Building engineering data processing method and system based on BIM

Abstract

The invention discloses a BIM-based building engineering data processing method and system, and belongs to the technical field of digital construction of building engineering. The method comprises the steps that construction scanning point cloud and BIM model data are acquired, and a feature skeleton is extracted; constructing a dynamic local coordinate system to establish a skeleton-model mapping relation; carrying out space division by adopting a locality sensitive hash index and executing ICP registration calculation; reversely deriving an error propagation path to generate a compensation instruction; adjusting a feature descriptor and reconstructing a mapping relation; and finally, model difference comparison and dynamic updating are realized. According to the method, automatic calibration of the construction point cloud and the BIM model is realized, and the problems of low efficiency and large error of traditional manual calibration are solved.

Description

Technical Field

[0001] This invention relates to the field of digital construction technology in building engineering, and in particular to a method and system for processing building engineering data based on BIM. Background Technology

[0002] During the construction process, discrepancies inevitably exist between the actual constructed structure and the BIM design model. Traditional methods, which rely primarily on manual measurement and comparison, suffer from the following technical limitations: Manual calibration is inefficient and struggles to process massive amounts of point cloud data, leading to delays in construction quality control. Existing automatic registration algorithms (such as traditional ICP) cannot adapt to the complex geometric features of building structures, and the matching accuracy is insufficient in key parts such as beam-column joints; The lack of an effective error propagation analysis mechanism makes it difficult to detect and block the cumulative effects of construction errors in a timely manner. When compensation and correction schemes are disconnected from construction techniques, the resulting BIM correction models often lack constructability.

[0003] The aforementioned problems are particularly prominent in complex buildings such as long-span steel structures and irregularly shaped curtain walls, severely restricting the effectiveness of BIM technology in construction quality control. Although automatic calibration methods based on point cloud registration have emerged in recent years, significant technical bottlenecks still exist in handling multi-scale building features and dynamic changes in the construction environment. Summary of the Invention

[0004] This invention provides a BIM-based building engineering data processing method and system to solve the problem of how to establish a dynamic mapping relationship between construction scan point cloud and BIM design model, realize automatic detection, propagation analysis and sub-millimeter accuracy compensation of construction errors, thereby solving the problems of low efficiency and error-proneness of manual calibration in traditional construction processes.

[0005] To address the aforementioned technical problems, this invention provides a BIM-based method for processing building engineering data, comprising: Acquire construction scan point cloud data and BIM model data, extract feature skeletons and analyze key points to generate BIM key point cloud data; Construct a dynamic local coordinate system and perform spatial alignment and weight coefficient adjustment to establish a skeleton-model mapping relationship; Locality-sensitive hashing index is used to spatially partition the feature skeleton data, generate an accelerated retrieval structure, perform ICP registration calculation, and output the coordinate offset matrix. Based on the reverse derivation of the coordinate offset matrix according to the construction sequence, error propagation modeling and compensation parameter calculation are performed to generate error compensation instructions and correction control instructions. Adjust the hierarchical feature descriptors of the feature skeleton data according to the error compensation instructions, perform bidirectional topology optimization processing, and re-establish the optimized skeleton-model mapping relationship. Based on the optimized skeleton-model mapping relationship, the system performs difference comparison and deviation warning, implements dynamic updates and quality monitoring of the BIM model, and outputs corrected BIM model data.

[0006] Furthermore, the generation of BIM key point cloud data includes: Acquire construction scanning point cloud data, perform noise filtering and topology analysis to obtain denoised point cloud data; Multi-scale feature skeletons are extracted from denoised point cloud data, and hierarchical feature descriptor calculations are performed to obtain feature skeleton data. Acquire BIM model data, analyze geometric key points and topological relationships, and generate BIM key point cloud data.

[0007] Furthermore, establishing the skeleton-model mapping relationship includes: A dynamic local coordinate system is constructed based on construction sequence progress and feature skeleton data. Spatially align the BIM key point cloud data with the dynamic local coordinate system to obtain the initial matching relationship; Adjust the weight coefficients of the feature skeleton data based on the initial matching relationship to establish the skeleton-model mapping relationship.

[0008] Further, the step of performing ICP registration calculation and outputting a coordinate offset matrix includes: Locality-sensitive hashing index is used to spatially partition the feature skeleton data, generating an accelerated retrieval structure; Based on the skeleton-model mapping relationship and the accelerated retrieval structure, GPU-accelerated ICP registration calculation is performed to generate a parallelized adaptive ICP architecture. Based on a parallelized adaptive ICP architecture, the coordinate offset matrix between the construction scan point cloud and the BIM model is output.

[0009] Furthermore, the generative parallelized adaptive ICP architecture includes: Spatial partitioning mapping of feature skeleton data is established using local sensitive hash index; The GPU parallel computing architecture is used to perform ICP iterative computation on each spatial partition; The calculation results from each partition are merged to generate global coordinate transformation parameters.

