Construction site bim modeling method and system based on three-dimensional laser scanning

Abstract

The application discloses a construction site BIM modeling method and system based on three-dimensional laser scanning, and relates to the technical field of three-dimensional modeling.The method comprises the following steps: in a construction site, three-dimensional laser scanning is performed to obtain three-dimensional laser data, laser data clustering is performed to obtain clustered laser data clusters; based on a preset construction structure, the clustered laser data clusters are iteratively fitted to obtain a fitting structure confidence set, and a matching construction structure is screened; according to construction progress information of the construction site, the fitting structure confidence set is combined to configure a low confidence proportion, and a low-confidence clustered laser data cluster is screened and obtained; the low-confidence clustered laser data cluster is combined with the nearest clustered laser data cluster respectively, preset occlusion combined structure fitting is performed, an occlusion structure confidence set is obtained, a matching occlusion combined structure is screened and obtained, BIM modeling is performed in combination with the matching construction structure, and a modeling result is obtained.The application effectively improves the BIM modeling precision of a construction site.

Description

Technical Field

[0001] This invention relates to the field of 3D modeling technology, specifically to a method and system for BIM modeling of construction sites based on 3D laser scanning. Background Technology

[0002] With the increasing application of Building Information Modeling (BIM) technology in construction progress, quality, and safety management, reverse modeling based on 3D laser scanning has become a key technology for obtaining the actual site conditions and comparing them with the design model. Existing methods typically collect site point cloud data, register and denoise it, and then directly perform geometric fitting or matching with a pre-set component library to achieve automated or semi-automated BIM model reconstruction.

[0003] However, the construction site environment is complex and ever-changing, and existing modeling methods have significant limitations. Due to dynamic interference on site and incomplete data acquisition, the generated 3D model often has deviations and omissions, making it difficult to accurately reflect the actual structural form. Ultimately, this results in insufficient accuracy of the BIM model, which cannot effectively support refined construction management. Summary of the Invention

[0004] This invention provides a method and system for BIM modeling of construction sites based on three-dimensional laser scanning, aiming to solve the technical problem of insufficient accuracy in existing BIM modeling technologies.

[0005] In view of the above problems, the present invention provides a method and system for BIM modeling of construction sites based on three-dimensional laser scanning.

[0006] In a first aspect, the present invention provides a construction site BIM modeling method based on three-dimensional laser scanning, including:

[0007] Within the construction site, three-dimensional laser scanning is performed to obtain three-dimensional laser data. Laser data clustering is then performed to obtain multiple clustered laser data clusters.

[0008] Based on various pre-set construction structures at the construction site, multiple clustered laser data clusters are iteratively fitted to obtain multiple sets of confidence scores for fitted structures, and multiple matching construction structures are selected.

[0009] Based on the construction progress information at the construction site, and combined with the low confidence ratio of multiple fitted structure confidence sets, multiple low confidence clustered laser data clusters were obtained.

[0010] Multiple low-confidence clustered laser data clusters are combined with the nearest clustered laser data cluster to fit various preset occlusion combination structures, resulting in multiple occlusion structure confidence sets. Multiple matching occlusion combination structures are then selected and combined with multiple matching construction structures to perform BIM modeling and obtain modeling results.

[0011] Secondly, the present invention provides a construction site BIM modeling system based on three-dimensional laser scanning, comprising:

[0012] The laser data clustering module is used to perform 3D laser scanning at the construction site to obtain 3D laser data, and then perform laser data clustering to obtain multiple clustered laser data clusters.

[0013] The structure fitting and matching module is used to iteratively fit multiple clustered laser data clusters based on various preset construction structures at the construction site, obtain multiple sets of confidence scores for fitted structures, and filter multiple matching construction structures.

[0014] The low-confidence screening module is used to filter and obtain multiple low-confidence clustered laser data clusters by combining the construction progress information at the construction site with the low-confidence ratio of multiple fitted structure confidence sets.

[0015] The occlusion modeling integration module is used to combine multiple low-confidence clustered laser data clusters with the nearest clustered laser data clusters to fit various preset occlusion combination structures, obtain multiple occlusion structure confidence sets, filter to obtain multiple matching occlusion combination structures, combine multiple matching construction structures, perform BIM modeling, and obtain modeling results.

[0016] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0017] This invention provides a method and system for BIM modeling of construction sites based on 3D laser scanning. By specifically addressing the problems of data loss caused by occlusion and modeling interference introduced by noise, and combining laser data clustering optimization, confidence screening of construction progress correlation, and occlusion combination structure fitting strategy, it effectively improves the accuracy of key feature extraction and point cloud registration accuracy, and finally achieves high-precision and high-reliability BIM modeling of construction sites, providing accurate data support for subsequent construction plan optimization, quality acceptance and other stages. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating the construction site BIM modeling method based on three-dimensional laser scanning provided in an embodiment of the present invention;

[0020] Figure 2A schematic diagram of the structure of a construction site BIM modeling system based on three-dimensional laser scanning provided in an embodiment of the present invention;

[0021] The components represented by each number in the attached diagram are explained below:

[0022] Laser data clustering module 11, structure fitting and matching module 12, low confidence screening module 13, occlusion modeling integration module 14. Detailed Implementation

[0023] This invention provides a construction site BIM modeling method and system based on three-dimensional laser scanning, which is used to address the technical problem of insufficient BIM modeling accuracy in existing technologies.

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0025] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0026] Example 1, as Figure 1 As shown, this invention provides a BIM modeling method for construction sites based on three-dimensional laser scanning, the method comprising:

[0027] S100: Within the construction site, perform 3D laser scanning to obtain 3D laser data, perform laser data clustering, and obtain multiple clustered laser data clusters.

[0028] In this embodiment of the invention, a three-dimensional laser scan is performed on the construction site to obtain three-dimensional laser data. Laser data clustering is then performed to obtain multiple clustered laser data clusters. The construction site environment is complex, and the raw point cloud data obtained from the three-dimensional laser scan contains a large number of discrete laser points. This includes valid data from target construction structures such as frame columns, beams, and wall panels, as well as interfering data from non-target objects such as construction equipment, safety nets, and temporary supports. Furthermore, all laser points are distributed randomly. Directly using this data for BIM modeling would lead to chaotic structural identification and low fitting accuracy. Therefore, it is necessary to first collect complete spatial data through three-dimensional laser scanning, and then use a clustering algorithm to group the laser points according to spatial distance characteristics. Laser points belonging to the same construction structure are aggregated into data clusters, achieving preliminary separation of valid data from interfering data and providing regular and clean basic data for subsequent structural fitting.

[0029] Step S100 in the method provided in this embodiment of the invention includes:

[0030] Within the construction site, a three-dimensional laser scan is performed to obtain three-dimensional laser data, which includes multiple laser point data, each laser point data including three-dimensional spatial coordinates;

[0031] Clustering optimization is performed on the three-dimensional laser data to obtain multiple clustered laser data clusters.

[0032] First, a 3D laser scan is performed at the construction site to obtain 3D laser data. This 3D laser data includes multiple laser point data, each with 3D spatial coordinates. 3D laser scanning refers to a technology that uses a 3D laser scanner to emit a laser beam to rapidly scan the construction site space and record the reflection information of the laser points, enabling omnidirectional, non-contact data acquisition. 3D laser data, also known as a 3D laser point cloud, refers to a collection of massive laser points generated by scanning, serving as the original data source for modeling. Laser point data refers to a single data unit in the 3D laser point cloud, with each laser point uniquely corresponding to a location in space. 3D spatial coordinates are parameters used to describe the position of a laser point in 3D space, represented using a Cartesian coordinate system (x, y, z), where the x / y axes correspond to the horizontal plane and the z-axis corresponds to the vertical height. A 3D laser scanner is set up at the construction site, with a scanning accuracy of 1 meter, and the scanner is activated to perform a 360° omnidirectional scan of the target area. The scanner records the (x, y, z) 3D spatial coordinates of each laser point using the principle of laser ranging, ultimately generating 3D laser data containing approximately 5 million laser points.

