighbor search is performed on the node set of the finite element model; if the distance between the feature point and the nearest node is less than a preset threshold, rigid node constraints are established; if the feature point is located inside the element, the interpolation weight is calculated using the shape function of the element to establish a weighted coupling relationship between the feature point and the element node.

[0020] For one-dimensional line features: the edge lines in the point cloud are mapped to the mesh surface of the finite element model using a curve projection algorithm; the element boundaries or node sequences crossed by the projection path are identified, and path constraint associations are established; for features at non-node locations on the path, their degrees of freedom are bound to two adjacent mesh nodes through linear interpolation.

[0021] For two-dimensional surface features: For continuous plate surfaces or curved surfaces in point clouds, bounding box overlap detection is used to filter candidate unit sets; further, the cosine similarity between the point cloud normal vector and the finite element shell element normal vector is calculated to eliminate nodes with inconsistent normals; the filtered finite element node set is defined as a subordinate surface region, and a region mapping relationship is established with the point cloud surface features.

[0022] Furthermore, the specific implementation process of the two-stage registration strategy in S4 is as follows:

[0023] In the coarse registration stage, let the set of homologous feature points selected in the 3D data model of the component to be assembled be... ,in , Let be the three-dimensional coordinates of the feature points in the local coordinate system of the component, and let be the set of homologous feature points corresponding to the design positions of the overall structural model. ,in These are the three-dimensional coordinates of the feature points in the global coordinate system of the overall structure.

[0024] Compute point set center of mass Calculate the point set center of mass Construct a decentralized set of nodes and Establish the covariance matrix ,right Singular value decomposition yields ,in and It is an orthogonal matrix. Given a diagonal matrix; then solve for the rotation matrix. With translation vector ,pass The initial adjustment of the component's attitude was completed, and coarse registration was achieved;

[0025] In the fine registration stage, the component posture obtained after coarse registration is taken as the initial state, and the point cloud set corresponding to the multi-dimensional key assembly features in the component's 3D data is set as follows: ,in The real-time coordinates of the feature point cloud in the global coordinate system; the point cloud set corresponding to the position of the overall structural model design is as follows: ,in To design the coordinates of the feature point cloud in the global coordinate system;

[0026] Define the iterative objective function as: ,in Represents Euclidean distance. For the first The coordinates of the feature point cloud after the next iteration and For the first The rotation matrix and translation vector for each iteration;

[0027] During the iteration process, based on the definition of points, lines, and surfaces according to the multi-dimensional key assembly feature attributes, for each exist Matching corresponding homologous feature points By eliminating outlier matching pairs corresponding to non-homologous features, and minimizing... renew and Repeat the iteration until... Less than the preset accuracy threshold ,in Determined based on the accuracy of 3D data preprocessing and the tolerance of assembly design;

[0028] Finally passed Achieve high-precision alignment between the 3D data model of the component and its bound mechanical analysis model and the overall structural model at the specified location in the design. and These are the rotation matrix and translation vector obtained from the final iteration.

[0029] Furthermore, the actual stress conditions in S5 include: the gravity of the component itself, the thermal expansion and contraction effect caused by changes in ambient temperature, the hoisting condition, and the forced displacement boundary conditions generated when forced assembly is performed to eliminate manufacturing errors.

[0030] Furthermore, S6 also includes: calculating the deviation of the assembly features of adjacent components and generating an assembly adjustment scheme adapted to the engineering requirements.

[0031] If the characteristic geometric deviation or nodal stress level exceeds the preset threshold, the virtual pose parameters of the component are fine-tuned with the goal of minimizing the deviation and stress value. The mechanical simulation and deformation calculation and related operations of this step are executed repeatedly until the preset tolerance requirements are met. Finally, the deformed three-dimensional assembly model, mechanical parameter analysis results and digital assembly accuracy report are output.

[0032] Furthermore, the specific process of fine-tuning the virtual pose parameters of the component is as follows:

[0033] With the objective of minimizing the deviation of the assembly features of adjacent components and the nodal stress values, let the set of assembly feature points of adjacent components after deformation correction be as follows: and ,in The three-dimensional coordinates of the assembly feature points of the first component are given. Given the three-dimensional coordinates of the assembly feature points corresponding to the second component, calculate the deviation of a single feature point. This leads to the overall feature geometric deviation evaluation index. ;

[0034] Let the stress values ​​at each node in the mechanical analysis model be... ,in, Define a nodal stress level evaluation index for the total number of nodes. Construct the objective function ,in This is the normalization coefficient for stress level and geometric deviation, used to bring the two evaluation indicators to the same order of magnitude;

[0035] Set the virtual pose adjustment parameters of the component to three-dimensional rotation angle. and three-dimensional translation ,in Respectively around the global coordinate system , , The rotation angle of the shaft, respectively along the global coordinate system , , The translation distance of the axis;

[0036] Based on the objective function Solving using the gradient descent method makes Minimize the pose adjustment parameters and calculate the partial derivatives of the objective function with respect to each adjustment parameter. The parameter adjustment trend is determined based on the direction of the partial derivatives, and the parameters are updated gradually. and The possible values ​​of ;

[0037] Each time the pose adjustment parameters are updated, the virtual pose of the component's 3D data model and its bound mechanical analysis model are updated synchronously. Mechanical simulation calculations are then re-executed to obtain new nodal displacement vectors and stress values. Based on the global correlation between geometric features and mechanical nodes, the assembly feature coordinates are corrected, and the calculations are repeated. and And update the objective function. ;

[0038] Repeat the above process of parameter updating, pose adjustment, simulation calculation, and deviation and stress evaluation until... Less than the preset geometric tolerance threshold and The stress is less than the preset safety threshold, at which point the obtained and This is the optimal virtual pose adjustment parameter.