[0010] Furthermore, the generation of error compensation instructions and correction control instructions includes: Based on the reverse derivation of the coordinate offset matrix according to the construction process, an error propagation model is established; Compensation parameters are calculated based on the error propagation model, and error compensation instructions are generated. Error compensation instructions are fed back to the feature skeleton data and BIM key point cloud data to generate correction control instructions.

[0011] Furthermore, the re-establishment of the optimized skeleton-model mapping relationship includes: Adjust the hierarchical feature descriptors of the feature skeleton data according to the error compensation instructions; Synchronously update the topological connection relationships of BIM key point cloud data; Re-establish the optimized skeleton-model mapping relationship.

[0012] Furthermore, the output corrected BIM model data includes: Based on the optimized skeleton-model mapping relationship, the geometric data of the BIM model is reconstructed; By comparing the differences between the model before and after correction, a quality deviation warning signal is triggered. Implement dynamic updates and quality monitoring of the BIM model, and output corrected BIM model data.

[0013] Furthermore, the comparison of the model differences before and after correction, triggering a quality deviation warning signal, includes: Calculate the geometric deviation between the corrected model and the design model; A level three warning signal is generated when the deviation exceeds the threshold. The early warning threshold parameters are dynamically adjusted according to the construction stage.

[0014] A BIM-based building engineering data processing system, applied to any of the methods described above, comprising: The data preprocessing module is used to acquire construction scan point cloud data and BIM model data, and generate feature skeleton data and BIM key point cloud data. The dynamic matching module is used to construct a dynamic local coordinate system and establish a skeleton-model mapping relationship; The parallel registration module is used to perform ICP registration calculations based on locality-sensitive hash indexes; The error compensation module is used to generate error compensation instructions and correction control instructions; The topology optimization module is used to re-establish the optimized skeleton-model mapping relationship; The quality monitoring module is used to output corrected BIM model data and trigger deviation warnings.

[0015] The key innovations of this invention include: (1) The feature skeleton point cloud is modeled as a viscous fluid, and key feature regions are identified by eddy field.

[0016] (2) Dynamic optimization of metric tensors based on Ricci flow equation.

[0017] (3) The spatial propagation law of error is quantified by the path integral of the connection coefficient.

[0018] The following are its main beneficial effects: (1) The dynamic mapping relationship between the construction point cloud and the BIM model was realized. Compared with the traditional ICP registration algorithm, this method further improves the matching accuracy of key nodes (such as steel structure connections), further controls the overall model calibration error, and meets the precision construction requirements of super high-rise buildings, large-span steel structures, etc.

[0019] (2) An error propagation model based on the Ricci flow equation can dynamically track the transmission path of construction deviations in the building structure. By quantifying the cumulative effect of errors, the error source can be quickly located and compensated, thereby reducing the cumulative amount of construction errors and significantly reducing the rework rate.

[0020] (3) Implement three-level feature extraction of 5mm / 2mm / 1mm (S120 module), and combine it with local sensitive hash space partitioning (S310 module) to enable the system to process the overall outline of the building and detailed features such as bolt hole positions at the same time. Attached Figure Description

[0021] Figure 1 A flowchart illustrating the BIM-based building engineering data processing method provided in this application embodiment; Figure 2 The structural block diagram of the BIM-based building engineering data processing system provided in the embodiments of this application is shown. Detailed Implementation

[0022] Example 1: Refer to Figure 1 This is a flowchart illustrating a BIM-based building engineering data processing method provided in an embodiment of the present invention. The process may include at least steps S100-S600: S100: Acquire construction scan point cloud data and BIM model data, extract feature skeleton and analyze key points to generate BIM key point cloud data. S200. Construct a dynamic local coordinate system and perform spatial alignment and weight coefficient adjustment to establish a skeleton-model mapping relationship; S300: Use Local Sensitive Hash Index to spatially partition the feature skeleton data, generate an accelerated retrieval structure, perform ICP registration calculation, and output the coordinate offset matrix; S400. Based on the reverse derivation of the coordinate offset matrix according to the construction procedure, perform error propagation modeling and compensation parameter calculation, and generate error compensation instructions and correction control instructions. S500: Adjust the hierarchical feature descriptors of the feature skeleton data according to the error compensation instruction, perform bidirectional topology optimization processing, and re-establish the optimized skeleton-model mapping relationship. S600, based on the optimized skeleton-model mapping relationship, performs difference comparison and deviation warning, implements dynamic updating and quality monitoring of BIM model, and outputs corrected BIM model data.

[0023] Step S100 includes at least steps S110-S130: S110. Acquire construction scan point cloud data, perform noise filtering and topology analysis to obtain denoised point cloud data.

[0024] Specifically, raw point cloud data of the construction site is acquired using a terrestrial 3D laser scanner. This raw point cloud data includes 3D coordinate information, reflection intensity information, and RGB color information. Understandably, the scanner employs a pulse ranging principle, achieving a single-point measurement accuracy of ±2mm, with a scanning density set to 5mm@10m. Furthermore, the raw point cloud data is stored in partitions according to the construction area, with each partition's point cloud data appended with a timestamp and construction stage identifier.