[0033] For example, some of the laser point data obtained after scanning are as follows: Laser point 1: (x=5.2, y=3.1, z=2.8m), corresponding to the middle position of the No. 2 frame column; Laser point 2: (x=8.7, y=6.3, z=3.5m), corresponding to the lower flange position of the No. 1 main beam; Laser point 3: (x=6.5, y=4.2, z=1.2m), corresponding to the position of the temporary scaffolding steel pipe.

[0034] Secondly, cluster optimization is performed on the three-dimensional laser data to obtain multiple clustered laser data clusters. Cluster optimization refers to the process of dividing the three-dimensional laser data into multiple non-overlapping subsets (data clusters) based on the spatial distance characteristics between laser points, so that the laser points in the same subset are spatially closer, and the laser points in different subsets are spatially farther apart.

[0035] The three-dimensional laser data is clustered and optimized to obtain multiple clustered laser data clusters, including:

[0036] Obtain the total number of structures within the construction site and set it as the number of cluster centers;

[0037] Within the three-dimensional laser data, the number of laser point data points of the specified number of cluster centers are randomly selected as multiple first cluster centers, and the distances between other laser point data points and the multiple first cluster centers are calculated to obtain multiple sets of first cluster distances;

[0038] Other laser point data are classified into laser data clusters with the smallest first cluster center within multiple first cluster distance sets, thus obtaining multiple first laser data clusters;

[0039] Multiple cluster center selection and cluster optimization processes are performed until convergence. The laser data clusters with the smallest average distance during the clustering process are retained as multiple clustered laser data clusters.

[0040] First, obtain the total number of structures within the construction site and set it as the number of cluster centers. The number of cluster centers refers to the target number of cluster groups, i.e., the final number of data clusters to be obtained, which must match the actual total number of structures on the construction site. For example, through verification of construction site drawings and on-site inspection, it is confirmed that the total number of structures in the scanned area is 8, including 6 frame columns and 2 main beams. Therefore, the number of cluster centers is set to 8.

[0041] Secondly, within the three-dimensional laser data, the number of laser point data points corresponding to the number of cluster centers is randomly selected as multiple first cluster centers. The distances between other laser point data points and these first cluster centers are calculated to obtain multiple sets of first cluster distances. The first cluster centers are laser points randomly selected in the initial stage of clustering, serving as the initial core points of each data cluster. The first cluster distance set refers to the set of distances from a single laser point to all first cluster centers, calculated using the Euclidean distance formula: .

[0042] For example, eight laser points are randomly selected from 5 million laser point data points as the first cluster centers, labeled C1 to C8, with the following three-dimensional coordinates: C1 (5.0, 3.0, 2.5), C2 (9.0, 4.0, 2.6), C3 (7.5, 7.0, 3.2), C4 (3.0, 8.0, 2.7), C5 (12.0, 5.0, 2.8), C6 (15.0, 9.0, 2.9), C7 (8.0, 12.0, 3.3), and C8 (11.0, 13.0, 3.1). The distances from other laser points to C1 to C8 are calculated using the Euclidean distance formula, generating the first cluster distance set. For example, the distance from laser point 1 (5.2, 3.1, 2.8) to each center is calculated as follows: Distance from laser point 1 to C1 = The distance from laser point 1 to C2 = Thus, the first cluster distance set of laser point 1 is {0.37, 3.91, 5.22, 6.13, 7.05, 9.87, 8.92, 7.93}m.

[0043] Furthermore, the other laser point data are categorized into laser data clusters with the smallest distance to the first cluster center within multiple first cluster distance sets, resulting in multiple first laser data clusters. The first laser data clusters refer to temporary data clusters formed after the initial clustering, and their accuracy needs to be improved through iterative optimization. Each laser point is categorized into the cluster corresponding to the first cluster center with the smallest distance within the first cluster distance set. For example, laser point 1 has the smallest distance to C1 (0.37m), so it is categorized into the first laser data cluster corresponding to C1; laser point 2 (8.7, 6.3, 3.5) has the smallest distance to C3 (1.05m), so it is categorized into the first laser data cluster corresponding to C3; laser point 3 (6.5, 4.2, 1.2) has the smallest distance to C1 (2.13m), so it is categorized into the first laser data cluster corresponding to C1. Ultimately, eight first laser data clusters are formed, each containing several laser points.

[0044] Finally, multiple cluster center selection and optimization processes are performed until convergence. The laser data clusters with the smallest average distance during the clustering process are retained as multiple clustered laser data clusters. Convergence means that after multiple iterations, the location of the cluster center no longer changes significantly (the change is less than 0.01m), or the average distance between laser points within a data cluster stabilizes within a fixed range; at this point, the clustering process has reached a stable state. The average distance from all laser points within a single data cluster to the cluster center is calculated to measure the aggregation effect of the data cluster; the smaller the average distance, the higher the aggregation degree. The clustered laser data clusters refer to the stable data clusters obtained after the final optimization; each cluster corresponds to a set of laser points from a construction structure or interfering object. For example, the geometric center of each first laser data cluster is recalculated. The original center of cluster C1 is (5.0, 3.0, 2.5). The new center is calculated as the average of the x, y, and z coordinates of all laser points within the cluster, resulting in (5.1, 3.2, 2.7). Using the new cluster center as a benchmark, the process of calculating the distance from the laser points to the center and classifying them is repeated for iterative optimization. After each iteration, the average distance of each cluster is calculated. When it reaches the 5th iteration, the maximum deviation between the new cluster center and the center of the previous round is 0.008m (less than 0.01m), and the average distance of each cluster is stable between 0.2 and 0.5m, indicating that the clustering has converged. By comparing the average distance of each cluster in all iterations, the 8 laser data clusters with the smallest average distance are retained, which are the final multiple clustered laser data clusters. These clusters correspond to the laser point sets of 6 frame columns and 2 main beams, respectively. Interference data such as temporary scaffolding form independent small clusters, which will be removed later through confidence screening.

[0045] In this embodiment of the invention, disordered laser points are divided into multiple targeted data clusters, each corresponding to a target construction structure. This organizes the originally chaotic raw data, providing a precise data source for subsequent structure fitting. Spatial distance clustering aggregates laser points belonging to the same construction structure, while laser points from interfering objects such as temporary supports and safety nets form independent small clusters, reducing the impact of interfering data on subsequent modeling. Each clustered data cluster contains only laser points of a single structure, avoiding fitting bias caused by mixed data from different structures, reducing the amount of data required for fitting calculations, and improving the efficiency and accuracy of subsequent iterative fitting.

[0046] S200: Based on multiple preset construction structures at the construction site, iteratively fit multiple clustered laser data clusters to obtain multiple sets of confidence scores for fitted structures, and then select multiple matching construction structures.