[0039] As a second aspect of the present invention, a digital assembly system for large-segment bridge components integrating point cloud and mechanical simulation is also provided, comprising:

[0040] The data input and preprocessing unit is used to acquire the measured three-dimensional data of the large component to be assembled and complete the preprocessing; at the same time, it acquires the theoretical mechanical analysis model corresponding to the component.

[0041] The multidimensional feature and mechanical analysis node extraction unit is used to perform two operations in parallel: identify and extract multidimensional key assembly features from the preprocessed three-dimensional data, and obtain the corresponding node position set, unit boundary information and mechanical parameters from the mechanical analysis model.

[0042] The multidimensional feature and mechanical node semantic binding unit is used to establish a deep semantic association between multidimensional key assembly features and mechanical analysis model nodes in three-dimensional data by adopting spatial geometric mapping method and hierarchical binding strategy, and to complete the global association between geometric features and mechanical nodes.

[0043] The component virtual positioning and attitude calibration unit is used to virtually assemble the three-dimensional data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model using a two-stage registration strategy of "coarse registration + fine registration".

[0044] The multi-condition mechanical simulation and deformation calculation unit is used to perform mechanical simulation calculations on the mechanical analysis model for the actual stress conditions of the component at the virtual positioning position, and to obtain the mechanical parameters of each node in the model.

[0045] The deformation correction unit is used to superimpose the nodal displacement vectors obtained from mechanical simulation onto the key assembly features of the three-dimensional data based on the global correlation between geometric features and mechanical nodes, and update the feature coordinates to obtain the assembly results that reflect the actual stress state.

[0046] As a third aspect of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored, which is executed by a processor as described in any one of the present invention, a digital assembly method for large-segment bridge components that integrates point cloud and mechanical simulation.

[0047] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0048] 1. The present invention provides a digital assembly method for large-segment bridge components that integrates point cloud and mechanical simulation. This method acquires and preprocesses measured 3D data of the components to be assembled, simultaneously establishing or importing corresponding theoretical mechanical analysis models to lay the data and model foundation for subsequent assembly analysis. It extracts multi-dimensional key assembly features from the 3D data and node location sets, element boundary information, and mechanical parameters from the mechanical analysis model in parallel, clarifying the geometric matching benchmark and mechanical calculation carrier. Employing a spatial geometric mapping method and a hierarchical binding strategy, it establishes a deep semantic association between multi-dimensional assembly features and mechanical nodes, achieving a global association between geometric features and mechanical nodes. This breaks the limitation of separating geometry and mechanics in traditional methods, allowing mechanical deformation data to accurately correspond to geometric assembly features, providing core logical support for subsequent deformation correction, and ensuring that the assembly results truly reflect the stress state of the components.

[0049] 2. The digital assembly method for large bridge segments integrating point cloud and mechanical simulation of this invention uses a two-stage registration strategy of "coarse registration + fine registration" to accurately and virtually assemble the 3D data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model. For the actual stress conditions of the component at this virtual positioning position, mechanical simulation calculations are performed on the mechanical analysis model to comprehensively capture the mechanical parameters of each node under conditions such as self-weight and temperature changes. Based on the global correlation between geometric features and mechanical nodes, the simulated node displacement vectors are superimposed onto the key assembly features of the 3D data to update the feature coordinates. This effectively solves the problem of "points matching but surfaces not fitting" caused by traditional sparse point correction, significantly improving the geometric accuracy and realism of digital assembly.

[0050] 3. The digital assembly method for large bridge segments integrating point cloud and mechanical simulation of this invention automatically generates an assembly adjustment scheme adapted to engineering requirements by calculating the deviation of assembly features of adjacent components. When the feature geometric deviation or node stress level exceeds a preset threshold, the virtual pose parameters of the components are fine-tuned with the goal of minimizing the deviation and stress value, and mechanical simulation, deformation calculation, and deviation and stress evaluation are performed iteratively. This iterative optimization mechanism can simulate the dynamic adjustment process of real forced assembly, avoiding the disconnect between theory and construction caused by the traditional one-time assembly process. The final output of the deformed and corrected three-dimensional assembly model and digital assembly accuracy report provides a feasible solution for on-site construction that balances geometric accuracy and mechanical safety. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of a digital assembly method for large bridge segment components that integrates point cloud and mechanical simulation according to an embodiment of the present invention;

[0052] Figure 2 This is a technical flowchart of an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of the original point cloud data of the steel box girder in an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of the outline features of a steel box girder according to an embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram of the discrete point features of the box girder to be assembled according to an embodiment of the present invention;

[0056] Figure 6 This is a system unit diagram of an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0058] Example 1

[0059] Please refer to Figure 1 This embodiment 1 provides a digital assembly method for large-segment bridge components that integrates point cloud and mechanical simulation, including:

[0060] S1. Obtain the measured three-dimensional data of the large component to be assembled and complete the preprocessing; at the same time, obtain the corresponding theoretical mechanical analysis model of the component;

[0061] S2. Perform two operations in parallel: identify and extract multi-dimensional key assembly features from the preprocessed 3D data, and simultaneously obtain the corresponding node location set, unit boundary information and mechanical parameters from the mechanical analysis model;

[0062] S3. Using spatial geometric mapping method and hierarchical binding strategy, establish deep semantic association between multi-dimensional key assembly features and mechanical analysis model nodes in three-dimensional data, and complete the global association between geometric features and mechanical nodes;

[0063] S4. A two-stage registration strategy of "coarse registration + fine registration" is adopted to virtually assemble the three-dimensional data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model;

[0064] S5. Based on the actual stress conditions of the component at the virtual positioning position, perform mechanical simulation calculations on the mechanical analysis model to obtain the mechanical parameters of each node in the model;

[0065] S6. Based on the global correlation between geometric features and mechanical nodes, the node displacement vectors obtained from mechanical simulation are superimposed onto the key assembly features of the three-dimensional data, and the feature coordinates are updated to obtain the assembly results that reflect the actual stress state.