[0025] The original point cloud data is processed using a noise filtering algorithm based on statistical outlier removal. Further, the average distance between each point and its 50 neighboring points is calculated; points with an average distance exceeding 1.5 times the global distance threshold are identified as noise points. Understandably, the global distance threshold is determined by the 85th percentile of the point cloud density distribution histogram. Further, the retained point cloud data is subjected to Gaussian smoothing filtering, with the kernel radius set to 3 times the average point cloud spacing.

[0026] Delaunay triangulation is performed on the filtered point cloud data to establish topological connections between the point clouds. Further, the side lengths of the triangular meshes are extracted as feature parameters; edges exceeding three times the average side length are identified as topologically anomalous edges. Understandably, local mesh reconstruction is performed on these anomalous edges to generate denoised point cloud data with a continuous topological structure.

[0027] S120. Extract multi-scale feature skeletons from denoised point cloud data, perform hierarchical feature descriptor calculations, and obtain feature skeleton data.

[0028] The denoised point cloud data is processed using a multi-scale feature extraction method based on the α-shape algorithm. Furthermore, three feature scale levels are set: First level: α=5mm, extract the overall outline skeleton of building components; Second level: α=2mm, extract beam-column node connection features; Third level: α=1mm, extract detailed features such as bolt hole positions; Understandably, each level of feature skeleton point set is appended with a scale identifier and confidence weights.

[0029] Specifically, the following descriptors are calculated for each feature skeleton point: normal vector feature: principal component direction based on neighborhood point PCA analysis; curvature feature: local curvature value calculated by fitting a quadratic surface; density feature: ratio of the number of adjacent points to the average spacing within a unit sphere. Further, the descriptors are normalized according to scale levels to generate a 128-dimensional composite feature description vector.

[0030] Specifically, the multi-scale feature skeleton point set is associated and stored with the corresponding feature descriptors. Understandably, the feature skeleton data structure includes: three-dimensional coordinate information; scale level identifier; 128-dimensional feature description vector; and topological connectivity index.

[0031] S130. Obtain BIM model data, analyze geometric key points and topological relationships, and generate BIM key point cloud data.

[0032] Geometric entity information is extracted from the BIM design model in IFC format. Further, the boundary representation data of the geometric entities is parsed, including vertex coordinate information, edge topological relationships, and surface parametric equations. Understandably, the geometric entities are classified and labeled according to construction component types.

[0033] Key points are extracted based on a geometric feature saliency detection algorithm: endpoints and midpoints are extracted for straight lines, start and end points and quadrant points are extracted for arcs, and Gaussian curvature extrema are extracted for curved surfaces. Furthermore, a component type identifier and design tolerance value are added to each key point.

[0034] The component connection relationships are reconstructed based on the Ifc RelConnects Path Elements entities in the IFC model. Further, a topological connection graph between key points is established, with edge weights determined by the following factors: physical connection type (welded / bolted connection); design load transfer path; construction sequence dependencies. Specifically, the parsed geometric key points and topological relationships are integrated and stored. Understandably, the BIM key point cloud data structure includes: 3D coordinates of key points, component type identifier, design tolerance values, and a topological connection weight matrix.

[0035] Step S200 includes at least steps S210-S230: S210. Based on the construction process progress and feature skeleton data, construct a dynamic local coordinate system.

[0036] Specifically, the process progress data of the current construction stage is obtained from the construction management system. The process progress data includes: a list of completed component installations, the design positioning coordinates of components under construction, and a table of dependencies on subsequent processes.

[0037] Furthermore, based on the process progress data, the spatial influence range of the current construction area is determined, and a three-dimensional boundary box of the construction influence area is generated.

[0038] Spatial filtering is performed on the feature skeleton data generated in S120 based on the three-dimensional bounding box of the construction impact area. Understandably, the filtering conditions include: the skeleton point coordinates are located within the bounding box, the scale level identifier matches the current construction accuracy requirements, and the confidence weight of the feature description vector is ≥0.7.

[0039] Furthermore, the selected feature skeleton points are grouped according to component type, and each group of skeleton points is assigned a construction procedure identifier.

[0040] Specifically, the following parameters are calculated for the feature skeleton point set for each component type: principal direction vector: the direction of the first principal component of the point set is determined by PCA analysis; reference plane: the best plane is fitted using the least squares method; local origin: the weighted average of the geometric center of the point set and the design positioning coordinates; further, the weight coefficients of the parameters are adjusted according to the construction procedure identifier to generate a component-level local coordinate system.

[0041] Multiple component-level local coordinate systems within the same construction area are integrated: the priority of the coordinate systems is determined by the construction load transfer path; the principal directions of each coordinate system are aligned using quaternion interpolation; and coordinate transformation is performed with the highest priority coordinate system as the reference. Understandably, the final generated dynamic local coordinate system includes: origin coordinates; three axial unit vectors; and a coordinate system confidence score.