[0047] In this embodiment of the invention, based on various preset construction structures at the construction site, multiple clustered laser data clusters are iteratively fitted to obtain multiple sets of confidence scores for fitted structures, and multiple matching construction structures are selected. After clustering optimization in S100, the original point cloud has been divided into 8 independent clustered laser data clusters, but the specific construction structure corresponding to each cluster is still unclear. Meanwhile, 3D laser scanning has inherent measurement errors, such as coordinate deviations of ±1mm, and the clustered data clusters may still retain a small number of interference points. To solve the problems of ambiguous correspondence between clusters and preset structures and inaccurate matching caused by scanning errors, it is necessary to accurately determine the actual construction structure corresponding to each clustered data cluster by iteratively fitting and calculating similarity based on the preset construction structures at the construction site, i.e., the 3D model of the target structure determined in the design phase, thus providing a clear structural type basis for subsequent BIM modeling.

[0048] Step S200 in the method provided in this embodiment of the invention includes:

[0049] Obtain multiple pre-designed construction structures at the construction site;

[0050] Based on various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the confidence set of the first fitted structure.

[0051] Continue iterative fitting within multiple other clustered laser data clusters to obtain multiple sets of fitting structure confidence scores;

[0052] Each of the multiple sets of fitted structure confidence scores selects the preset construction structure corresponding to the maximum fitted structure confidence score as multiple matching construction structures.

[0053] First, various pre-designed construction structures were acquired from the construction site. Pre-designed construction structures refer to the 3D design models of the target structures planned for construction, determined during the site design phase. Their dimensions, shapes, and spatial parameters are completely consistent with the actual structures to be built, serving as the benchmark template for fitting and matching. The 3D design model is a digital model created using BIM software, containing the structure's geometric parameters and spatial location design information. For example, from the construction design drawings and BIM design files, the pre-designed construction structures corresponding to the scanned area were extracted, and their 3D design model parameters were defined: Pre-designed structure 1: Frame column (circular cross-section, diameter 0.8m, design height 4m, material: concrete); Pre-designed structure 2: Main beam (rectangular cross-section, width 0.6m × height 1.0m, design length 12m, material: concrete). The 3D models of both pre-designed construction structures were exported in a common BIM format and used as the fitting benchmark. The pre-designed construction structures in this scanned area only include frame columns and main beams; therefore, the acquired set of pre-designed construction structures is {3D model of frame columns, 3D model of main beams}, whose parameters completely match the 6 frame columns and 2 main beams to be built on the actual construction site.

[0054] Secondly, according to various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the confidence set of the first fitted structure.

[0055] Among them, according to various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the first set of confidence scores for the fitted structure, including:

[0056] The first preset construction structure is randomly fitted multiple times within the first cluster of laser data. The proportion of laser point data within the first cluster of laser data that are less than the preset distance threshold from the first preset construction structure is calculated to obtain the first structure fitting accuracy set.

[0057] The maximum value within the first set of structure fitting accuracy is selected as the first fitting structure confidence of the first preset construction structure.

[0058] The other preset construction structures are used to perform multiple random fittings within the first cluster of laser data to obtain the confidence set of the first fitted structure.

[0059] First, the first preset construction structure is randomly fitted multiple times within the first cluster of laser data. The proportion of laser points within the first cluster of laser data that are less than a preset distance threshold from the first preset construction structure is calculated, thus obtaining the first structure fitting accuracy set. The first preset construction structure refers to the first structure in the preset construction structure set that is fitted first. Multiple random fittings refer to the process of randomly adjusting the three-dimensional coordinates and orientation of the preset structure within the spatial boundary of the first cluster of laser data, and repeating the fitting operation. The number of fittings must cover different spatial positions to avoid random errors. The first structure fitting accuracy set is the set composed of the proportion of fitted laser points calculated after each random fitting of the first preset construction structure, with each element corresponding to the accuracy value of a single fitting.

[0060] For example, the first preset construction structure is a three-dimensional model of a frame column with a circular cross-section diameter of 0.8m and a height of 4m. The fitting range is limited to the spatial range of cluster 1 to avoid exceeding the laser point distribution area within the cluster. Ten random fittings are performed: during each fitting, the center coordinates of the three-dimensional model of the frame column are randomly translated within the range of cluster 1. The single translation amount in the x, y, and z directions is ≤0.2m, and the rotation angle is ≤5°. Since the frame column is an axisymmetric structure, the rotation has little impact on the fitting. The laser point distribution is mainly adapted through translation. The accuracy is calculated after each fitting: Fitting accuracy = (Number of laser points in the cluster that are <0.1m away from the preset structure / Total number of laser points in the cluster) × 100%. For example, cluster 1 has a total of 62,000 laser points. The accuracy calculation results of 10 fittings are as follows: First fitting: 54,560 laser points, accuracy = 54,560 / 62,000 × 100% = 88%; Third fitting: 57,040 laser points, accuracy = 57,040 / 62,000 × 100% = 92%; Sixth fitting: 57,660 laser points, accuracy = 57,660 / 62,000 × 100% = 93%; The accuracy of the remaining 7 fittings are 90%, 89%, 91%, 87%, 90%, 92%, and 89% respectively; The final set of accuracy for the first structure fitting is: {88%, 90%, 92%, 89%, 91%, 93%, 87%, 90%, 92%, 89%}.

[0061] Secondly, the maximum value within the first set of fitting accuracies is selected as the first fitting structure confidence score for the first preset construction structure. The first fitting structure confidence score is the final matching index between the first preset construction structure and the first clustered laser data cluster. Taking the maximum value of multiple fitting accuracies can minimize the impact of scanning errors and positional deviations in single fittings. For example, the first set of fitting accuracies {88%, 90%, 92%, 89%, 91%, 93%, 87%, 90%, 92%, 89%} is numerically sorted; the maximum value of 93% is extracted and determined as the first fitting structure confidence score corresponding to the first preset construction structure. The maximum value of 10 fitting accuracies is 93%, representing that at the optimal fitting position, 93% of the laser points in cluster 1 are in contact with the surface of the frame column model (distance < 0.1m), directly reflecting the high matching degree between cluster 1 and the frame column structure. Therefore, 93% is used as the final confidence score for the frame column.

[0062] Furthermore, other preset construction structures are used to perform multiple random fittings within the first clustered laser data cluster to obtain the first fitted structure confidence set. Other preset construction structures refer to the remaining structures in the preset construction structure set excluding the first preset construction structure. The first fitted structure confidence set is a set composed of the first fitted structure confidence scores of all preset construction structures with the first clustered laser data cluster, including the final matching degree between each preset structure and the cluster. Other preset construction structures are imported, and the above operation is repeated: 10 random fittings are performed within the cluster 1 space, the accuracy of each fitting is calculated, and the maximum value is extracted as the first fitted structure confidence score of that structure; the confidence scores of the first preset construction structure and other preset construction structures (main beams) are collected to form the first fitted structure confidence set. For example, the second preset construction structure is the three-dimensional model of the main beam: a rectangular cross-section of 0.6m × 1.0m and a length of 12m. The accuracy set of the main beam three-dimensional model in cluster 1 for 10 fittings is {6%, 8%, 7%, 5%, 9%, 7%, 6%, 8%, 5%, 7%}, with a maximum value of 9%, that is, the confidence level of the first fitting structure corresponding to the main beam is 9%. Finally, the confidence set of the first fitting structure is obtained: {frame column: 93%, main beam: 9%}, which fully presents the matching degree between the first clustered laser data cluster and all preset construction structures.

