A precise construction method for a complex special-shaped structure based on BIM

Abstract

This invention discloses a precise construction method for complex irregular structures based on BIM, belonging to the field of digital construction technology. A multi-scale BIM parametric model is established, and irregular curved surfaces are classified into control patches according to curvature and assigned unique spatial codes to form a coded digital benchmark model. Through rigid registration and non-rigid iterative matching of site point cloud data with the benchmark model, a dynamically corrected BIM model is constructed by integrating site and component prefabrication deviations. Based on the corrected model, installation control points are extracted, a multi-factor error positive prediction model is constructed, six-degree-of-freedom pre-adjustment parameters are generated, component coordinates are collected and compared with the model, and a spatial pose closed-loop control link is constructed. Using patch codes as indexes, the deviation data throughout the entire process is integrated with the as-built model to form a digital as-built result with full lifecycle accuracy traceability. This invention achieves high-precision management and control of irregular structures throughout the entire process, from benchmark modeling, dynamic correction, error prediction, closed-loop control to as-built traceability.

Description

Technical Field

[0001] This invention relates to the field of digital construction technology, specifically to a precise construction method for complex irregular structures based on BIM. Background Technology

[0002] Large-span spatial structures, free-form curtain walls, and complex irregular steel structures have become the mainstream forms for super high-rise landmarks and large public buildings. Their forms are characterized by multi-curvature coupling, strict spatial positioning constraints, and extremely high interface precision requirements. Traditional construction relies on two-dimensional drawing layout, manual measurement and correction, experience-based pre-adjustment, and post-construction inspection. This results in distorted surface representation, large deviations in point transmission, accumulation of errors from multiple processes, and delayed correction response, making it difficult to achieve millimeter-level installation accuracy and consistent forming. This has become the core bottleneck restricting the high-quality construction of complex irregular structures.

[0003] While digital construction currently widely employs BIM modeling, 3D laser scanning, and virtual pre-assembly, it suffers from systemic deficiencies in the precision control of irregular structures throughout the entire process: the BIM static design model lacks dynamic linkage with on-site measurements, component prefabrication deviations, structural deformation, and environmental disturbances; point clouds and models often use rigid registration, which cannot adapt to the non-rigid deformation of components and site errors; there is a lack of unified quantitative prediction models for temperature, welding shrinkage, self-weight, and wind disturbance coupled deformation; and there is no real-time posture closed-loop control for hoisting and positioning, resulting in delayed correction, frequent deviations, high rework costs, and the lack of an integrated technical system.

[0004] The academic community has achieved a series of high-level research results on the precision control of irregular structures. Liu Junjin et al., in their 2022 paper "A Precision Control Method for Steel Structures Based on the Fusion of 3D Laser Scanning and BIM" published in the *Journal of Civil Engineering*, systematically elaborated on the point cloud-BIM registration and deviation detection process. Du Zhenyu et al., in their 2023 paper "Multi-factor Deformation Prediction and Installation Pre-adjustment Technology for Large-Span Irregular Curved Surface Structures" published in the *Journal of Building Structures*, constructed a coupled deformation calculation model for temperature, welding, and self-weight. These studies confirm that multi-scale modeling, non-rigid matching, and positive error prediction can significantly improve accuracy; however, they are mostly limited to single steps or algorithms and have not formed a closed-loop construction method that can be implemented in engineering, lacking systematicity and practicality.

[0005] Chinese patent CN117947954A discloses a high-precision installation method for steel structures based on BIM and 3D laser scanning. This method uses BIM and point cloud registration to detect deviations and guide installation adjustments, improving the positioning efficiency of irregularly shaped components to some extent. However, this technology still has significant shortcomings: it does not implement multi-level patch coding and refined coordinate control for irregular curved surfaces; it does not establish a multi-factor positive error prediction model for the coupling of temperature, welding shrinkage, self-weight, and wind disturbance; it does not construct a spatial pose closed-loop control link during hoisting; it does not support dynamic correction of hydraulic supports; and it lacks a full life-cycle accuracy traceability mechanism, making it difficult to meet the construction needs of ultra-high precision complex irregularly shaped structures.

[0006] Therefore, there is an urgent need for self-controllable accuracy throughout the entire process from design benchmarks to project completion traceability, so as to fundamentally solve the industry problems of low construction accuracy, delayed correction, difficult deformation prediction, and untraceable data for complex and irregular structures. Summary of the Invention

[0007] To address the aforementioned technical issues, this application discloses a BIM-based method for precise construction of complex irregular structures, specifically including the following steps: Establish a multi-scale BIM parametric model for complex irregular structures, decompose the irregular curved surface into multi-level control patches, and assign a unique spatial code and theoretical coordinate system to each patch to form a coded digital benchmark model. Obtain point cloud data of benchmark points, constructed structures and components to be installed at the construction site. Construct a dynamically corrected BIM model that includes site deviations and component prefabrication deviations through rigid registration and non-rigid iterative matching of point clouds and BIM digital benchmark models. Based on the dynamically corrected BIM model, the target points for component installation are extracted. Combined with the effects of temperature field, welding shrinkage, structural self-weight and wind disturbance coupling, a multi-factor error positive prediction model is established to generate component installation pre-adjustment parameters. During the component hoisting and positioning process, the real-time coordinates of key points of the component are collected and compared with the dynamically corrected BIM model in real time to construct a spatial pose closed-loop control link. When the real-time pose deviation exceeds the preset threshold, the fine-tuning compensation amount is output, and the spatial attitude of the component is dynamically corrected through the adjustable support system until the deviation converges to the allowable range. After construction is completed, the coded deviation data and correction records of the entire process will be integrated with the as-built model to form a digital as-built result with full life-cycle accuracy traceability.