[0042] S220. Spatially align the BIM key point cloud data with the dynamic local coordinate system to obtain the initial matching relationship.

[0043] Specifically, based on the 3D boundary box of the construction impact area determined in S210, matching point sets are selected from the BIM key point cloud data generated in S130. Further, the selection criteria include: key point coordinates are within the boundary box, component type identifier matches the current construction stage, and design tolerance value is ≤3mm.

[0044] Understandably, the selected key points are grouped by component type, and each group of key points is accompanied by a design positioning identifier.

[0045] For each component type's BIM key point set, perform the following operations: transform the design coordinates to the dynamic local coordinate system; calculate the average positional offset of the key points before and after the transformation; and solve for the optimal rotation matrix using singular value decomposition (SVD).

[0046] Furthermore, the rotation matrix is ​​smoothed according to the component connection relationship to generate a component-level coordinate transformation matrix.

[0047] Match each feature skeleton point with BIM key points: calculate the Euclidean distance in the dynamic local coordinate system; establish candidate matching pairs when the distance is ≤ 2 times the average spacing of the point cloud; verify the reliability of the matching by using the cosine similarity of the feature description vector.

[0048] Furthermore, the generated initial matching relationships include: a correspondence table between skeleton points and key points; a confidence score for each matching pair; and anomaly markers for unmatched points.

[0049] S230. Adjust the weight coefficients of the feature skeleton data according to the initial matching relationship, and establish the skeleton-model mapping relationship.

[0050] Specifically, the feature skeleton data is processed based on the initial matching relationship generated in S220 as follows: For successfully matched skeleton points: increase their confidence weight by 20%; For unmatched skeleton points: reduce their weight or remove them based on anomaly markers; For low-confidence matching pairs: manual verification and labeling are performed; Furthermore, the weight adjustment range is dynamically adjusted according to the progress of the construction process.

[0051] Using the BIM key point topology connection weight matrix generated by S130, the topological rationality of feature skeleton points is verified: the consistency between the skeleton point connection relationship and the design topology is calculated; topology correction is triggered when the connection direction deviation is >5°; and the positions of abnormal skeleton points are adjusted using a virtual force field algorithm. Understandably, the corrected feature skeleton data requires recalculation of the hierarchical feature descriptors.

[0052] The integrated data generates the final mapping relationship: a two-way index table of skeleton points and BIM key points is established; coordinate transformation parameters are recorded for each mapping pair; and a mapping relationship matrix containing confidence weights is generated.

[0053] Furthermore, the data structure of the mapping relationship matrix includes: the mapping relationship between skeleton point IDs and key point IDs; local coordinate transformation parameters (translation vector + rotation matrix); and mapping confidence weights (range 0-1).

[0054] Step S300 includes at least steps S310-S330: S310. Use locality-sensitive hash index to spatially partition the feature skeleton data to generate an accelerated retrieval structure; Construct the Navier-Stokes equations, specifically by using the feature skeleton data output from S230. (in Modeling the fluid as a three-dimensional incompressible viscous fluid (considering the total number of skeleton points), and establishing the governing equations:

[0055] in: : The velocity field of the feature skeleton points in three-dimensional space; Time variable; Mapping confidence weights output by S230 Defined density field; Fluid pressure field, used to balance inertial forces; Dynamic viscosity coefficient, reflecting feature similarity; Surface tension, generated by S130 topological weights; Gradient operator; : Laplace operator; Furthermore, vortex feature extraction is performed to solve for the vortex field. Identify feature regions:

[0056] in: : Represents the three-dimensional coordinate vector of the j-th feature point in the feature skeleton dataset; j: is the point index variable; : Vortex field, reflecting the local rotation intensity; : The permutation tensor used for cross product calculation; : No. A total of 10 high-vorticity characteristic regions (in total) indivual); : Vortex intensity threshold (usually set to 0.3); Furthermore, we construct the Riemannian metric: in each Define a metric tensor above:

[0057] in: Subregion The metric tensor; , , Mean curvature along the three principal directions; Scale adjustment factor Technical effect: Adaptive spatial partitioning of feature skeleton is achieved through fluid dynamics simulation, which further improves the accuracy of region segmentation (compared to the traditional LSH method).

[0058] The Navier-Stokes equations model the motion of feature skeleton points as a viscous fluid, achieving dynamic spatial partitioning through pressure terms and surface tension.

[0059] Vortex field: Identifies high curvature regions (such as beam-column joints) in the feature skeleton, providing key points for subsequent registration.

[0060] Riemannian metric: Constructing a local coordinate system through curvature weighting, allowing ICP registration to adapt to the geometric characteristics of the building structure.

[0061] S320, based on skeleton-model mapping relationship and accelerated retrieval structure, performs GPU-accelerated ICP registration calculation and generates parallelized adaptive ICP architecture.