[0063] Furthermore, iterative fitting is performed within multiple other clustered laser data clusters to obtain multiple sets of fitting structure confidence scores. For example, for the remaining 7 clustered laser data clusters (cluster 2 to cluster 8), the iterative fitting process is repeated: each cluster is iteratively fitted 10 times using the 3D models of frame columns and main beams, and the confidence score of the fitted structure is calculated each time; for example, the confidence score set of cluster 3 fitted with the main beam model is {89%, 91%, 88%, 93%, 90%, 87%, 92%, 89%, 91%, 88%}, and the confidence score set of cluster 3 fitted with the frame column model is {7%, 9%, 8%, 6%, 10%, 7%, 8%, 9%, 6%, 8%}; finally, 8 sets of fitting structure confidence scores are obtained, one set for each cluster, and each set contains 10 iterations of confidence score data for both frame columns and main beams.

[0064] Finally, within each set of fitted structure confidence scores, the pre-defined construction structures corresponding to the maximum fitted structure confidence scores are selected as multiple matching construction structures. A matching construction structure refers to the pre-defined construction structure corresponding to the maximum value in the fitted structure confidence score set for each cluster of laser data, i.e., the target structure with the highest matching degree to the laser points within the cluster. For each fitted structure confidence score set, the maximum confidence scores of the two types of pre-defined structures are extracted. For example, the maximum confidence score for the frame columns in cluster 1 is 93%, and the maximum confidence score for the main beams is 9%; the maximum confidence score for the main beams in cluster 3 is 93%, and the maximum confidence score for the frame columns is 10%. The pre-defined construction structure corresponding to the maximum value is selected as the matching construction structure for that cluster. For example, the maximum confidence scores for clusters 1 to 6 all come from frame columns, so they are matched as frame columns; the maximum confidence scores for clusters 7 to 8 all come from main beams, so they are matched as main beams. Ultimately, eight matching construction structures are obtained: {6 frame columns, 2 main beams}, which are completely consistent with the actual number and type of structures to be built on the construction site, achieving a precise correspondence between each cluster and the pre-defined structures.

[0065] In this embodiment of the invention, clustered laser data clusters are accurately matched to the construction structure through iterative fitting, providing a clear structural type basis for BIM modeling. By using multiple random iterative fittings to obtain the maximum value, deviations caused by scanning errors in a single fitting are avoided, making the confidence calculation more representative and improving the matching accuracy. Using the proportion of laser points as the core confidence index, the similarity between the cluster and the preset structure is intuitively reflected, avoiding subjective judgment errors and ensuring that the matching results are highly consistent with the actual structure at the construction site.

[0066] S300: Based on the construction progress information at the construction site, and combined with multiple fitted structure confidence sets, a low confidence ratio is configured to screen and obtain multiple low confidence clustered laser data clusters.

[0067] In this embodiment of the invention, based on the construction progress information at the construction site, and combined with the low confidence ratio configured from multiple fitted structure confidence sets, multiple low-confidence clustered laser data clusters are obtained. After iterative fitting in S200, although matching construction structures for each clustered laser data cluster have been obtained, two key interfering factors exist at the construction site: first, the construction progress may be delayed, with some structures not being constructed or not completed, resulting in naturally low fitting confidence for the corresponding clusters; second, some clusters may have inaccurate fitting due to occlusion or residual noise, manifesting as significantly lower fitting confidence than other clusters. If all clusters are directly used for modeling, the model accuracy will be reduced due to invalid clusters with unconstructed structures or incorrectly fitted clusters. Therefore, it is necessary to determine whether a structure should exist based on the construction progress, and to determine the accuracy of the matching based on the fitting confidence. A low confidence ratio should be scientifically configured to screen out low-confidence clustered laser data clusters that may not fit correctly, providing target objects for subsequent targeted optimization.

[0068] Step S300 in the method provided in this embodiment of the invention includes:

[0069] Obtain construction progress information from the construction site;

[0070] Based on the construction progress information, the low confidence ratio of the progress is calculated. The formula for calculating the low confidence ratio of the progress is: Low confidence ratio = (1 - actual completion ratio) × progress weight coefficient.

[0071] Based on the confidence sets of multiple fitted structures, the proportion of low-confidence fit is calculated. The proportion of low-confidence fit is the ratio of the proportion of the multiple largest fitted structures whose confidence is less than the confidence threshold multiplied by the fitting weight coefficient.

[0072] The low confidence ratio is calculated based on the aforementioned low confidence ratio for progress and the fitted low confidence ratio, wherein the low confidence ratio is obtained by weighting the low confidence ratio for progress and the fitted low confidence ratio.

[0073] Based on the low confidence ratio and the total number of multiple clustered laser data clusters, multiple low confidence clustered laser data clusters are obtained through screening.

[0074] First, obtain the construction progress information of the construction site. Construction progress information refers to the comparison data between the actual construction progress and the planned progress at the construction site, including the planned completion percentage, actual completion percentage, and types of unconstructed structures of the target structure, used to determine whether the structure corresponding to the cluster exists. The planned completion percentage refers to the proportion of the target structure that should be completed at the current time node in the construction plan, such as the completion percentage of the pouring and installation of frame columns and main beams. The actual completion percentage refers to the proportion of the target structure that has actually been completed at the construction site, confirmed through on-site inspection and supervision records. For example, from the construction site's progress management system, construction logs, and supervision reports, the construction progress information for the scanned area is extracted: Planned progress: The current node should complete the concrete pouring of all frame columns (6 columns) and main beams (2 beams) on the first floor, with a planned completion rate of 100%; Actual progress: After on-site verification, all 6 frame columns have been poured (completion rate 100%), and only 1 of the 2 main beams has been poured (completion rate 50%), with an overall actual completion rate of (6×100%+2×50%) / (6+2)=87.5%; Unconstructed structure: 1 main beam (corresponding to main beam No. 2 in the design drawings). Therefore, the core data of the construction progress information obtained this time is: {planned completion rate 100%, actual completion rate 87.5%, unconstructed structure: 1 main beam}, which clearly indicates that there is currently an unconstructed structure, and the corresponding cluster may have a low fitting confidence due to the absence of the structure.

[0075] Secondly, based on the construction progress information, the low confidence ratio of the progress is calculated. The formula for calculating the low confidence ratio is: Low Confidence Ratio = (1 - Actual Completion Ratio) × Progress Weight Coefficient. The low confidence ratio of the progress is a ratio calculated based on the construction progress deviation, representing the probability that the cluster fitting confidence is low due to the structure not being constructed or not being completed. The larger the progress deviation, the lower the actual completion ratio, and the higher the low confidence ratio. The low confidence ratio of the progress is calculated as (1 - Actual Completion Ratio) × Progress Weight Coefficient. The progress weight coefficient ranges from 0 to 1, adjusted according to the construction type. For building structure construction, it is taken as 0.6, taking into account the impact of progress deviation on the existence of the structure. For example, given an actual completion ratio of 87.5% and a progress weight coefficient of 0.6, the low confidence ratio of the progress is (1 - 87.5%) × 0.6 = 12.5% ​​× 0.6 = 7.5%. Because the actual completion rate did not meet the plan (87.5% < 100%), there are main beams that have not been constructed. The corresponding cluster fitting confidence may be low. Therefore, the calculated low confidence ratio of progress is 7.5%, which reflects the expected proportion of low confidence clusters caused by progress factors.