[0008] Preferably, the step of decomposing the irregular curved surface into multi-level control patches specifically includes: Based on the rate of curvature change of the irregular surface, a curvature threshold is set. The first region with curvature higher than the curvature threshold is divided into first-level patches, and the second region with curvature lower than or equal to the curvature threshold is divided into second-level patches. The first-level surface is fitted with a quadratic surface, and the second-level surface is fitted with a planar or single-curved surface. A unique spatial code containing structural hierarchy, spatial quadrant and sequence number is generated for each first-level and second-level surface.

[0009] Preferably, the step of constructing a dynamically corrected BIM model through rigid registration and non-rigid iterative matching of point clouds and BIM digital benchmark models specifically includes: The singular value decomposition algorithm is used to rigidly register the point cloud data with the BIM digital benchmark model to obtain the initial transformation matrix. Based on the initial transformation matrix, an energy function for non-rigid iterative matching is constructed. The non-rigid deformation field is solved by minimizing the energy function. The non-rigid deformation field is then applied to the BIM digital benchmark model to generate a dynamically corrected BIM model.

[0010] Preferably, the step of establishing a multi-factor error positive prediction model to generate component installation pre-adjustment parameters specifically includes: Construct a multi-factor error positive prediction model, the formula is as follows: ,in, For the pre-tuning bit parameter vector, The coefficient of thermal expansion of the material. For the internal temperature gradient field of the component, This is the weld shrinkage coefficient. The residual stress distribution along the weld path, For the component's own weight, The length of the cantilever. For elastic modulus, Let the moment of inertia of the cross section be... This is the wind load shape coefficient. For real-time wind speed, The windward projected area of ​​the component; Calculate the pre-adjustment parameter vector based on real-time collected environmental parameters and component properties. And apply it in reverse to the target installation point of the component.

[0011] Preferably, the construction of the spatial pose closed-loop control link specifically includes: The extended Kalman filter algorithm is used to fuse real-time coordinate data collected by multiple total stations to eliminate measurement noise and obtain the optimal estimated coordinates of key points of the component. Based on the optimal estimated coordinates and the theoretical coordinates in the dynamically corrected BIM model, a pose deviation matrix is ​​constructed, as shown in the formula: ,in, The pose deviation matrix is... For rotational deviation components, For translational deviation components, and These are the measured rotation matrix and translation vector, respectively. and These are the theoretical rotation matrix and translation vector, respectively; The pose deviation matrix is ​​compared element by element with the preset threshold matrix to determine whether the real-time pose deviation exceeds the preset threshold.

[0012] Preferably, the automatic calculation and output of fine-tuning compensation, and the dynamic correction of the spatial attitude of the component through the adjustable support system, specifically includes: When the real-time pose deviation exceeds a preset threshold, the fine-tuning compensation amount is calculated based on the pose deviation matrix, and its expression is: ,in, To fine-tune the compensation vector, This is the proportional gain coefficient. The differential gain coefficient, This is the rotational gain coefficient. The Lie algebraic representation of the rotational deviation matrix; The fine-tuning compensation vector is converted into stroke commands for each hydraulic jack in the adjustable support system, driving the jacks to move and dynamically correcting the spatial posture of the components.

[0013] Preferably, the dynamic correction of the spatial attitude of the component through the adjustable support system includes: Multiple hydraulic jacks are installed at the key stress points of the component, and each hydraulic jack is equipped with a displacement sensor and a pressure sensor. Receive the fine-tuning compensation amount and decompose it into the target stroke of each hydraulic jack; The system collects the actual stroke and support pressure of each hydraulic jack in real time. When the deviation between the actual stroke and the target stroke or the support pressure exceeds the safety threshold, the system pauses the correction and outputs an early warning signal.

[0014] Preferably, the step of linking and fusing the coded deviation data, correction records, and as-built model throughout the entire process includes: Using the unique spatial code of each patch as the index key, the real-time pose deviation, fine-tuning compensation amount, correction timestamp and environmental parameters corresponding to the patch are packaged to generate a deviation data package. By embedding the deviation data package into the corresponding patch attributes of the as-built BIM model, a digital as-built result containing accuracy traceability information throughout the entire lifecycle is formed.

[0015] Preferably, the method further includes: Before the components are hoisted, virtual pre-assembly is carried out based on the dynamically modified BIM model to detect interference between adjacent components; If interference is detected, the pre-positioning parameters of the component to be installed are adjusted according to the amount of interference until the virtual pre-assembly is successful.