[0066] Please refer to Figure 2 This embodiment 1 further elaborates on the above steps.

[0067] (1) Data input and preprocessing

[0068] Against the backdrop of bridge engineering developing towards larger scale, prefabrication, and precision, the digital pre-assembly of large segmental components requires accurate data and reliable models as a foundation. Data acquisition and model preparation are key links connecting physical components and digital analysis, directly affecting the accuracy of subsequent assembly analysis.

[0069] The acquisition of measured 3D data needs to be adapted to different types of large bridge segments, including bolted, welded, or concrete trusses, box girders, arch ribs, and mixed irregular segments composed of the above structures.

[0070] Taking a steel box girder segment as an example, a Faro Focus S150 3D laser scanner was used for omnidirectional scanning with a resolution of 1 / 4. This parameter corresponds to a scan point spacing of approximately 6mm at a distance of 10 meters, which can fully capture the detailed features of the component surface. To achieve accurate stitching of multi-station scan data, no fewer than six target spheres were placed around and inside the girder segment as stitching references. After scanning, the raw data was imported into SCENE software. First, noise reduction was performed to remove invalid points caused by environmental interference and equipment errors, and then multi-station data stitching was performed. Figure 3 As shown, a complete 3D point cloud model is finally formed, ensuring that the average point spacing of the model is better than 2mm, which meets the accuracy requirements of subsequent assembly feature extraction. At the same time, the ambient temperature during scanning is recorded to provide actual working condition data support for subsequent mechanical simulation.

[0071] The theoretical mechanics analysis model adopts the finite element analysis model, which can be constructed by selecting beam elements, shell elements or solid elements according to the structural characteristics of the components, and the mesh density must meet the mapping accuracy requirements of the assembly features.

[0072] For steel box girder segments, a plate-shell finite element model was established using Midas Civil software. The model rigorously reproduces the actual structure of the component, including key components such as the top plate, bottom plate, web, and stiffeners, ensuring structural consistency between the model and the solid component. Simultaneously, the element mesh size was controlled within 50mm; reasonable mesh refinement improved the model's simulation accuracy of the component's stress and deformation, providing a reliable basis for subsequent mechanical simulation calculations. For other types of large-segment bridge components, pre-constructed finite element analysis models can be directly imported based on their structural form, ensuring that the model matches the actual structure of the component and the analysis requirements.

[0073] (2) Extraction of multidimensional features and mechanical analysis nodes

[0074] In the process of digital assembly, accurately extracting key features and obtaining core information for mechanical analysis are prerequisites for establishing geometric and mechanical connections. This needs to be done in parallel to ensure the matching and integrity of the two types of data, laying the foundation for subsequent full-domain connections.

[0075] The multidimensional key assembly features encompass zero-dimensional point features, one-dimensional line features, and two-dimensional surface features. Zero-dimensional point features include bolt hole center points, component corner points, measurement reference points, and temporary hoisting points. One-dimensional line features include plate unit edge lines, arch rib central axis, welds, box girder outlines, and component edge lines. Two-dimensional surface features include splicing contact surfaces, web planes, top and bottom plate planes, and curved surface skin areas.

[0076] Please refer to Figure 4 and Figure 5 Taking a steel box girder segment as an example, the pre-processed clean point cloud data is first imported into Cloudcompare point cloud processing software. With the help of the software's visual interface, points are manually drawn along the top edge, bottom edge, and outer edge of the web of the section to be spliced ​​through human-computer interaction, so as to accurately extract the key one-dimensional line feature of the section outline.

[0077] To adapt to the subsequent assembly calculation requirements, the manually drawn continuous contour lines are automatically resampled at 10cm intervals to generate a discrete contour point set. This point set can serve as a supplement to the zero-dimensional point features, together forming the key feature data for subsequent assembly. For other types of large-segment bridge components, corresponding point, line, and surface multi-dimensional features can be extracted according to their structural characteristics to ensure that the features accurately reflect the assembly benchmark of the components.

[0078] While extracting multidimensional key assembly features, the corresponding node location sets, element boundary information, and mechanical parameters are simultaneously obtained from the mechanical analysis model. The element type of the mechanical analysis model is adapted to the component structure. By extracting the node locations, element boundaries, and related mechanical parameters corresponding to the multidimensional assembly features in the model, a correspondence between the mechanical analysis data and the geometric feature data is ensured. This provides data support for establishing a deep semantic relationship between the two through spatial geometric mapping methods, achieving accurate matching between geometric features and mechanical nodes.

[0079] (3) Semantic binding of multidimensional features and mechanical nodes

[0080] The global association between geometric features and mechanical nodes is the core link to achieve deep integration of geometric assembly and mechanical simulation. Through spatial geometric mapping methods and hierarchical binding strategies, precise semantic associations can be established between assembly features of different dimensions and nodes of mechanical analysis models, ensuring that subsequent mechanical deformation data can be accurately fed back to the geometric assembly process.