[0076] Furthermore, based on multiple sets of confidence scores for fitted structures, the proportion of low-confidence fits is calculated. This proportion is the ratio of the percentage of the highest-confidence fitted structures whose confidence scores are less than a confidence threshold, multiplied by the fitting weight coefficient. The low-confidence fit proportion refers to the percentage of clusters with low confidence scores due to inaccurate fitting, based on the statistical proportion of the highest-confidence fitted structures across all clusters. The more clusters with low confidence scores, the higher this proportion. The confidence threshold is the critical value for judging the accuracy of the fit. For example, considering the fitting accuracy of S200, a confidence threshold of 89% can be set; clusters below this value are considered to have poor fitting performance. The low-confidence fit proportion = (Number of clusters with highest-confidence fitted structures < confidence threshold / Total number of clusters) × fitting weight coefficient. The fitting weight coefficient ranges from 0 to 1, with a value of 0.8 to highlight the direct impact of the fitting performance on low-confidence clusters. For example, the total number of clusters is 8, and the confidence threshold is 89%. The cluster with the maximum fit confidence score < 89% is cluster 6 (88%), totaling 1. The low-confidence fit ratio is (1 / 8) × 0.8 = 12.5% ​​× 0.8 = 10%. Among the 8 clusters, only cluster 6 has a maximum fit confidence score of 88% that is lower than the threshold of 89%, indicating a poor fit. Therefore, the calculated low-confidence fit ratio is 10%, directly reflecting the proportion of low-confidence clusters in the fitted data itself.

[0077] Furthermore, based on the aforementioned low confidence ratios for progress and fitting, a low confidence ratio is calculated. This low confidence ratio is obtained by weighting the low confidence ratios for progress and fitting. The low confidence ratio is the proportion of low-confidence clusters to be screened out of the total number of clusters, determined by considering both progress and fitting factors. It is calculated using a weighted average method, taking into account the impact of both factors on low-confidence clusters. Low confidence ratio = low confidence ratio for progress × 0.4 + low confidence ratio for fitting × 0.6. Since the fitting confidence directly reflects the matching accuracy, the fitting factor has a higher weight. For example, if the low confidence ratio for progress is 7.5% and the low confidence ratio for fitting is 10%, the low confidence ratio is 7.5% × 0.4 + 10% × 0.6 = 9%. Taking into account the unconstructed structures caused by progress deviations and clusters with poor fitting effects, the final low confidence ratio is determined to be 9%, meaning that approximately 9% of the total clusters will be selected as low-confidence clusters in the laser data analysis.

[0078] Finally, based on the low confidence ratio and the total number of multiple clustered laser data clusters, multiple low confidence clustered laser data clusters are obtained through screening.

[0079] Among them, based on the low confidence ratio and the total number of multiple clustered laser data clusters, multiple low confidence clustered laser data clusters are obtained, including:

[0080] The number of low-confidence data clusters is calculated based on the low-confidence ratio and the total number of clustered laser data clusters.

[0081] Calculate the maximum fitting confidence scores of multiple clustered laser data clusters and multiple matched construction structures, and sort them in descending order;

[0082] Select the lowest-confidence cluster of laser data as multiple low-confidence clusters.

[0083] First, the number of low-confidence clusters is calculated based on the low-confidence ratio and the total number of clustered laser data sets. The number of low-confidence clusters refers to the specific number of low-confidence clustered laser data sets to be screened. It is calculated by multiplying the low-confidence ratio by the total number of clusters. Since the number of clusters is an integer, it needs to be rounded up to avoid missing potential low-confidence clusters. Low-confidence cluster = Total number of clustered laser data sets × Low-confidence ratio. For example, if the total number of clustered laser data sets is 8 and the low-confidence ratio is 9%, substituting into the formula, the low-confidence cluster = 8 × 9% = 0.72. Rounding up, this is 1, meaning 1 low-confidence cluster needs to be screened. If the low-confidence ratio is 15% and the total number of clusters is 8, then the low-confidence cluster = 8 × 15% = 1.2. Rounding up, this is 2, ensuring that more clusters that may not be properly fitted are covered.

[0084] Secondly, the confidence scores of multiple clustered laser data clusters and multiple matching construction structures are calculated based on their maximum fitting scores, and then sorted in descending order. The maximum fitting score is the highest confidence score obtained after iterative fitting of all preset construction structures within S200 for each clustered laser data cluster. For example, the maximum score for cluster 6 is 88%, which is a core indicator for measuring the accuracy of cluster-structure matching. The sorting logic involves ranking the clusters by their maximum fitting scores from highest to lowest, placing the cluster with the lowest confidence score at the end of the sort for easier subsequent selection. For example, the maximum fitting structure confidence scores of all clusters are extracted and organized into a set: {Cluster 1 (93%), Cluster 2 (89%), Cluster 3 (93%), Cluster 4 (94%), Cluster 5 (90%), Cluster 6 (88%), Cluster 7 (92%), Cluster 8 (91%)}. These clusters are then sorted from largest to smallest. If the confidence scores are the same, such as Cluster 1 and Cluster 3 both being 93%, they are sorted by cluster number in ascending order. The final sorted result is: Cluster 4 (94%), Cluster 1 (93%), Cluster 3 (93%), Cluster 7 (92%), Cluster 8 (91%), Cluster 5 (90%), Cluster 2 (89%), Cluster 6 (88%). The sorting clearly shows the matching accuracy of each cluster: Cluster 4 has the highest confidence score (94%) and the most accurate match; Cluster 6 has the lowest confidence score (88%) and the worst matching accuracy, providing an intuitive basis for subsequent selection.

[0085] Finally, the clusters with the lowest confidence count at the end of the ranking are selected as multiple low-confidence clusters. Low-confidence clusters refer to the clusters with the lowest confidence count at the end of the ranking; their maximum value has the lowest confidence in the fitted structure and is the most likely to have incorrect fits. For example, if the lowest confidence count is 1, the last cluster in the ranking is cluster 6 (88%); this cluster is determined as a low-confidence cluster, i.e., the selection result is {cluster 6}. If the lowest confidence count is 2, then the last two clusters in the ranking (cluster 2: 89%, cluster 6: 88%) are selected as low-confidence clusters; in this example, only 1 needs to be selected, so cluster 6 becomes the only low-confidence cluster. Cluster 6 has the lowest confidence, which may be due to some areas being obscured by scaffolding during construction or a small number of interfering points being mixed in during clustering, leading to inaccurate fitting with the frame columns.

[0086] In this embodiment of the invention, the combination of construction progress and fitting confidence avoids misjudgment caused by a single factor; the abstract progress deviation and fitting effect are transformed into specific proportions through weighted calculation, making the screening logic more scientific and operable; the selected low-confidence clusters are the core objects for subsequent S400 occlusion combination fitting, providing a clear direction for targeted solutions to the problem of inaccurate fitting and improving the reliability of the overall modeling.

[0087] S400: Multiple low-confidence clustered laser data clusters are combined with the nearest clustered laser data clusters to fit various preset occlusion combination structures, obtain multiple occlusion structure confidence sets, filter to obtain multiple matching occlusion combination structures, combine multiple matching construction structures, perform BIM modeling, and obtain modeling results.

[0088] In this embodiment of the invention, multiple low-confidence clustered laser data clusters are combined with the nearest clustered laser data cluster to perform various preset occlusion combination structure fittings, obtaining multiple occlusion structure confidence sets. Multiple matching occlusion combination structures are then selected and combined with multiple matching construction structures for BIM modeling to obtain the modeling results. The low-confidence clustered laser data clusters selected by S300 likely have low fitting confidence due to occlusion interference from the construction site, such as scaffolding or temporary supports obscuring the surface of frame columns, resulting in incomplete and inaccurate point cloud data. The point cloud of a single cluster cannot fully reflect the spatial relationship between the target structure and the occluding object, and direct fitting is prone to deviation. Therefore, it is necessary to combine low-confidence clusters with the nearest normal clusters, and use preset occlusion combination structures for iterative fitting to accurately identify occlusion scenarios, correct the fitting deviation of low-confidence clusters, and finally combine all matching structures to achieve high-precision BIM modeling.