[0016] Preferably, the method of simultaneously acquiring real-time coordinates of key points of the component using multiple total stations includes: At least three total stations should be set up at the construction site to form a measurement control network; A forced synchronization trigger signal is used to control at least three total stations to simultaneously acquire coordinate data of key points of the component, and the three-dimensional coordinates of the key points are calculated by spatial resection algorithm.

[0017] Compared with the prior art, the technical solution of this application has the following technical effects: This invention constructs a multi-scale BIM parametric model and a coded digital benchmark, which can perform refined spatial decomposition and unified coordinate control of complex irregular curved surfaces, realize the complete digital expression of structural morphology, provide a stable and reliable spatial positioning basis for the entire construction process, and ensure the consistency and traceability of the installation benchmark of irregular structures.

[0018] This invention employs rigid registration and non-rigid iterative matching of point cloud and BIM model, which can effectively integrate actual on-site working conditions and component processing deviations to form a dynamic correction model. This keeps the model information synchronized with the on-site conditions, providing data support that fits the actual engineering conditions for subsequent installation positioning and error prediction.

[0019] This invention establishes a multi-factor error positive prediction model and a spatial pose closed-loop control link, which can predict the structural deformation trend in advance and generate pre-adjustment parameters. During the hoisting process, the component posture is monitored in real time and the correction amount is automatically calculated. Dynamic adjustment is completed with the help of an adjustable support system to ensure that the component installation posture is continuously stable and controllable.

[0020] This invention integrates deviation data from the entire construction process with the as-built model, using a unique spatial code as an index to form a digital as-built result. This enables the full lifecycle recording and retention of structural installation accuracy and the correction process, providing complete and reliable data support for project acceptance, subsequent operation and maintenance, and structural health monitoring.

[0021] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the preferred embodiments of this application are described in detail below with reference to the accompanying drawings.

[0022] The above and other objects, advantages and features of this application will become more apparent to those skilled in the art from the following detailed description of specific embodiments in conjunction with the accompanying drawings. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0024] Based on the description of the figures and their corresponding technical content in the document, the titles of the figures are as follows: Figure 1Schematic diagram of the entire process architecture for multi-scale BIM-based precision construction of complex irregular structures; Figure 2 Schematic diagram of multi-level patch division and coding benchmark construction of irregular curved surfaces; Figure 3 Schematic diagram of multi-level preprocessing and dynamic correction of point cloud data for BIM model construction; Figure 4 Schematic diagram of multi-factor error prediction and installation pre-adjustment parameter generation control. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. In the following description, specific details such as specific configurations and components are provided merely to help fully understand the embodiments of this application. Therefore, those skilled in the art should understand that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. In addition, for clarity and brevity, descriptions of known functions and structures are omitted in the embodiments.

[0026] It should be understood that the phrase "an embodiment" or "this embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "an embodiment" or "this embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0027] Furthermore, reference numerals and / or letters may be repeated in different examples within this application. Such repetition is for the purpose of simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or settings discussed.

[0028] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" in this article describes another type of relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the related objects before and after it are in an "or" relationship.

[0029] In this article, the term "at least one" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, "at least one of A and B" can mean: A exists alone, A and B exist simultaneously, or B exists alone.

[0030] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion.

[0031] Example 1 This embodiment mainly describes a precise construction method for complex irregular structures based on BIM, such as... Figure 1 As shown, it specifically includes: Establish a multi-scale BIM parametric model for complex irregular structures, decompose the irregular curved surface into multi-level control patches, and assign a unique spatial code and theoretical coordinate system to each patch to form a coded digital benchmark model. Obtain point cloud data of benchmark points, constructed structures and components to be installed at the construction site. Construct a dynamically corrected BIM model that includes site deviations and component prefabrication deviations through rigid registration and non-rigid iterative matching of point clouds and BIM digital benchmark models. Based on the dynamically corrected BIM model, the target points for component installation are extracted. Combined with the effects of temperature field, welding shrinkage, structural self-weight and wind disturbance coupling, a multi-factor error positive prediction model is established to generate component installation pre-adjustment parameters. During the component hoisting and positioning process, the real-time coordinates of key points of the component are collected and compared with the dynamically corrected BIM model in real time to construct a spatial pose closed-loop control link. When the real-time pose deviation exceeds the preset threshold, the fine-tuning compensation amount is output, and the spatial attitude of the component is dynamically corrected through the adjustable support system until the deviation converges to the allowable range. After construction is completed, the coded deviation data and correction records of the entire process will be integrated with the as-built model to form a digital as-built result with full life-cycle accuracy traceability.