[0081] First, the mechanical analysis model and the 3D data model need to be imported into the same coordinate system to eliminate the correlation deviation caused by coordinate differences, thus providing a foundation for accurate matching of features and nodes. For different dimensional attributes of multi-dimensional key assembly features, corresponding hierarchical binding strategies are adopted: For zero-dimensional point features, for discrete key points such as bolt hole center points and component corner points in the point cloud, nearest neighbor search is performed on the node set of the mechanical analysis model. If the distance between the feature point and the nearest node is less than a preset threshold, rigid node constraints are directly established to make the feature point move synchronously with the node; if the feature point is located inside the element, the shape function of the element is used to determine the interpolation relationship, establishing a weighted coupling relationship between the feature point and each node of the element, ensuring that the deformation of the feature point reflects the overall deformation state of the element.

[0082] For one-dimensional line features, such as plate element edge lines and box girder outlines, a curve projection algorithm is used to map them onto the mesh surface of the mechanical analysis model. This accurately identifies the element boundaries or node sequences traversed by the projection path and establishes path constraint associations between line features and node sequences. If there are feature points at non-node locations on the line feature, the degree of freedom of the feature point is bound to two adjacent mesh nodes using a linear interpolation method, ensuring that the continuous deformation of the line feature can be accurately reflected by node displacements.

[0083] For two-dimensional surface features, including continuous plate surfaces or curved surfaces such as spliced ​​contact surfaces and curved skin areas, a bounding box overlap detection method is first used to filter out the candidate element set corresponding to the surface feature. Then, by calculating the cosine similarity between the normal vector of the point cloud surface feature and the normal vector of the finite element shell element, nodes with inconsistent normals are eliminated to avoid invalid associations. The filtered finite element node set is defined as a subordinate surface region, and a region mapping relationship is established with the point cloud surface feature to achieve precise coupling between the global deformation of the surface feature and the displacement of the nodes. Taking the contour feature of a steel box girder segment as an example, through the above binding logic, the extracted point cloud contour point set is projected onto the boundary of the end element of the finite element model. For each contour point, the nearest node or element boundary in the finite element model is searched and an index relationship is established. For example, the top left corner point of the top plate in the point cloud is bound to the corresponding top left corner node in the finite element model to form a follow-up relationship, ensuring that the deformation of the contour feature can be synchronized with the displacement of the nodes.

[0084] (4) Virtual positioning and attitude calibration of components

[0085] The core objective of initial digital assembly is to precisely and virtually place the 3D data model of the components and its associated mechanical analysis model into the designated positions of the overall structural model through a two-stage strategy of "coarse registration + fine registration," providing an initial orientation that closely matches actual working conditions for subsequent mechanical simulations. This process must balance registration efficiency and accuracy, achieving precise geometric alignment through step-by-step optimization.

[0086] The coarse registration stage aims to quickly reduce the spatial deviation between the component and its design position. First, at least three pairs of common feature points are selected at the design positions in both the 3D data model of the component to be assembled and the overall structural model. These feature points must have significant geometric recognizability, such as the two corner points of the top plate of a steel box girder and the center point of the bottom plate. By calculating the centroids of each of the two sets of feature points, a decentralized point set is constructed to eliminate the influence of translation. Then, a covariance matrix is ​​established based on the two decentralized point sets. After singular value decomposition of the covariance matrix, the rotation matrix and translation vector are solved. Using this rotation matrix and translation vector, the orientation of the component is adjusted, completing the initial alignment between the component to be assembled and its design position. This provides an initial state close to the optimal solution for fine registration, significantly reducing the computational complexity of subsequent fine adjustments. The specific quantification process is as follows:

[0087] In the coarse registration stage, let the set of homologous feature points selected in the 3D data model of the component to be assembled be... ,in , Let be the three-dimensional coordinates of the feature points in the local coordinate system of the component, and let be the set of homologous feature points corresponding to the design positions of the overall structural model. ,in These are the three-dimensional coordinates of the feature points in the global coordinate system of the overall structure.

[0088] Compute point set center of mass Calculate the point set center of mass Construct a decentralized set of nodes and Establish the covariance matrix ,right Singular value decomposition yields ,in and It is an orthogonal matrix. Given a diagonal matrix; then solve for the rotation matrix. With translation vector ,pass The initial adjustment of the component's attitude was completed, and coarse registration was achieved;

[0089] The fine registration stage focuses on further improving registration accuracy, building upon the component posture after coarse registration. It uses the point cloud set corresponding to multi-dimensional key assembly features in the component's 3D data as the object, matching it with the point cloud set of features corresponding to the design positions in the overall structural model. During iteration, based on the attributes of multi-dimensional key assembly features such as points, lines, and surfaces, each measured point cloud feature is matched with a corresponding design point cloud feature, eliminating abnormal matching pairs caused by non-homologous features to avoid error interference. A target function is defined to quantify the spatial distance difference between the measured and design point clouds. The rotation matrix and translation vector are continuously updated to minimize this target function, iterating repeatedly until the target function value is less than a preset accuracy threshold, which is determined based on the 3D data preprocessing accuracy and assembly design tolerances. Finally, the rotation matrix and translation vector obtained from the final iteration achieve high-precision alignment between the component's 3D data model and its bound mechanical analysis model with the designated positions in the overall structural model, completing the initial pure geometric assembly. The specific quantification process is as follows:

[0090] In the fine registration stage, taking the component posture obtained after coarse registration as the initial state, let the point cloud set corresponding to the multi-dimensional key assembly features in the component's 3D data be denoted as . ,in The real-time coordinates of the feature point cloud in the global coordinate system; the point cloud set corresponding to the position of the overall structural model design is as follows: ,in To design the coordinates of the feature point cloud in the global coordinate system;

[0091] Define the iterative objective function as: ,in Represents Euclidean distance. For the first The coordinates of the feature point cloud after the next iteration and For the first The rotation matrix and translation vector for each iteration;

[0092] During the iteration process, based on the definition of points, lines, and surfaces according to the multi-dimensional key assembly feature attributes, for each exist Matching corresponding homologous feature points By eliminating outlier matching pairs corresponding to non-homologous features, and minimizing... renew and Repeat the iteration until... Less than the preset accuracy threshold ,in Determined based on the accuracy of 3D data preprocessing and the tolerance of assembly design;

[0093] Finally passed Achieve high-precision alignment between the 3D data model of the component and its bound mechanical analysis model and the overall structural model at the specified location in the design. and These are the rotation matrix and translation vector obtained from the final iteration.

[0094] (5) Multi-condition mechanical simulation and deformation calculation

[0095] Mechanical simulation and deformation calculation are crucial links connecting geometric assembly with actual stress states. By simulating the actual stress conditions of components at virtual positioning locations, nodal mechanical parameters are obtained, providing data support for subsequent feedback of mechanical deformation to geometric assembly. This avoids the disconnect between assembly analysis based solely on ideal geometric states and actual construction. The actual stress conditions must comprehensively cover the main stress forms during component assembly, including the effects of its own gravity, thermal expansion and contraction caused by changes in ambient temperature, hoisting conditions, and forced displacement boundary conditions during forced assembly, ensuring that the simulation results closely match actual construction conditions.

[0096] Taking the segmental assembly of steel box girders as an example, the specific working parameters must first be defined according to the construction scenario. In the gravity condition, the material density and gravitational acceleration are set based on the material properties of the steel box girder, enabling the model to accurately reflect the influence of its own weight on the deformation of the component. The hoisting condition needs to be set according to the on-site construction plan. In the finite element model, the nodes corresponding to the lifting lugs are set as vertical support boundaries to simulate the lifting action of the hooks. It should be noted that the lifting lugs are only used as the location for applying mechanical boundary conditions to simulate the influence of different lifting postures on the overall deformation of the beam's assembly profile; their local deformation is not a key feature of the assembly matching. In the temperature condition, the expected ambient temperature during on-site assembly is input, and the temperature difference is calculated based on the ambient temperature recorded during the previous scan. An overall temperature load is applied in the finite element model, and the coefficient of thermal expansion of the steel is set to reflect the thermal expansion and contraction effect caused by temperature changes.

[0097] After the working condition parameters are defined, based on the virtual positioning of the component, simultaneous simulation calculations of gravity, hoisting, and temperature conditions are performed in the finite element solver. During the calculation, the focus is on obtaining the mechanical parameters of key nodes in the finite element model that are already bound to geometric features, especially the node displacement vectors under the combined action of multiple working conditions. Through this calculation, the model can realistically reflect the spatial position migration and shape distortion of the end face contour caused by the overall stiffness of the beam segment under specific temperatures and hoisting support methods, providing a reliable mechanical basis for subsequently superimposing deformation data onto geometric features to achieve precise assembly and alignment. For scenarios requiring forced assembly, corresponding forced displacement boundary conditions are applied to the model to simultaneously obtain the node mechanical parameters under that working condition.

[0098] (6) Deformation correction

[0099] Deformation correction and iterative optimization are the core steps to achieving a balance between geometric accuracy and mechanical safety. Based on the established global correlation between geometric features and mechanical nodes, the deformation data obtained from mechanical simulation is fed back to the geometric assembly model. Through iterative adjustments and optimizations, a feasible assembly scheme is finally output. This process breaks through the limitations of the traditional separation of geometry and mechanics, realistically restoring the assembly form of components under stress.

[0100] In the feature reverse correction stage, relying on the binding relationship between geometric features and mechanical nodes, the comprehensive displacement vector of the nodes obtained from mechanical simulation is superimposed onto the key assembly features of the 3D data to update the feature coordinates. Taking a steel box girder segment as an example, the superimposed point cloud contour is no longer an ideal geometric shape, but presents a real physical shape under the combined effects of gravity, hoisting posture, and temperature difference, including the spatial position migration and shape distortion of the end face contour, making the assembly analysis more consistent with the actual stress scenario on site.

[0101] After the correction is completed, the deviation of the assembly characteristics of adjacent components is calculated, such as the uniformity of weld gaps and the amount of misalignment, and an assembly adjustment plan adapted to the project requirements is generated accordingly.

[0102] If the characteristic geometric deviation or nodal stress level exceeds the preset threshold, the virtual pose parameters of the component are fine-tuned with the goal of minimizing the deviation and stress value. The mechanical simulation and deformation calculation and related operations of this step are executed repeatedly until the preset tolerance requirements are met. Finally, the deformed three-dimensional assembly model, mechanical parameter analysis results and digital assembly accuracy report are output.