[0089] Step S400 in the method provided in this embodiment of the invention includes:

[0090] This involves combining multiple low-confidence clustered laser data clusters with the nearest clustered laser data cluster to perform various occlusion combination structure fitting, resulting in multiple occlusion structure confidence sets, including:

[0091] Calculate the distances between other clustered laser data clusters and the cluster centers of multiple low-confidence clustered laser data clusters, and select the clustered laser data cluster with the smallest distance from each low-confidence clustered laser data cluster to obtain a combination of multiple clustered laser data clusters;

[0092] Obtain multiple pre-set shielding combinations within the construction site;

[0093] Multiple preset occlusion combinations are used, and iterative fitting is performed within multiple clustered laser data clusters to obtain multiple sets of occlusion structure confidence.

[0094] First, the distances between other clustered laser data clusters and the cluster centers of multiple low-confidence clustered laser data clusters are calculated. The cluster with the smallest distance among the low-confidence clustered laser data clusters is then selected, resulting in multiple clustered laser data cluster combinations. The cluster center refers to the geometric center of each clustered laser data cluster, i.e., the average of the x, y, and z coordinates of all laser points within the cluster, serving as the reference point for calculating inter-cluster distances. Inter-cluster distance refers to the straight-line distance between the cluster centers of two clustered laser data clusters, calculated using the Euclidean distance formula. A clustered laser data cluster combination is a set consisting of a low-confidence cluster and its nearest normal cluster, used to completely cover the point cloud range of the target structure and occlusions. For example, extract the optimized cluster center coordinates of all normal clusters (cluster 1~cluster 5, cluster 7~cluster 8): C1(5.1,3.2,2.7), C2(9.2,4.1,2.6), C3(7.6,7.1,3.3), C4(3.1,8.2,2.8), C5(12.1,5.3,2.9), C7(8.1,12.2,3.4), C8(11.2,13.1,3.2); calculate the distance between C6 and the other 7 cluster centers using the Euclidean distance formula, such as the distance from C6(15.2,9.3,2.8) to C5(12.1,5.3,2.9) = Similarly, other distances were calculated: C6-C1≈12.3m, C6-C2≈6.8m, C6-C3≈8.5m, C6-C4≈12.1m, C6-C7≈7.2m, C6-C8≈8.1m. The cluster with the smallest distance was selected: the distance from C6 to C5 (5.06m) was the smallest. Therefore, cluster 6 was combined with cluster 5 to form a clustered laser data cluster combination: {cluster 5 (normal cluster, 90% confidence, corresponding to frame column 5), cluster 6 (low confidence cluster)}. Since cluster 6 and cluster 5 are spatially closest, it is speculated that frame column 6 corresponding to cluster 6 is adjacent to frame column 5 corresponding to cluster 5. Furthermore, cluster 6 may be affected by occlusions around cluster 5, such as scaffolding connecting the two frame columns. Therefore, combining the point clouds of the two clusters can completely cover the spatial range occluded by frame column 5, frame column 6, and the scaffolding, providing complete data for subsequent occlusion combination fitting.

[0095] Secondly, multiple pre-defined occlusion combinations within the construction site are obtained. These pre-defined occlusion combinations are 3D models of the target construction structure and its occluders, pre-constructed based on common occlusion scenarios at the construction site. They include the geometric parameters of the target structure and the type and shape information of the occluders, serving as a benchmark template for fitting occlusion scenarios. Occluders refer to temporary objects at the construction site that may obstruct the target structure, such as scaffolding, safety nets, temporary supports, and construction machinery. Combining the characteristics of the construction site, common occlusion combinations related to frame columns and main beams are extracted from the BIM design documents and construction organization design to form a set of pre-defined occlusion combination structures. For example, Combination Structure 1: Frame column + steel pipe scaffolding (covering material: φ48mm steel pipe scaffolding, grid spacing 0.8m×1.2m); Combination Structure 2: Frame column + safety net (covering material: dense mesh safety net, thickness 0.02m); Combination Structure 3: Frame column + temporary support (covering material: steel temporary support, cross section 0.1m×0.1m); Combination Structure 4: Double frame column + connecting scaffolding (covering material: horizontal scaffolding connecting two frame columns, suitable for the double-column scenario of cluster 5 + cluster 6). The preset covering combination structure set is {Combination 1, Combination 2, Combination 3, Combination 4}, all designed for frame columns. Among them, Combination 4 is specifically adapted to the scenario of two adjacent columns + scaffolding connection, and is highly compatible with the combination of cluster 5 + cluster 6.

[0096] Furthermore, multiple preset occlusion combination structures are employed, and iterative fitting is performed within multiple clustered laser data cluster combinations to obtain multiple sets of occlusion structure confidence scores. Occlusion structure confidence score is an indicator measuring the degree of matching between a clustered laser data cluster combination and a specific preset occlusion combination structure. Its calculation method is consistent with the fitting structure confidence score: Occlusion structure confidence score = (Number of laser points within the cluster combination whose distance to the preset occlusion combination structure is less than a threshold / Total number of laser points in the cluster combination) × 100%. Higher confidence scores indicate more accurate occlusion scene matching. The occlusion structure confidence score set refers to the set of iterative fitting confidence scores for all preset occlusion combination structures for a single cluster combination. For example, using cluster 5 + cluster 6 as the fitting object, 10 random iterations are performed on each of the four preset occlusion combination structures: Fitting constraints: Within the spatial range of the cluster combination (… Within the range, the position (translation ≤ 0.3m) and orientation (rotation ≤ 5°) of the combined model are randomly adjusted; after each fitting, the proportion of laser points that meet the distance requirements is calculated, which is the confidence of the occlusion structure, and all results are recorded: for example, the confidence of the 10 fittings of combined structure 4 (double frame columns + connecting scaffolding) is: {86%, 88%, 91%, 89%, 93%, 87%, 90%, 89%, 92%, 94%}; finally, the confidence set of the occlusion structure is formed: {Combination 1: [65%,…,68%], Combination 2: [58%,…,61%], Combination 3: [72%,…,76%], Combination 4: [86%,…,94%]}.

[0097] This process involves selecting multiple matching occlusion combinations, combining them with multiple matching construction structures, performing BIM modeling, and obtaining the modeling results, including:

[0098] Within multiple sets of occlusion structure confidence scores, preset occlusion combination structures corresponding to the maximum occlusion structure confidence scores are selected as multiple matching occlusion combination structures.

[0099] Based on multiple matching construction structures and multiple matching occlusion combination structures, BIM modeling is performed to obtain the modeling results. Among these, there are multiple duplicate matching construction structures within the multiple matching construction structures and multiple matching occlusion combination structures.

[0100] First, within multiple sets of occlusion structure confidence scores, preset occlusion combination structures corresponding to the highest occlusion structure confidence scores are selected as multiple matching occlusion combination structures. A matching occlusion combination structure refers to the optimal matching result between a cluster combination and a preset occlusion combination structure; that is, the preset occlusion combination structure corresponding to the highest confidence score in the occlusion structure confidence score set most accurately reflects the actual occlusion scenario. For example, the highest confidence scores of each preset occlusion combination structure are extracted: Combination 1: 70%, Combination 2: 63%, Combination 3: 76%, Combination 4: 94%. The preset occlusion combination structure corresponding to the highest confidence score of 94% (Combination 4: double frame columns + connecting scaffolding) is selected and determined as the matching occlusion combination structure of cluster 5 + cluster 6. The highest confidence score of 94% for Combination 4 means that at the optimal fitting position, 94% of the laser points in the cluster combination are aligned with the model surface of the double frame columns + connecting scaffolding, perfectly matching the actual scenario of two adjacent frame columns being occluded by the connecting scaffolding at the construction site. Therefore, it is selected as the matching occlusion combination structure.