[0032] Furthermore, based on the construction drawings, 3D geometric model, structural calculation sheets, material performance parameters, boundary constraints, design load conditions, construction conditions, and installation accuracy requirements of the complex irregular structure, a five-level multi-scale BIM parametric model is constructed using a BIM parametric modeling engine. This model covers the overall structural scale, the scale of segmented components, the detailed scale of node connections, the detailed scale of weld construction, and the scale of embedded parts positioning. The model adopts a hybrid modeling method of NURBS surfaces and BREP boundary representation entities. The number of surface control points is no less than 8,000, and the number of entity topological surfaces is no less than 3,000. The geometric analytical calculation accuracy of the model is set to 0.0005 mm. The model node coordinates are stored using double-precision 64-bit floating-point numbers, with 14 decimal places retained. The model coordinate system is strictly unified with the engineering coordinate system of the construction site, the coordinate origin coincides with the site's primary control network point, and the coordinate axis direction is completely consistent with the design coordinate system. Parametric differential geometry analysis is performed on the entire irregular surface. The parameters u and v are used as variables to traverse the entire surface. The sampling step size is set to 0.5 mm. The Gaussian curvature, mean curvature, principal curvature, and rate of change of curvature are calculated for each discrete sampling point. The rate of change of curvature is solved by jointly calculating the surface parameter biased derivatives, mixed biased derivatives, and the normal vector. The formula is as follows: In the formula: Let u be the first-order deflection vector of the surface parameter u. Let v be the first-order deflection vector of the surface parameter v. For the second-order deflection of the surface parameter u, For the uv-axis hybrid second-order biased guide coefficient of the surface parameters, For the second-order deflection of the surface parameter v, The result is the vector cross product of the parametric bias vectors in the u and v directions. This is the magnitude of the surface normal vector; like Figure 2 As shown, the curvature threshold is comprehensively set based on the statistical characteristics of the curvature distribution of irregular curved surfaces, the distribution of key structural stress areas, the installation control accuracy level, and the feasibility of on-site construction. , The value is fixed at 0.012 mm⁻¹, and the curvature calculation result is greater than... The continuous connected regions are divided into first-order patches, and the curvature calculation results are less than or equal to The continuous connected region is divided into two-level patches. The first-level patches correspond to regions with high curvature, high deformation sensitivity, and high precision control, with a minimum division unit area of ​​25cm² and a boundary fitting error of ≤0.03mm. The second-level patches correspond to regions with low curvature, low sensitivity, and conventional precision control, with a minimum division unit area of ​​100cm² and a boundary fitting error of ≤0.08mm.

[0033] After the facets are divided, topological relationship reconstruction and boundary consistency checks are performed to eliminate facet overlaps, gaps, jumps, and inconsistent normal vectors, ensuring that the deviation of the normal vector angle between adjacent faces is ≤0.05° and the overlap of boundary node coordinates is ≤0.001mm. A quadratic surface global fitting algorithm is used for geometric reconstruction of the first-level faces. The general standard fitting equation for quadratic surfaces is: In the formula: The coefficient of the quadratic term, The coefficient of the linear term, The constant term coefficients are obtained by solving the orthogonal least squares method based on a uniformly distributed discrete point set within the patch. An overdetermined system of equations is constructed and solved by QR decomposition. The maximum fitting residual is ≤0.05mm, and the sum of squared residuals is ≤1×10⁻⁻⁴. 6 mm².

[0034] For secondary surface patches, either plane fitting or single-curved surface fitting is used. The standard equation for plane fitting is: In the formula: For the unit normal vector components of the plane, For the planar position constant term, the maximum fitting residual is ≤0.1mm, and the flatness error is ≤0.08mm; For fitting single-curved surfaces, the standard analytical equations of cylindrical and conical surfaces are preferred. The standard equation for a cylindrical surface is: In the formula: Let X be the projected coordinates of the central axis of the cylindrical surface in the XY plane. The radius of the cylindrical surface is calculated with an accuracy of 0.001 mm, and the straightness error of the axis is ≤0.05 mm. A globally unique and permanent spatial code is generated for each first-level and second-level surface. The code structure consists of 9 decimal digits. The first 2 digits are the structural level code, ranging from 01 to 99, corresponding to five levels: overall structure, zone, component, unit, and node. The middle 3 digits are the spatial quadrant code, divided into 8 main quadrants, 24 sub-quadrants, and 120 sub-quadrants based on the three-dimensional rectangular coordinate system of the construction site, ranging from 001 to 999. The last 4 digits are the continuous sequence number of the surface, ranging from 0001 to 9999. After the code is generated, it is strongly bound to the surface's theoretical coordinates, curvature parameters, fitting parameters, boundary node coordinates, topological adjacency relationships, and accuracy level to form a coded digital benchmark model. All surfaces in the model have independent index identifiers, spatial positioning benchmarks, geometric attribute sets, and accuracy control standards.