[0062] After obtaining the spatial partitioning results, the system invokes GPU parallel computing resources to execute an improved ICP registration algorithm. First, based on the initial matching relationship generated by the S220 module, the system loads the corresponding BIM keypoint dataset for each spatial partition. During registration, the system employs a two-stage optimization strategy: in the first stage, coarse matching is performed based on the feature description vectors stored in the accelerated retrieval structure to quickly establish candidate correspondences between feature skeleton points and BIM keypoints; in the second stage, fine matching is performed using the geometric features within the partition, optimizing rigid body transformation parameters through an iterative nearest-point algorithm. To improve computational efficiency, the system designs an adaptive registration termination condition, automatically terminating the calculation when the error change over three consecutive iterations is less than a dynamic threshold. For large-scale building projects, the system adopts a block-based pipeline processing mechanism, where the CPU prepares data for the next partition while the GPU calculates the registration result for the current partition, ensuring full utilization of computing resources. After completing the registration calculation for all partitions, the system performs weighted fusion based on the confidence weights of each partition to generate globally consistent coordinate transformation parameters, constructing a complete parallelized adaptive ICP architecture.

[0063] S330, based on a parallelized adaptive ICP architecture, outputs the coordinate offset matrix between the construction scan point cloud and the BIM model.

[0064] The system integrates the registration results from various spatial partitions to construct a complete coordinate offset matrix. First, it performs boundary consistency checks on the registration results of adjacent partitions to ensure smooth connections in transition areas. For conflicting boundary areas, the system arbitrates based on the confidence weights in the skeleton-model mapping relationship, prioritizing the calculation results from the high-confidence partition. Subsequently, the system transforms the local coordinate transformation parameters to the global construction coordinate system, generating a complete offset matrix containing translation vectors and rotation matrices. This matrix not only records the displacement of each feature point but also includes key parameters required for error propagation path analysis, such as material deformation coefficients and connection node stiffness. The final output coordinate offset matrix is ​​organized according to construction areas, with each area unit appended with a timestamp and version identifier, supporting incremental updates for subsequent processes. The system also generates a registration quality report, marking low-confidence areas and key nodes requiring manual review, providing a basis for construction quality control decisions.

[0065] Step S400 includes at least steps S410-S430: S410. Based on the reverse derivation of the coordinate offset matrix according to the construction process, establish an error propagation model.

[0066] The system first obtains complete process data records from the construction management platform, including the component installation sequence, the completion time of each process, and the design change history. Using a timestamp parsing engine, this data is sorted in reverse order by construction stage, establishing a time chain tracing back from the current construction node to the initial process.

[0067] Based on the coordinate offset matrix output by S330, the system creates a three-dimensional spatial error distribution map. Through structural mechanics analysis algorithms, the main load transfer paths are identified. Combined with the component connection relationship data provided by S130, key force transmission locations such as steel structure nodes and concrete joints are marked, forming a weighted error propagation network.

[0068] The system employs the finite element method to divide the construction area into a three-dimensional mesh with millimeter-level precision. Within each mesh cell, the cumulative error intensity is calculated based on the error propagation coefficient of adjacent components and material properties. Through iterative calculations, a visualized error intensity distribution field is ultimately generated, accurately reflecting the propagation path and impact range of errors within the building structure.

[0069] S420: Calculate compensation parameters based on the error propagation model and generate error compensation instructions.

[0070] The system calculates the optimal compensation direction for each feature skeleton point in the error field. Using a gradient descent algorithm, it finds the adjustment vector that can maximally offset the impact of the error. Simultaneously, construction process limitations are considered to ensure that the compensation amount is within the allowable adjustment range of the component.

[0071] Depending on the component type and construction stage, the system adopts differentiated compensation strategies: for the main steel structure, rigid displacement compensation is used to maintain the overall structural stability; for the curtain wall units, flexible deformation compensation is used to adapt to local deformation requirements; and for the pipeline system, segmented compensation is used to ensure system functionality.

[0072] The calculated compensation parameters are grouped according to the construction area to generate a structured instruction set. Each instruction unit contains complete parameters such as the target component ID, three-dimensional compensation vector, execution priority, and process requirements.

[0073] S430: Feedback the error compensation command to the feature skeleton data and BIM key point cloud data to generate correction control command.

[0074] The system establishes a dynamic data channel to transmit compensation commands to the feature skeleton dataset in real time. An incremental update algorithm is employed to adjust the coordinates of skeleton points point by point without affecting the overall model stability. Simultaneously, automatic updates of feature descriptors are triggered to ensure the consistency of geometric features.

[0075] The system transmits the compensation amount to the BIM model via the IFC data interface. A parametric adjustment method is used to synchronously update the geometric data and topological relationships of relevant components. During the correction process, clash detection and process feasibility verification are continuously performed to ensure that the corrected model conforms to construction specifications.

[0076] The system establishes a correction effect evaluation mechanism, verifying the compensation effect by comparing scan data before and after correction. A secondary compensation process is initiated for areas that fail to meet standards until the error is controlled within acceptable limits. All correction records are automatically archived, forming a complete quality traceability chain.

[0077] Step 500 includes at least steps S510-S530: S510. Adjust the hierarchical feature descriptor of the feature skeleton data according to the error compensation instruction.