[0103] Specifically, the process of fine-tuning the virtual pose parameters of the component with the goal of minimizing the deviation and stress value is as follows:

[0104] With the objective of minimizing the deviation of the assembly features of adjacent components and the nodal stress values, let the set of assembly feature points of adjacent components after deformation correction be as follows: and ,in The three-dimensional coordinates of the assembly feature points of the first component are given. Given the three-dimensional coordinates of the assembly feature points corresponding to the second component, calculate the deviation of a single feature point. This leads to the overall feature geometric deviation evaluation index. ;

[0105] Let the stress values ​​at each node in the mechanical analysis model be... ,in, Define a nodal stress level evaluation index for the total number of nodes. Construct the objective function ,in This is the normalization coefficient for stress level and geometric deviation, used to bring the two evaluation indicators to the same order of magnitude;

[0106] Set the virtual pose adjustment parameters of the component to three-dimensional rotation angle. and three-dimensional translation ,in Respectively around the global coordinate system , , The rotation angle of the shaft, respectively along the global coordinate system , , The translation distance of the axis;

[0107] Based on the objective function Solving using the gradient descent method makes Minimize the pose adjustment parameters and calculate the partial derivatives of the objective function with respect to each adjustment parameter. The parameter adjustment trend is determined based on the direction of the partial derivatives, and the parameters are updated gradually. and The possible values ​​of ;

[0108] Each time the pose adjustment parameters are updated, the virtual pose of the component's 3D data model and its bound mechanical analysis model are updated synchronously. Mechanical simulation calculations are then re-executed to obtain new nodal displacement vectors and stress values. Based on the global correlation between geometric features and mechanical nodes, the assembly feature coordinates are corrected, and the calculations are repeated. and And update the objective function. ;

[0109] Repeat the above process of parameter updating, pose adjustment, simulation calculation, and deviation and stress evaluation until... Less than the preset geometric tolerance threshold and The stress is less than the preset safety threshold, at which point the obtained and This is the optimal virtual pose adjustment parameter.

[0110] The final output includes a deformed and corrected 3D assembly model, mechanical parameter analysis results, and a digital assembly accuracy report, providing precise guidance for on-site construction.

[0111] The digital assembly method proposed in this embodiment has strong practical application prospects in the field of bridge engineering prefabrication and assembly. For various types of large-segment components such as steel box girders, steel trusses, and steel arch ribs, it can effectively solve the problems of large site occupation, high cost, and great susceptibility to environmental influences associated with traditional physical prefabrication. At the same time, it can make up for the shortcomings of existing digital assembly technologies in terms of adaptability to heterogeneous components and the realism of deformation simulation. In the factory prefabrication stage, digital prefabrication can detect assembly conflicts caused by manufacturing errors and stress deformation in advance, reducing on-site rework. In the construction stage, the output of accurate assembly schemes and mechanical parameter reports can directly guide key processes such as hoisting posture adjustment and temperature control, significantly reducing the risks of high-altitude operations and structural damage hazards caused by forced assembly. It is especially suitable for high-standard engineering requirements such as long-span bridges and complex irregular segmental bridges.

[0112] From an industry development perspective, this method aligns with the development trend of bridge engineering towards "large-scale, prefabricated, and precise" construction, providing core technical support for the application of digital twin technology in bridge construction. Its established multi-dimensional feature and mechanical node association mechanism, two-stage registration strategy, and iterative optimization logic can be further extended to similar engineering scenarios such as railway bridges and municipal elevated roads, providing a standardized technical paradigm for the assembly of large segments with different structural forms.

[0113] Example 2

[0114] Please refer to Figure 6 This embodiment 2 provides a digital assembly system for large bridge segment components that integrates point cloud and mechanical simulation, including:

[0115] The data input and preprocessing unit is used to acquire the measured three-dimensional data of the large component to be assembled and complete the preprocessing; at the same time, it acquires the theoretical mechanical analysis model corresponding to the component.

[0116] The multidimensional feature and mechanical analysis node extraction unit is used to perform two operations in parallel: identify and extract multidimensional key assembly features from the preprocessed three-dimensional data, and obtain the corresponding node position set, unit boundary information and mechanical parameters from the mechanical analysis model.

[0117] The multidimensional feature and mechanical node semantic binding unit is used to establish a deep semantic association between multidimensional key assembly features and mechanical analysis model nodes in three-dimensional data by adopting spatial geometric mapping method and hierarchical binding strategy, and to complete the global association between geometric features and mechanical nodes.

[0118] The component virtual positioning and attitude calibration unit is used to virtually assemble the three-dimensional data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model using a two-stage registration strategy of "coarse registration + fine registration".

[0119] The multi-condition mechanical simulation and deformation calculation unit is used to perform mechanical simulation calculations on the mechanical analysis model for the actual stress conditions of the component at the virtual positioning position, and to obtain the mechanical parameters of each node in the model.

[0120] The deformation correction unit is used to superimpose the nodal displacement vectors obtained from mechanical simulation onto the key assembly features of the three-dimensional data based on the global correlation between geometric features and mechanical nodes, and update the feature coordinates to obtain the assembly results that reflect the actual stress state.

[0121] Example 3

[0122] This embodiment 3 also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement any step of a digital assembly method for large-segment bridge components that integrates point cloud and mechanical simulation.

[0123] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] For a description of the computer-readable storage medium provided in this application, please refer to the above method embodiments; further details will not be repeated here.