[0035] Furthermore, a ground-based long-range 3D laser scanner, a handheld close-range 3D laser scanner, and a high-precision structured light 3D scanner were used to collaboratively collect point cloud data of the primary and secondary benchmark points, the main structure already constructed, installed components, embedded parts, and components to be installed at the construction site. The sampling point spacing of the scanned point cloud was set to 0.2 mm, the mean square error of single-point coordinate measurement was ≤ ±0.03 mm, and the total number of point cloud data collected was not less than 1.2 × 10⁻⁶. 8 The point cloud normal vector calculation accuracy is ≤0.1°. like Figure 3 As shown, after the data acquisition is completed, multi-level preprocessing is performed on the point cloud data. The first stage is outlier removal, which uses the k-neighbor standard deviation discrimination method. The number of neighboring points k=16, the standard deviation multiple is 3.0, and extreme outliers with a probability of less than 0.1% are removed. The second stage is Gaussian filtering, with a convolution kernel size of 3×3 and a filter window standard deviation σ=1.2, to suppress measurement noise and environmental interference noise; The third stage is voxel mesh simplification, with the voxel resolution set to 0.3mm. While maintaining the geometric shape, the amount of data redundancy is reduced, and the data compression ratio after simplification is ≥60%. The fourth stage is the recalculation of normal vectors. The normal vectors are reconstructed by fitting the neighborhood point set with least squares plane fitting, ensuring that the consistency of normal vectors is ≥98.5%. After preprocessing, the effective data ratio of point cloud is ≥99.2%, the effective point cloud coordinate error is ≤±0.04mm, and there are no obvious noise, holes, jitter, or misalignment phenomena.

[0036] The preprocessed point cloud dataset is rigidly registered with the coded digital reference model. Singular Value Decomposition (SVD) is used to perform high-precision registration calculations. First, based on curvature features, boundary features, and extreme point features, corresponding feature point pairs between the point cloud and the BIM model are automatically extracted. The total number of feature point pairs is no less than 1200 pairs, evenly distributed across the entire structure to avoid local concentration. The point cloud dataset is then calculated separately. With BIM model point set The centroid coordinates are calculated using the following formula: In the formula: This represents the total number of feature point pairs. The three-dimensional coordinates of the feature points in the point cloud. The three-dimensional coordinates of the feature points corresponding to the BIM model.

[0037] The covariance matrix is ​​constructed based on the decentralized point set, and its expression is: For the covariance matrix Perform full singular value decomposition: In the formula: It is an orthogonal matrix. Given a non-negative singular value diagonal matrix, find the optimal rotation matrix. With translation vector The formula is: After rigid registration, the average global registration deviation is ≤0.15mm, the maximum registration deviation is ≤0.3mm, the rotation angle error is ≤±2″, and the translation error is ≤±0.08mm. A non-rigid iterative matching energy function is constructed based on the initial transformation matrix obtained from rigid registration. The energy function is a weighted combination of data error terms and regularization constraint terms. In the formula: This refers to the data error term between the point cloud and the model. For deformation smoothing regularization, The weighting coefficient ranges from 0.08 to 0.15 and is adaptively adjusted based on the surface curvature. The formula for the data error term is: ; The formula for the regularization constraint term is: In the formula: It is a non-rigid deformation mapping function. Let i be a spherical neighborhood set of points, with a fixed neighborhood search radius of 5 mm. The Gauss-Newton iterative method is used to minimize the energy function, with an adaptively decreasing iteration step size. The initial step size is 0.1, and the minimum step size is 1 × 10⁻⁻⁻⁶. 6 The iteration termination condition is that the change in the energy function is less than 1 × 10⁻ for three consecutive iterations. 6 The maximum number of iterations is capped at 80, resulting in a globally smooth, continuous, distortion-free, and abrupt non-rigid deformation field. This deformation field is applied node-by-node, patch-by-patch, and control vertex-by-control-vertex to the BIM digital baseline model, correcting the model's geometry point-by-point. Simultaneously, patch codes, coordinate systems, topological relationships, and geometric attributes are updated. This generates a fully dynamic, corrected BIM model that includes site construction deviations, component prefabrication deviations, cumulative installation positioning deviations, and structural foundation settlement deviations. The overall spatial positioning error of the dynamically corrected BIM model is ≤±0.1mm, the patch geometric conformity is ≥99.7%, and the node coordinate update rate is 100%.

[0038] Furthermore, such as Figure 4 As shown, the three-dimensional theoretical coordinates of component installation target points, control node points, boundary constraint points, weld positioning points, and embedded part center points are extracted in batches from the dynamically corrected BIM model according to the coded index. The point extraction density is no less than 15 control points per linear meter and no less than 30 control points per square meter of curved surface. The coordinate storage format is double-precision floating-point numbers. A multi-factor error positive prediction model is constructed under the coupled effects of four fields: temperature field, welding shrinkage, structural self-weight, and wind disturbance. The overall formula of the model remains unchanged. In the formula: It is a six-degree-of-freedom pre-adjustment position parameter vector, which includes three translational components in the X, Y, and Z directions and three rotational components in the Rx, Ry, and Rz directions; The coefficient of thermal expansion of the material is taken as 1.2 × 10⁻⁻. 5 / ℃, for stainless steel, the value is 1.6×10⁻ 5 / ℃, for aluminum alloys, the value is 2.3×10⁻ 5 / ℃; The three-dimensional temperature gradient field inside the component is collected in real time by an array of temperature sensors. The volume of the component element; The weld shrinkage coefficient is 0.05-0.12 mm / kN, which is determined based on the weld height, groove type, and welding process. The length of the weld path element; The value is 0.92 to 1.08, which is the self-weight deformation correction factor, and is taken as 0.92 to 1.08 according to the cantilever state and support conditions of the structure. The self-weight load of the component is automatically calculated from the BIM model; The cantilever length of the component; For the elastic modulus of the material, 2.06 × 10⁻⁶ is taken as 2.06 × 10⁻⁶. 5 N / mm²; The moment of inertia of the component section; The wind load shape coefficient is determined by taking a value of 0.8 to 1.3 based on the surface shape of the component and its windward angle. Real-time wind speed, collected in real time by an anemometer; The windward projected area of ​​the component is automatically calculated from the BIM model. Each error component is independently solved and coupled together. The complete integral formula for the temperature deformation component is as follows: ; Complete integral formula for welding shrinkage deformation component: In the formula: The residual welding stress distributed along the weld path is determined by the coupling calculation of welding process parameters and thermodynamics.