[0078] The system first receives an error compensation instruction set from S430, which includes a three-dimensional compensation vector and adjustment weights for each feature skeleton point. Through spatial index lookup, the compensation instructions are precisely matched to the multi-scale feature skeleton data generated by S120. Specifically, the system performs hierarchical processing of the compensation amount according to the scale level identifier (5mm / 2mm / 1mm) of the skeleton points, ensuring that feature points at different precision levels receive appropriate adjustment strategies.

[0079] For each feature skeleton point, the system performs the following processing steps: Coordinate update: Adjust the spatial position of the skeleton point based on the compensation vector, and use a smooth transition algorithm to avoid abrupt changes; Normal vector recalculation: Perform PCA analysis again in the updated neighborhood point set to generate new normal vector features; Curvature feature correction: Calculate the adjusted Gaussian curvature and average curvature through local surface fitting; Density feature calibration: Recalculate the point density within the unit sphere according to the spatial distribution after displacement.

[0080] The updated geometric features are normalized according to scale levels: Level 1 (5mm): retain the overall outline features and weaken the details; Level 2 (2mm): balance macro and micro features; Level 3 (1mm): strengthen the details.

[0081] Finally, a 128-dimensional composite feature description vector with compensatory adaptation is generated.

[0082] S520: Synchronously update the topological connection relationship of BIM key point cloud data; Based on the compensated skeleton point distribution, the system recalculates the topological connection relationship between BIM key points: Physical connection analysis: Based on the component type identification of S130, the adaptability of welding / bolted connections is verified; Load path optimization: Based on the compensated geometry, the load transfer efficiency coefficient is recalculated; Process dependency update: Combined with the construction progress, the temporal dependency relationship between key points is adjusted.

[0083] A graph-based topological relationship model is constructed: Nodes: compensated BIM key points; Edges: updated physical connections; Edge weights: composite weights that integrate load transfer, construction procedures, and geometric matching degree; Through graph traversal algorithms, key transmission paths and vulnerable nodes are identified, providing a basis for establishing subsequent mapping relationships.

[0084] S530. Re-establish the optimized skeleton-model mapping relationship.

[0085] The system executes an improved matching algorithm: Forward matching: finding the optimal BIM key points from the compensated feature skeleton points; Reverse verification: verifying the rationality of skeleton point matching from BIM key points; Conflict resolution: manually reviewing or recalculating inconsistent matching pairs; Constructing a mapping relationship matrix containing the following elements: Spatial transformation parameters: a rigid body transformation matrix that integrates compensation amounts; Confidence assessment: a composite score based on feature similarity and topological consistency; Error tolerance: the allowable deviation range set according to component type and construction stage.

[0086] Establish a real-time verification mechanism: Geometric consistency: check whether the mapped model conforms to the design specifications; Topological consistency: verify whether the connection relationship meets the structural mechanics requirements; Technological consistency: ensure that the modified scheme is feasible for construction.

[0087] Step S600 includes at least steps S610-S630: S610. Based on the optimized skeleton-model mapping relationship, reconstruct the geometric data of the BIM model.

[0088] The system first loads the optimized skeleton-model mapping matrix generated by S530, which fully records the precise correspondence between feature skeleton points and BIM key points. During the parsing process, the system categorizes and processes the mapping data according to component type, extracting the corresponding feature skeleton point subset and BIM key point subset for each building component (including structural elements such as steel beams and concrete columns). Through a specially developed spatial transformation engine, the system converts the abstract skeleton-model mapping relationship into executable coordinate transformation parameters for each specific component. In the geometric reconstruction phase, the system adopts differentiated processing strategies for different types of building components according to the data structure requirements of the IFC standard: for linear components such as beams and columns, the system focuses on adjusting the coordinates of axis control points and end connections; for curved components, including complex geometries such as curtain walls and roofs, the vertex positions of the control grid are recalculated; and for equipment pipeline systems, the pipeline routing and connection point positions are precisely updated according to the results of segmented compensation. Throughout the reconstruction process, the system continuously verifies the validity of geometric data to ensure that component length variations are controlled within the allowable range of materials, angle deviations meet construction specifications, and, in particular, the matching accuracy of connection parts must strictly meet design requirements. At the same time, the system also maintains the non-geometric attribute data of the BIM model, including adjusting the material stress state markers according to the displacement, updating the component installation progress markers, and recording detailed quality information such as the time and operator of each correction, forming a complete modification traceability chain.

[0089] S620: Compare the differences between the model before and after correction, and trigger a quality deviation warning signal.