[0125] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A digital assembly method for large-segment bridge components integrating point cloud and mechanical simulation, characterized in that, include: S1. Obtain the measured three-dimensional data of the large component to be assembled and complete the preprocessing; at the same time, obtain the corresponding theoretical mechanical analysis model of the component; S2. Perform two operations in parallel: identify and extract multi-dimensional key assembly features from the preprocessed 3D data, and simultaneously obtain the corresponding node location set, unit boundary information and mechanical parameters from the mechanical analysis model; S3. Using spatial geometric mapping method and hierarchical binding strategy, establish deep semantic association between multi-dimensional key assembly features and mechanical analysis model nodes in three-dimensional data, and complete the global association between geometric features and mechanical nodes; S4. A two-stage registration strategy of "coarse registration + fine registration" is adopted to virtually assemble the three-dimensional data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model; S5. Based on the actual stress conditions of the component at the virtual positioning position, perform mechanical simulation calculations on the mechanical analysis model to obtain the mechanical parameters of each node in the model; S6. Based on the global correlation between geometric features and mechanical nodes, the nodal displacement vectors obtained from mechanical simulation are superimposed on the key assembly features of the three-dimensional data, and the feature coordinates are updated to obtain the assembly results that reflect the real stress state. The multi-dimensional key assembly features in S2 include: Zero-dimensional point features include bolt hole center points, component corner points, measurement reference points, and temporary hoisting points; One-dimensional line features include plate element edge lines, arch rib center axis, welds, box girder outline, and component edge lines; Two-dimensional surface features include splicing contact surfaces, web planes, top and bottom plate planes, and curved skin areas; The process of establishing a global association between geometric features and mechanical nodes in S3 includes a hierarchical binding strategy for features of different dimensions: For zero-dimensional point features: For discrete key points in the point cloud, K-nearest neighbor search is performed on the node set of the finite element model; if the distance between the feature point and the nearest node is less than a preset threshold, rigid node constraints are established; if the feature point is located inside the element, the interpolation weight is calculated using the shape function of the element to establish a weighted coupling relationship between the feature point and the element node. For one-dimensional line features: the edge lines in the point cloud are mapped to the mesh surface of the finite element model using a curve projection algorithm; the element boundaries or node sequences crossed by the projection path are identified, and path constraint associations are established; for features at non-node locations on the path, their degrees of freedom are bound to two adjacent mesh nodes through linear interpolation. For two-dimensional surface features: For continuous plate surfaces or curved surfaces in point clouds, bounding box overlap detection is used to filter candidate unit sets; further, the cosine similarity between the point cloud normal vector and the finite element shell element normal vector is calculated to eliminate nodes with inconsistent normals; the filtered finite element node set is defined as a subordinate surface region, and a region mapping relationship is established with the point cloud surface features.

2. The digital assembly method for large-segment bridge components integrating point cloud and mechanical simulation according to claim 1, characterized in that, The types of large components in S1 include bridge components such as bolted, welded, or concrete trusses, box girders, and arch ribs, as well as mixed irregular segments composed of the above structures. The mechanical analysis model adopts the finite element analysis model, which is constructed by using beam elements, shell elements or solid elements, and its mesh density meets the mapping accuracy requirements of the assembly features.

3. The digital assembly method for large-segment bridge components integrating point cloud and mechanical simulation according to claim 1, characterized in that, The specific implementation process of the two-stage registration strategy in S4 is as follows: In the coarse registration stage, let the set of homologous feature points selected in the 3D data model of the component to be assembled be... ,in , Let be the three-dimensional coordinates of the feature points in the local coordinate system of the component, and let be the set of homologous feature points corresponding to the design positions of the overall structural model. ,in These are the three-dimensional coordinates of the feature points in the global coordinate system of the overall structure. Compute point set center of mass Calculate the point set center of mass Construct a decentralized set of nodes and Establish the covariance matrix ,right Singular value decomposition yields ,in and It is an orthogonal matrix. Given a diagonal matrix; then solve for the rotation matrix. With translation vector ,pass The initial adjustment of the component's attitude was completed, and coarse registration was achieved; In the fine registration stage, the component posture obtained after coarse registration is taken as the initial state, and the point cloud set corresponding to the multi-dimensional key assembly features in the component's 3D data is set as follows: ,in The real-time coordinates of the feature point cloud in the global coordinate system; the point cloud set corresponding to the position of the overall structural model design is as follows: ,in To design the coordinates of the feature point cloud in the global coordinate system; Define the iterative objective function as: ,in Represents Euclidean distance. For the first The coordinates of the feature point cloud after the next iteration and For the first The rotation matrix and translation vector for each iteration; During the iteration process, based on the definition of points, lines, and surfaces according to the multi-dimensional key assembly feature attributes, for each exist Matching corresponding homologous feature points By eliminating outlier matching pairs corresponding to non-homologous features, and minimizing... renew and Repeat the iteration until... Less than the preset accuracy threshold ,in Determined based on the accuracy of 3D data preprocessing and the tolerance of assembly design; Finally passed Achieve high-precision alignment between the 3D data model of the component and its bound mechanical analysis model and the overall structural model at the specified location in the design. and These are the rotation matrix and translation vector obtained from the final iteration.

4. The digital assembly method for large-segment bridge components integrating point cloud and mechanical simulation according to claim 1, characterized in that, The actual stress conditions in S5 include: the gravity of the component itself, the thermal expansion and contraction effect caused by changes in ambient temperature, the hoisting condition, and the forced displacement boundary conditions generated when forced assembly is performed to eliminate manufacturing errors.

5. The digital assembly method for large-segment bridge components integrating point cloud and mechanical simulation according to claim 1, characterized in that, S6 further includes: calculating the deviation of the assembly features of adjacent components and generating an assembly adjustment scheme adapted to the engineering requirements. If the characteristic geometric deviation or nodal stress level exceeds the preset threshold, the virtual pose parameters of the component are fine-tuned with the goal of minimizing the deviation and stress value. The mechanical simulation and deformation calculation and related operations of this step are executed repeatedly until the preset tolerance requirements are met. Finally, the deformed three-dimensional assembly model, mechanical parameter analysis results and digital assembly accuracy report are output.