[0039] Complete formula for structural self-weight deformation components: ; Complete formula for wind disturbance deformation components: ; The total pre-adjustment parameter vector is obtained by vector superposition of the four components. The pre-adjustment parameters are applied to the component installation target point in a reverse compensation manner to form an installation control target point with multi-factor error compensation. The point pre-adjustment compensation accuracy is ≤ ±0.05mm, and the attitude pre-adjustment compensation accuracy is ≤ ±3. Furthermore, at least three high-precision fully automatic total stations will be deployed at the construction site to form a spatially redundant measurement control network. The total station's angle measurement accuracy will be ≤ ±0.5, and its distance measurement accuracy will be ≤ ±(0.3mm + 1×10⁻⁻⁴).6 • D) The total station deployment height meets the conditions of unobstructed line of sight, full observation coverage, and redundant observation. The station spacing is 15m~25m, forming a stable triangular measurement structure. Hardware synchronous triggering signals are used to achieve synchronous data acquisition by multiple total stations, with a synchronous triggering time error ≤1ms. This ensures that all total stations acquire the coordinate data of key points of the component at the same time. The three-dimensional coordinates of the key points of the component are jointly calculated using a spatial resection algorithm. The calculation formula is: In the formula: The number of total stations. The original coordinates of a single total station. The normalization coefficient is... The extended Kalman filter algorithm is used to suppress noise, remove outliers, and estimate the optimal state of multi-source measurement data. The system state equation and observation equation are as follows: In the formula: Let k be the system state vector at time k, which includes spatial position, velocity, attitude angle, and attitude angular velocity. This is the state transition matrix; To control the input matrix; External control vector; For process noise, the variance matrix ; For observation vectors; The observation matrix; For observation noise, the variance matrix .

[0040] The filtering, updating, prediction, and correction process has an execution cycle of 50ms. After filtering, the optimal estimated coordinates of the key points of the component are obtained, with the optimal coordinate estimation error ≤ ±0.04mm and the optimal attitude estimation error ≤ ±2″. Based on the optimal estimated coordinates and the theoretical coordinates of the dynamically corrected BIM model, a homogeneous pose deviation matrix is ​​constructed, with the following formula: In the formula: This is the pose deviation matrix; It is a 3×3 rotational deviation component matrix; It is a 3×1 translational deviation component vector; This is the measured rotation matrix; The measured translation vector; This is the theoretical rotation matrix; This is the theoretical translation vector.

[0041] Construct a preset threshold matrix: In the formula: The allowable threshold for rotational deviation is set to ±15″. The allowable threshold for translational deviation is set to ±0.3 mm. The pose deviation matrix is ​​compared element-by-element with the threshold matrix. or If the real-time pose is determined to be outside the allowable range, the dynamic correction control process is immediately initiated.

[0042] When the pose deviation exceeds the limit, the fine-tuning compensation vector is calculated based on the pose deviation matrix. The compensation amount is solved using a proportional-derivative-rotation composite control algorithm, and the formula remains unchanged: In the formula: This is a six-degree-of-freedom fine-tuning compensation vector; This is the proportional gain coefficient, with a value ranging from 800 to 1200; This is the differential gain coefficient, with a value ranging from 20 to 50; This is the rotational gain coefficient, with a value ranging from 30 to 60; Let be the Lie algebra representation of the rotational deviation matrix.

[0043] Complete computation formula for Lie algebraic mappings: In the formula: For the equivalent rotation angle, satisfying , Let be the trace of the rotational deviation matrix. The fine-tuning compensation vector is projected and decomposed onto the direction of action of each hydraulic jack in the adjustable support system. The decomposition formula is: In the formula: Let i be the target stroke of the i-th hydraulic jack; Let be the unit vector representing the direction of action of the i-th jack. Hydraulic jacks are evenly distributed at key stress points, control points, and constraint points of the component, with a minimum of six jacks. The support point layout satisfies the requirements of spatial six-degree-of-freedom adjustable attitude, uniform stress distribution, and no additional bending moment. Each jack is equipped with a high-precision grating displacement sensor and a strain gauge pressure sensor. The displacement sensor has a resolution of 0.001 mm, a measurement range of ±50 mm, and a linearity ≤0.02%FS; the pressure sensor has an accuracy of 0.1%FS and a response time ≤1 ms.