[0090] The system establishes a multi-layered, comprehensive difference detection mechanism to ensure the accuracy of model correction. In terms of geometric difference detection, the system performs point-to-point distance analysis to compare the positional offset of key points before and after correction, conducts shape change analysis to calculate the deformation rate of component sections, and verifies the relative positional relationships between components through spatial relationship analysis. At the topology level, the system focuses on verifying connection relationships, checking the effectiveness of welding and bolted connections, and analyzing changes in load transfer paths. Process detection emphasizes assessing construction feasibility, ensuring that the corrected model meets actual construction process requirements without negatively impacting subsequent procedures. Based on these precise analyses, the system implements a three-level early warning mechanism: a level one warning is triggered when a local deviation is detected within the range of 1-3mm, and the system automatically records the deviation without interrupting the construction process; a level two warning is activated when the deviation of a critical part exceeds 3mm or affects structural performance, notifying technical personnel for on-site verification; if a serious deviation occurs that may lead to structural safety hazards or make the process infeasible, a level three warning is immediately triggered, suspending the relevant procedures and initiating emergency response procedures. It is worth noting that these warning thresholds are not fixed, but are dynamically adjusted according to factors such as construction stage, component importance and material properties. For example, stricter standards are adopted during the main structure construction stage, higher precision requirements are set for major load-bearing components, and the differences in allowable deformation values ​​of different building materials are fully considered.

[0091] S630. Implement dynamic updates and quality monitoring of the BIM model, and output the corrected BIM model data.

[0092] In terms of model version management, the system has established a comprehensive version control mechanism. Each revision is automatically recorded with a precise timestamp, generating a detailed change log that records the modified content and responsible personnel information. It also supports historical version query and recovery functions, providing complete process traceability capabilities for project management. The quality monitoring system forms a complete closed loop, acquiring on-site point cloud data through regular laser scanning, digitizing construction records, and automatically comparing on-site data with the revised model, performing deviation trend analysis and prediction. When anomalies are detected, the system automatically triggers a review process, dynamically optimizing the model based on monitoring results. In the final output stage, the system performs rigorous data integrity verification and specification compliance checks, automatically generating delivery documents and supporting output in multiple formats such as IFC, DWG, and PDF to meet the needs of different application scenarios. The entire implementation process fully demonstrates the application value of digital technology in construction quality management. Through continuous monitoring and feedback optimization, it ensures that the BIM model and the actual building remain highly consistent, providing the project with an accurate and reliable digital twin model, effectively improving construction quality and management efficiency.

[0093] Example 2: Figure 2A structural block diagram of a BIM-based building engineering data processing system according to an embodiment of the present invention is shown. Figure 2 As shown, the structure may include: The data preprocessing module 10 is configured to acquire raw point cloud data of the construction site using a terrestrial 3D laser scanner, and perform noise filtering and topological analysis on the raw point cloud data to generate denoised point cloud data. Specifically, this module executes a multi-scale feature skeleton extraction algorithm to identify the overall outline of building components, beam-column node connection features, and detailed features such as bolt hole positions from the denoised point cloud data, generating feature skeleton data with hierarchical feature descriptors. Simultaneously, this module parses geometric key points and topological relationships from the IFC format BIM design model, constructing BIM key point cloud data that includes component type identifiers, design tolerance values, and topological connection weight matrices.

[0094] The dynamic matching module 20 constructs a dynamic local coordinate system based on construction sequence progress and feature skeleton data. This module first filters feature skeleton data according to the 3D boundary box of the construction influence area, determines the principal direction vector of each component point set through PCA analysis, and aligns multiple component-level local coordinate systems using quaternion interpolation. Further, this module spatially aligns the BIM key point cloud data with the dynamic local coordinate system, solves for the optimal rotation matrix through singular value decomposition, and establishes a skeleton-model mapping relationship matrix including confidence weights.

[0095] The parallel registration module 30 uses a locality-sensitive hash index to spatially partition the feature skeleton data, generating a multi-resolution accelerated retrieval structure. This module performs improved ICP registration calculations based on a GPU-accelerated architecture, achieving high-precision matching between the feature skeleton data and the BIM key point cloud through a curvature-driven optimization algorithm. During the registration process, the module dynamically adjusts the weight coefficients of each spatial partition, ultimately outputting a coordinate offset matrix between the construction scan point cloud and the BIM model.

[0096] The error compensation module 40 establishes an error propagation model by deriving the propagation path of the coordinate offset matrix in reverse order of the construction process. This module calculates the three-dimensional spatial error distribution using finite element analysis and employs differentiated compensation strategies for different component types (steel structures, curtain walls, pipelines, etc.), generating a structured compensation instruction set containing the target component ID, three-dimensional compensation vector, and execution priority. The compensation parameter calculation process considers constraints such as material properties, connection methods, and construction techniques.

[0097] The topology optimization module 50 adjusts the hierarchical feature descriptors of the feature skeleton data according to error compensation instructions, including recalculating normal vector features, curvature features, and density features. This module synchronously updates the topological connectivity of the BIM key point cloud data, identifies key transmission paths through a graph traversal algorithm, and establishes an optimized skeleton-model mapping relationship. During the mapping relationship reconstruction process, the module implements a triple verification mechanism for geometric consistency, topological consistency, and process consistency.