[0101] Secondly, BIM modeling was performed based on multiple matching construction structures and multiple matching occlusion combinations to obtain the modeling results. Among these, multiple matching construction structures and multiple matching occlusion combinations contained duplicate matching construction structures. Duplicate matching construction structures refer to target construction structures included in the matching occlusion combinations that are the same type as the matching construction structures obtained from S200. These need to be integrated into a unified structure during modeling to avoid redundancy. Using BIM software such as Revit and Bentley, the 3D models of the matching construction structures and matching occlusion combinations were stitched and integrated according to their actual spatial locations to generate a complete digital model of the construction site.

[0102] For example, all valid matching structures are organized: Matching construction structures (S200 results): {frame columns × 5, main beams × 2} corresponding to clusters 1~5 and clusters 7~8; Matching occlusion combination structures: Combination 4 (double frame columns + connecting scaffolding), including the core target structure double frame columns (corresponding to clusters 5 and 6) and the occlusion connecting scaffolding. Duplicate structures are handled: The double frame columns in the matched occlusion combination structure are integrated with the frame columns of clusters 5 and 6 in the original matched construction structure, preserving the precise spatial relationship of the double frame columns and eliminating duplicate single frame column models. BIM software modeling: Import the 3D models of all matching structures (4 frame columns, 2 main beams, double frame columns + connecting scaffolding); adjust the model position according to the actual spatial coordinates of the cluster center (e.g., cluster 1 center (5.1m, 3.2m, 2.7m)) to ensure consistency with the construction site space; verify model accuracy: the distance error between all model surfaces and the corresponding point cloud is ≤0.1m, which meets the construction requirements; generate the final BIM model, which includes the digital information of all frame columns (6, including double column connections), main beams (2), and key obstructions (connecting scaffolding).

[0103] In this embodiment of the invention, by fitting the low-confidence cluster with the nearest cluster combination and the occlusion combination structure, the actual occlusion scenario was successfully identified, and the fitting deviation of the low-confidence cluster was corrected. By integrating the matching construction structure and the matching occlusion combination structure, the accurate information of the normal structure was preserved, and the digital model of the occlusion object was supplemented, avoiding the defect of only modeling the target structure and ignoring the occlusion interference. The BIM model includes the target structure and key occlusion objects, providing more comprehensive and accurate digital support for subsequent construction scheme optimization and quality acceptance.

[0104] Through the specific implementation methods described above, the embodiments of the present invention achieve the following technical effects:

[0105] This invention provides a method and system for BIM modeling of construction sites based on 3D laser scanning. It collects point cloud data from the construction site using 3D laser scanning and performs cluster optimization. Then, iteratively fits the clustered data clusters to match the target structure based on a preset construction structure. Next, it scientifically configures the low-confidence ratio and filters low-confidence clusters by combining construction progress and fitting confidence. Finally, it combines the low-confidence clusters with the nearest normal clusters and performs a preset occlusion combination structure fitting. This invention specifically solves problems such as data loss due to occlusion at the construction site, modeling interference introduced by noise, and inaccurate fitting. Ultimately, it achieves high-precision and high-reliability BIM modeling of construction sites, providing accurate and comprehensive digital support for construction progress control, quality supervision, scheme optimization, and quality acceptance.

[0106] Example 2, as Figure 2 As shown, this invention provides a construction site BIM modeling system based on three-dimensional laser scanning, the system comprising:

[0107] The laser data clustering module 11 is used to perform three-dimensional laser scanning at the construction site, obtain three-dimensional laser data, perform laser data clustering, and obtain multiple clustered laser data clusters.

[0108] The structure fitting and matching module 12 is used to iteratively fit multiple clustered laser data clusters based on multiple preset construction structures at the construction site, obtain multiple sets of fitting structure confidence, and filter multiple matching construction structures.

[0109] The low confidence screening module 13 is used to select and obtain multiple low confidence clustered laser data clusters by combining the construction progress information of the construction site with the low confidence ratio of multiple fitted structure confidence sets.

[0110] The occlusion modeling integration module 14 is used to combine multiple low-confidence clustered laser data clusters with the nearest clustered laser data clusters to fit multiple preset occlusion combination structures, obtain multiple occlusion structure confidence sets, filter to obtain multiple matching occlusion combination structures, combine multiple matching construction structures to perform BIM modeling, and obtain modeling results.

[0111] In one embodiment, the laser data clustering module 11 is further configured to:

[0112] Within the construction site, a three-dimensional laser scan is performed to obtain three-dimensional laser data, which includes multiple laser point data, each laser point data including three-dimensional spatial coordinates;

[0113] Clustering optimization is performed on the three-dimensional laser data to obtain multiple clustered laser data clusters.

[0114] The three-dimensional laser data is clustered and optimized to obtain multiple clustered laser data clusters, including:

[0115] Obtain the total number of structures within the construction site and set it as the number of cluster centers;

[0116] Within the three-dimensional laser data, the number of laser point data points of the specified number of cluster centers are randomly selected as multiple first cluster centers, and the distances between other laser point data points and the multiple first cluster centers are calculated to obtain multiple sets of first cluster distances;

[0117] Other laser point data are classified into laser data clusters with the smallest first cluster center within multiple first cluster distance sets, thus obtaining multiple first laser data clusters;

[0118] Multiple cluster center selection and cluster optimization processes are performed until convergence. The laser data clusters with the smallest average distance during the clustering process are retained as multiple clustered laser data clusters.

[0119] In one embodiment, the structure fitting and matching module 12 is further configured to:

[0120] Obtain multiple pre-designed construction structures at the construction site;

[0121] Based on various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the confidence set of the first fitted structure.

[0122] Continue iterative fitting within multiple other clustered laser data clusters to obtain multiple sets of fitting structure confidence scores;

[0123] Each of the multiple sets of fitted structure confidence scores selects the preset construction structure corresponding to the maximum fitted structure confidence score as multiple matching construction structures.

[0124] Among them, according to various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the first set of confidence scores for the fitted structure, including:

[0125] The first preset construction structure is randomly fitted multiple times within the first cluster of laser data. The proportion of laser point data within the first cluster of laser data that are less than the preset distance threshold from the first preset construction structure is calculated to obtain the first structure fitting accuracy set.

[0126] The maximum value within the first set of structure fitting accuracy is selected as the first fitting structure confidence of the first preset construction structure.

[0127] The other preset construction structures are used to perform multiple random fittings within the first cluster of laser data to obtain the confidence set of the first fitted structure.

[0128] In one embodiment, the low-confidence screening module 13 is further configured to:

[0129] Obtain construction progress information from the construction site;

[0130] Based on the construction progress information, the low confidence ratio of the progress is calculated. The formula for calculating the low confidence ratio of the progress is: Low confidence ratio = (1 - actual completion ratio) × progress weight coefficient.

[0131] Based on the confidence sets of multiple fitted structures, the proportion of low-confidence fit is calculated. The proportion of low-confidence fit is the ratio of the proportion of the multiple largest fitted structures whose confidence is less than the confidence threshold multiplied by the fitting weight coefficient.

[0132] The low confidence ratio is calculated based on the aforementioned low confidence ratio for progress and the fitted low confidence ratio, wherein the low confidence ratio is obtained by weighting the low confidence ratio for progress and the fitted low confidence ratio.

[0133] Based on the low confidence ratio and the total number of multiple clustered laser data clusters, multiple low confidence clustered laser data clusters are obtained through screening.