6. The digital assembly method for large-segment bridge components integrating point cloud and mechanical simulation according to claim 5, characterized in that, The specific process for fine-tuning the virtual pose parameters of the component is as follows: With the objective of minimizing the deviation of the assembly features of adjacent components and the nodal stress values, let the set of assembly feature points of adjacent components after deformation correction be as follows: and ,in The three-dimensional coordinates of the assembly feature points of the first component are given. Given the three-dimensional coordinates of the assembly feature points corresponding to the second component, calculate the deviation of a single feature point. This leads to the overall feature geometric deviation evaluation index. ; Let the stress values ​​at each node in the mechanical analysis model be... ,in, Define a nodal stress level evaluation index for the total number of nodes. Construct the objective function ,in This is the normalization coefficient for stress level and geometric deviation, used to bring the two evaluation indicators to the same order of magnitude; Set the virtual pose adjustment parameters of the component to three-dimensional rotation angle. and three-dimensional translation ,in Respectively around the global coordinate system , , The rotation angle of the shaft, respectively along the global coordinate system , , The translation distance of the axis; Based on the objective function Solving using the gradient descent method makes Minimize the pose adjustment parameters and calculate the partial derivatives of the objective function with respect to each adjustment parameter. The parameter adjustment trend is determined based on the direction of the partial derivatives, and the parameters are updated gradually. and The possible values ​​of ; Each time the pose adjustment parameters are updated, the virtual pose of the component's 3D data model and its bound mechanical analysis model are updated synchronously. Mechanical simulation calculations are then re-executed to obtain new nodal displacement vectors and stress values. Based on the global correlation between geometric features and mechanical nodes, the assembly feature coordinates are corrected, and the calculations are repeated. and And update the objective function. ; Repeat the above process of parameter updating, pose adjustment, simulation calculation, and deviation and stress evaluation until... Less than the preset geometric tolerance threshold and The stress is less than the preset safety threshold, at which point the obtained and This is the optimal virtual pose adjustment parameter.

7. A digital assembly system for large bridge segmental components integrating point cloud and mechanical simulation, characterized in that, include: The data input and preprocessing unit is used to acquire the measured three-dimensional data of the large component to be assembled and complete the preprocessing; at the same time, it acquires the theoretical mechanical analysis model corresponding to the component. The multidimensional feature and mechanical analysis node extraction unit is used to perform two operations in parallel: identify and extract multidimensional key assembly features from the preprocessed three-dimensional data, and obtain the corresponding node position set, unit boundary information and mechanical parameters from the mechanical analysis model. The multidimensional feature and mechanical node semantic binding unit is used to establish a deep semantic association between multidimensional key assembly features and mechanical analysis model nodes in three-dimensional data by adopting spatial geometric mapping method and hierarchical binding strategy, and to complete the global association between geometric features and mechanical nodes. The component virtual positioning and attitude calibration unit is used to virtually assemble the three-dimensional data model of the component and its bound mechanical analysis model to the design-specified position of the overall structural model using a two-stage registration strategy of "coarse registration + fine registration". The multi-condition mechanical simulation and deformation calculation unit is used to perform mechanical simulation calculations on the mechanical analysis model for the actual stress conditions of the component at the virtual positioning position, and to obtain the mechanical parameters of each node in the model. The deformation correction unit is used to superimpose the nodal displacement vector obtained from mechanical simulation onto the key assembly features of the three-dimensional data based on the global correlation between geometric features and mechanical nodes, and update the feature coordinates to obtain the assembly result that reflects the real stress state. The multidimensional key assembly features in the multidimensional feature and mechanical analysis node extraction unit include: Zero-dimensional point features include bolt hole center points, component corner points, measurement reference points, and temporary hoisting points; One-dimensional line features include plate element edge lines, arch rib center axis, welds, box girder outline, and component edge lines; Two-dimensional surface features include splicing contact surfaces, web planes, top and bottom plate planes, and curved skin areas; The process of completing the global association between geometric features and mechanical nodes in the multidimensional feature and mechanical node semantic binding unit includes a hierarchical binding strategy for features of different dimensions: For zero-dimensional point features: For discrete key points in the point cloud, K-nearest neighbor search is performed on the node set of the finite element model; if the distance between the feature point and the nearest node is less than a preset threshold, rigid node constraints are established; if the feature point is located inside the element, the interpolation weight is calculated using the shape function of the element to establish a weighted coupling relationship between the feature point and the element node. For one-dimensional line features: the edge lines in the point cloud are mapped to the mesh surface of the finite element model using a curve projection algorithm; the element boundaries or node sequences crossed by the projection path are identified, and path constraint associations are established; for features at non-node locations on the path, their degrees of freedom are bound to two adjacent mesh nodes through linear interpolation. For two-dimensional surface features: For continuous plate surfaces or curved surfaces in point clouds, bounding box overlap detection is used to filter candidate unit sets; further, the cosine similarity between the point cloud normal vector and the finite element shell element normal vector is calculated to eliminate nodes with inconsistent normals; the filtered finite element node set is defined as a subordinate surface region, and a region mapping relationship is established with the point cloud surface features.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed by a processor as described in any one of claims 1-6: a digital assembly method for large bridge segment components that integrates point cloud and mechanical simulation.

Images