[0044] Real-time data collection of the actual stroke of each jack Support and pressure Construct dual security constraint judgment conditions: In the formula: The travel safety deviation threshold is set to 0.05 mm. To support the pressure safety threshold, 80% of the jack's rated load is used. If the safety conditions are not met, the correction drive signal is immediately cut off, all actions are stopped, and an audible and visual warning signal is output. If the safety conditions are met, the jack is continuously driven to dynamically adjust its spatial attitude. This process is repeated until the position deviation converges to the allowable threshold range. The convergence criterion is that the deviation is less than the threshold for 5 consecutive control cycles, and the cycle length is fixed at 50ms.

[0045] Using the unique spatial code of each facet as the core index key, a standardized deviation data package is generated by packaging the real-time pose deviation, fine-tuning compensation, correction action timestamps, ambient temperature, real-time wind speed, jack target stroke, actual jack stroke, support pressure, original coordinate data, filtered data, registration error, and fitting error of the corresponding facet throughout the entire construction process. The data package adopts a binary compact storage format, with a single point data length of 64 bytes, a data storage accuracy of 0.001mm, a timestamp accuracy to the millisecond level, and a data sampling frequency of 20Hz. The deviation data package is then associated with the corresponding facet nodes, control nodes, support nodes, and weld nodes in the as-built BIM model through attribute embedding. This achieves the integrated fusion of the as-built geometric model, coding index system, full-process deviation data, correction control records, environmental monitoring data, sensor raw data, and accuracy evaluation data, forming a digital as-built result with full life-cycle accuracy traceability capabilities. The digitalized as-built deliverables include: as-built BIM geometric model, patch coding and theoretical coordinate library, installation deviation time series dataset, pre-adjustment parameter dataset, closed-loop control parameter dataset, environmental monitoring dataset, sensor raw dataset, correction execution log, error statistics report, and accuracy evaluation report. All data can be quickly retrieved, located, traced, verified, calculated, and exported through coding. The data query response time is ≤100ms, the data integrity is ≥99.99%, and the data storage format supports long-term archiving and cross-platform reading.

[0046] Before the components are officially hoisted into place, full-size virtual pre-assembly and global interference detection are carried out based on the dynamically modified BIM model. This constructs the component's axis-aligned bounding box (AABB) and directed bounding box (OBB). The standard formula for the bounding box is as follows: ; The separation axis theorem is used to perform full-domain interference detection on adjacent components, components and existing structures, components and support systems, and components and embedded parts. The detection covers all 6 main separation axes and 15 diagonal separation axes. The interference determination formula is as follows: In the formula: Minimum safe distance between components The coordinates of the geometric center of the component are, Let be the equivalent geometric radius of the component. When geometric interference is determined to exist, the interference quantity is calculated. The interference calculation accuracy is 0.01mm. The pre-adjustment parameters are iteratively corrected according to the interference magnitude, with a correction step of 0.05mm. Virtual pre-assembly and interference detection are performed once for each correction until there is no interference between all components, the installation gap meets the design requirements, and the positioning accuracy meets the specification requirements.

[0047] This detailed implementation describes how a coded BIM benchmark model, dynamic point cloud correction, multi-factor error prediction, and hoisting closed-loop correction achieve high-precision control of the installation of complex irregular structures, effectively avoiding interference risks, reducing construction deviations, ensuring that component installation accuracy meets standards, and generating full life-cycle accuracy traceability results, thereby improving construction efficiency and project quality.

[0048] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any changes, modifications, substitutions, integrations, and parameter changes made to these embodiments within the spirit and principles of the present invention, without departing from the principles and spirit of the present invention, through conventional substitutions or to achieve the same function, fall within the scope of protection of the present invention.

Claims

1. A precise construction method for complex irregular structures based on BIM, characterized in that, Includes the following steps: Establish a multi-scale BIM parametric model for complex irregular structures, decompose the irregular curved surface into multi-level control patches, and assign a unique spatial code and theoretical coordinate system to each patch to form a coded digital benchmark model. Obtain point cloud data of benchmark points, constructed structures and components to be installed at the construction site. Construct a dynamically corrected BIM model that includes site deviations and component prefabrication deviations through rigid registration and non-rigid iterative matching of point clouds and BIM digital benchmark models. Based on the dynamically corrected BIM model, the target points for component installation are extracted. Combined with the effects of temperature field, welding shrinkage, structural self-weight and wind disturbance coupling, a multi-factor error positive prediction model is established to generate component installation pre-adjustment parameters. The positive prediction model has the following formula: ， in, For the pre-tuning bit parameter vector, The coefficient of thermal expansion of the material. For the internal temperature gradient field of the component, This is the weld shrinkage coefficient. The residual stress distribution along the weld path, For the component's own weight, The length of the cantilever. For elastic modulus, Let the moment of inertia of the cross section be... This is the wind load shape coefficient. For real-time wind speed, The windward projected area of ​​the component; Calculate the pre-adjustment parameter vector based on real-time collected environmental parameters and component properties. And apply it in reverse to the target point of component installation; During the component hoisting and positioning process, the real-time coordinates of key points of the component are collected and compared with the dynamically corrected BIM model in real time to construct a spatial pose closed-loop control link. When the real-time pose deviation exceeds the preset threshold, the fine-tuning compensation amount is output, and the spatial attitude of the component is dynamically corrected through the adjustable support system until the deviation converges to the allowable range. After construction is completed, the coded deviation data and correction records of the entire process will be integrated with the as-built model to form a digital as-built result with full life-cycle accuracy traceability.