[0098] The quality monitoring module 60 reconstructs the BIM model's geometric data based on the optimized skeleton-model mapping relationship and performs multi-dimensional difference detection and analysis. This module sets dynamically adjustable three-level early warning thresholds, triggering corresponding early warning signals when geometric deviations, topological anomalies, or process conflicts are detected. The module also maintains a complete version control record, including timestamps, change logs, and version rollback functionality, ultimately outputting corrected BIM model data conforming to IFC standards.

[0099] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A BIM-based method for processing building engineering data, characterized in that, include: Acquire construction scan point cloud data and BIM model data, extract feature skeletons and analyze key points to generate BIM key point cloud data; Construct a dynamic local coordinate system and perform spatial alignment and weight coefficient adjustment to establish a skeleton-model mapping relationship; Locality-sensitive hash index is used to spatially partition the feature skeleton data, generate an accelerated retrieval structure, perform ICP registration calculation, and output the coordinate offset matrix. Based on the reverse derivation of the coordinate offset matrix according to the construction sequence, error propagation modeling and compensation parameter calculation are performed to generate error compensation instructions and correction control instructions. Adjust the hierarchical feature descriptors of the feature skeleton data according to the error compensation instruction, perform bidirectional topology optimization processing, and re-establish the optimized skeleton-model mapping relationship. Based on the optimized skeleton-model mapping relationship, the system performs difference comparison and deviation warning, implements dynamic updates and quality monitoring of the BIM model, and outputs corrected BIM model data.

2. The method according to claim 1, characterized in that, The generation of BIM key point cloud data includes: Acquire construction scanning point cloud data, perform noise filtering and topology analysis to obtain denoised point cloud data; Multi-scale feature skeletons are extracted from denoised point cloud data, and hierarchical feature descriptor calculations are performed to obtain feature skeleton data. Acquire BIM model data, analyze geometric key points and topological relationships, and generate BIM key point cloud data.

3. The method according to claim 1, characterized in that, The establishment of the skeleton-model mapping relationship includes: A dynamic local coordinate system is constructed based on construction sequence progress and feature skeleton data. Spatially align the BIM key point cloud data with the dynamic local coordinate system to obtain the initial matching relationship; Adjust the weight coefficients of the feature skeleton data based on the initial matching relationship to establish the skeleton-model mapping relationship.

4. The method according to claim 1, characterized in that, The ICP registration calculation is performed, and the coordinate offset matrix is ​​output, including: Locality-sensitive hashing index is used to spatially partition the feature skeleton data, generating an accelerated retrieval structure; Based on the skeleton-model mapping relationship and the accelerated retrieval structure, GPU-accelerated ICP registration calculation is performed to generate a parallelized adaptive ICP architecture. Based on a parallelized adaptive ICP architecture, the coordinate offset matrix between the construction scan point cloud and the BIM model is output.

5. The method according to claim 4, characterized in that, The generated parallelized adaptive ICP architecture includes: Spatial partitioning mapping of feature skeleton data is established using local sensitive hash index; The GPU parallel computing architecture is used to perform ICP iterative computation on each spatial partition; The calculation results from each partition are merged to generate global coordinate transformation parameters.

6. The method according to claim 1, characterized in that, The generation of error compensation instructions and correction control instructions includes: Based on the reverse derivation of the coordinate offset matrix according to the construction process, an error propagation model is established; Compensation parameters are calculated based on the error propagation model, and error compensation instructions are generated. Error compensation instructions are fed back to the feature skeleton data and BIM key point cloud data to generate correction control instructions.

7. The method according to claim 1, characterized in that, The process of re-establishing the optimized skeleton-model mapping relationship includes: Adjust the hierarchical feature descriptors of the feature skeleton data according to the error compensation instructions; Synchronously update the topological connection relationships of BIM key point cloud data; Re-establish the optimized skeleton-model mapping relationship.

8. The method according to claim 1, characterized in that, The output corrected BIM model data includes: Based on the optimized skeleton-model mapping relationship, the geometric data of the BIM model is reconstructed; By comparing the differences between the model before and after correction, a quality deviation warning signal is triggered. Implement dynamic updates and quality monitoring of the BIM model, and output corrected BIM model data.

9. The method according to claim 8, characterized in that, The comparison of the model before and after correction triggers a quality deviation warning signal, including: Calculate the geometric deviation between the corrected model and the design model; A level three warning signal is generated when the deviation exceeds the threshold. The early warning threshold parameters are dynamically adjusted according to the construction stage.

10. A BIM-based building engineering data processing system, applied to the method according to any one of claims 1-9, characterized in that, include: The data preprocessing module is used to acquire construction scan point cloud data and BIM model data, and generate feature skeleton data and BIM key point cloud data. The dynamic matching module is used to construct a dynamic local coordinate system and establish a skeleton-model mapping relationship; The parallel registration module is used to perform ICP registration calculations based on locality-sensitive hash indexes; The error compensation module is used to generate error compensation instructions and correction control instructions; The topology optimization module is used to re-establish the optimized skeleton-model mapping relationship; The quality monitoring module is used to output corrected BIM model data and trigger deviation warnings.

Images