2. The method for precise construction of complex irregular structures based on BIM according to claim 1, characterized in that, The process of decomposing the irregular curved surface into multi-level control patches specifically includes: Based on the rate of curvature change of the irregular surface, a curvature threshold is set. The first region with curvature higher than the curvature threshold is divided into first-level patches, and the second region with curvature lower than or equal to the curvature threshold is divided into second-level patches. The first-level surface is fitted with a quadratic surface, and the second-level surface is fitted with a planar or single-curved surface. A unique spatial code containing structural hierarchy, spatial quadrant and sequence number is generated for each first-level and second-level surface.

3. The method for precise construction of complex irregular structures based on BIM according to claim 1, characterized in that, The process of constructing a dynamically corrected BIM model through rigid registration and non-rigid iterative matching between point clouds and the BIM digital benchmark model specifically includes: The singular value decomposition algorithm is used to rigidly register the point cloud data with the BIM digital benchmark model to obtain the initial transformation matrix. Based on the initial transformation matrix, an energy function for non-rigid iterative matching is constructed. The non-rigid deformation field is solved by minimizing the energy function. The non-rigid deformation field is then applied to the BIM digital benchmark model to generate a dynamically corrected BIM model.

4. The method for precise construction of complex irregular structures based on BIM according to claim 1, characterized in that, The construction of the spatial pose closed-loop control link specifically includes: The extended Kalman filter algorithm is used to fuse real-time coordinate data collected by multiple total stations to eliminate measurement noise and obtain the optimal estimated coordinates of key points of the component. Based on the optimal estimated coordinates and the theoretical coordinates in the dynamically corrected BIM model, a pose deviation matrix is ​​constructed, as shown in the formula: ， in, The pose deviation matrix is... For rotational deviation components, For translational deviation components, and These are the measured rotation matrix and translation vector, respectively. and These are the theoretical rotation matrix and translation vector, respectively; The pose deviation matrix is ​​compared element by element with the preset threshold matrix to determine whether the real-time pose deviation exceeds the preset threshold.

5. The BIM-based precise construction method for complex irregular structures according to claim 4, characterized in that, The automatic calculation and output of fine-tuning compensation amounts, through an adjustable support system, dynamically corrects the spatial attitude of the components, specifically including: When the real-time pose deviation exceeds a preset threshold, the fine-tuning compensation amount is calculated based on the pose deviation matrix, and its expression is: ,in, To fine-tune the compensation vector, This is the proportional gain coefficient. The differential gain coefficient, This is the rotational gain coefficient. The Lie algebraic representation of the rotational deviation matrix; The fine-tuning compensation vector is converted into stroke commands for each hydraulic jack in the adjustable support system, driving the jacks to move and dynamically correcting the spatial posture of the components.

6. The method for precise construction of complex irregular structures based on BIM according to claim 1, characterized in that, The dynamic correction of the spatial attitude of the component through an adjustable support system includes: Multiple hydraulic jacks are installed at the key stress points of the component, and each hydraulic jack is equipped with a displacement sensor and a pressure sensor. Receive the fine-tuning compensation amount and decompose it into the target stroke of each hydraulic jack; The system collects the actual stroke and support pressure of each hydraulic jack in real time. When the deviation between the actual stroke and the target stroke or the support pressure exceeds the safety threshold, the system pauses the correction and outputs an early warning signal.

7. The method for precise construction of complex irregular structures based on BIM according to claim 1, characterized in that, The process of integrating and linking coded deviation data, correction records, and as-built models throughout the entire process includes: Using the unique spatial code of each patch as the index key, the real-time pose deviation, fine-tuning compensation amount, correction timestamp and environmental parameters corresponding to the patch are packaged to generate a deviation data package. By embedding the deviation data package into the corresponding patch attributes of the as-built BIM model, a digital as-built result containing accuracy traceability information throughout the entire lifecycle is formed.

8. The method for precise construction of complex irregular structures based on BIM according to claim 1, characterized in that, The method further includes: Before the components are hoisted, virtual pre-assembly is carried out based on the dynamically modified BIM model to detect interference between adjacent components; If interference is detected, the pre-positioning parameters of the component to be installed are adjusted according to the amount of interference until the virtual pre-assembly is successful.

9. The BIM-based precise construction method for complex irregular structures according to claim 1, characterized in that, The method of simultaneously acquiring real-time coordinates of key points of the component using multiple total stations includes: At least three total stations should be set up at the construction site to form a measurement control network; A forced synchronization trigger signal is used to control at least three total stations to simultaneously acquire coordinate data of key points of the component, and the three-dimensional coordinates of the key points are calculated by spatial resection algorithm.

Images