[0134] Among them, based on the low confidence ratio and the total number of multiple clustered laser data clusters, multiple low confidence clustered laser data clusters are obtained, including:

[0135] The number of low-confidence data clusters is calculated based on the low-confidence ratio and the total number of clustered laser data clusters.

[0136] Calculate the maximum fitting confidence scores of multiple clustered laser data clusters and multiple matched construction structures, and sort them in descending order;

[0137] Select the lowest-confidence cluster of laser data as multiple low-confidence clusters.

[0138] In one embodiment, the occlusion modeling integration module 14 is further configured to:

[0139] This involves combining multiple low-confidence clustered laser data clusters with the nearest clustered laser data cluster to perform various occlusion combination structure fitting, resulting in multiple occlusion structure confidence sets, including:

[0140] Calculate the distances between other clustered laser data clusters and the cluster centers of multiple low-confidence clustered laser data clusters, and select the clustered laser data cluster with the smallest distance from each low-confidence clustered laser data cluster to obtain a combination of multiple clustered laser data clusters;

[0141] Obtain multiple pre-set shielding combinations within the construction site;

[0142] Multiple preset occlusion combinations are used, and iterative fitting is performed within multiple clustered laser data clusters to obtain multiple sets of occlusion structure confidence.

[0143] This process involves selecting multiple matching occlusion combinations, combining them with multiple matching construction structures, performing BIM modeling, and obtaining the modeling results, including:

[0144] Within multiple sets of occlusion structure confidence scores, preset occlusion combination structures corresponding to the maximum occlusion structure confidence scores are selected as multiple matching occlusion combination structures.

[0145] Based on multiple matching construction structures and multiple matching occlusion combination structures, BIM modeling is performed to obtain the modeling results. Among these, there are multiple duplicate matching construction structures within the multiple matching construction structures and multiple matching occlusion combination structures.

[0146] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0147] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0148] This specification and accompanying drawings are merely illustrative examples of the invention and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its scope. Therefore, if such modifications and modifications fall within the scope of the invention and its equivalents, the invention is intended to include these modifications and modifications.

Claims

1. A construction site BIM modeling method based on 3D laser scanning, characterized in that, The method includes: Within the construction site, three-dimensional laser scanning is performed to obtain three-dimensional laser data. Laser data clustering is then performed to obtain multiple clustered laser data clusters. Based on various pre-set construction structures at the construction site, multiple clustered laser data clusters are iteratively fitted to obtain multiple sets of confidence scores for fitted structures, and multiple matching construction structures are selected. Based on the construction progress information at the construction site, and combined with the low confidence ratio of multiple fitted structure confidence sets, several low-confidence clustered laser data clusters were obtained, including: Obtain construction progress information from the construction site; Based on the construction progress information, the low confidence ratio of the progress is calculated. The formula for calculating the low confidence ratio of the progress is: Low confidence ratio = (1 - actual completion ratio) × progress weight coefficient. Based on the confidence sets of multiple fitted structures, the proportion of low-confidence fit is calculated. The proportion of low-confidence fit is the ratio of the proportion of the multiple largest fitted structures whose confidence is less than the confidence threshold multiplied by the fitting weight coefficient. The low confidence ratio is calculated based on the aforementioned low confidence ratio for progress and the fitted low confidence ratio, wherein the low confidence ratio is obtained by weighting the low confidence ratio for progress and the fitted low confidence ratio. Based on the aforementioned low confidence ratio and the total number of multiple clustered laser data clusters, multiple low-confidence clustered laser data clusters were obtained, including: The number of low-confidence data clusters is calculated based on the low-confidence ratio and the total number of clustered laser data clusters. Calculate the maximum fitting confidence scores of multiple clustered laser data clusters and multiple matched construction structures, and sort them in descending order; Select the clustered laser data with the lowest confidence count at the end of the sorting as multiple low-confidence clustered laser data clusters; Multiple low-confidence clustered laser data clusters are combined with the nearest clustered laser data cluster to perform various occlusion combination structure fitting, obtaining multiple occlusion structure confidence sets, including: Calculate the distances between other clustered laser data clusters and the cluster centers of multiple low-confidence clustered laser data clusters, and select the clustered laser data cluster with the smallest distance from each low-confidence clustered laser data cluster to obtain a combination of multiple clustered laser data clusters; Obtain multiple pre-set shielding combinations within the construction site; Multiple preset occlusion combinations are used, and iterative fitting is performed within multiple clustered laser data clusters to obtain multiple sets of occlusion structure confidence. Multiple matching occlusion combinations are obtained through screening. These are then combined with multiple matching construction structures to perform BIM modeling and obtain the modeling results.

2. The construction site BIM modeling method based on three-dimensional laser scanning according to claim 1, characterized in that, Within the construction site, 3D laser scanning was performed to obtain 3D laser data. Laser data clustering was then performed to obtain multiple clustered laser data clusters, including: Within the construction site, a three-dimensional laser scan is performed to obtain three-dimensional laser data, which includes multiple laser point data, each laser point data including three-dimensional spatial coordinates; Clustering optimization is performed on the three-dimensional laser data to obtain multiple clustered laser data clusters.

3. The construction site BIM modeling method based on three-dimensional laser scanning according to claim 2, characterized in that, Clustering optimization is performed on the three-dimensional laser data to obtain multiple clustered laser data clusters, including: Obtain the total number of structures within the construction site and set it as the number of cluster centers; Within the three-dimensional laser data, the number of laser point data points of the specified number of cluster centers are randomly selected as multiple first cluster centers, and the distances between other laser point data points and the multiple first cluster centers are calculated to obtain multiple sets of first cluster distances; Other laser point data are classified into laser data clusters with the smallest first cluster center within multiple first cluster distance sets, thus obtaining multiple first laser data clusters; Multiple cluster center selection and cluster optimization processes are performed until convergence. The laser data clusters with the smallest average distance during the clustering process are retained as multiple clustered laser data clusters.

4. The construction site BIM modeling method based on three-dimensional laser scanning according to claim 1, characterized in that, Based on various pre-designed construction structures at the construction site, iterative fitting is performed on multiple clustered laser data clusters to obtain multiple sets of confidence scores for the fitted structures. Then, multiple matching construction structures are selected, including: Obtain multiple pre-designed construction structures at the construction site; Based on various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the confidence set of the first fitted structure. Continue iterative fitting within multiple other clustered laser data clusters to obtain multiple sets of fitting structure confidence scores; Each of the multiple sets of fitted structure confidence scores selects the preset construction structure corresponding to the maximum fitted structure confidence score as multiple matching construction structures.

5. The construction site BIM modeling method based on three-dimensional laser scanning according to claim 4, characterized in that, Based on various preset construction structures, iterative fitting is performed within the first cluster of laser data to obtain the first set of confidence scores for the fitted structure, including: The first preset construction structure is randomly fitted multiple times within the first cluster of laser data. The proportion of laser point data within the first cluster of laser data that are less than the preset distance threshold from the first preset construction structure is calculated to obtain the first structure fitting accuracy set. The maximum value within the first set of structure fitting accuracy is selected as the first fitting structure confidence of the first preset construction structure. The other preset construction structures are used to perform multiple random fittings within the first cluster of laser data to obtain the confidence set of the first fitted structure.

6. The construction site BIM modeling method based on three-dimensional laser scanning according to claim 1, characterized in that, Multiple matching occlusion combinations were obtained through screening. These were then combined with multiple matching construction structures to perform BIM modeling, yielding the following modeling results: Within multiple sets of occlusion structure confidence scores, preset occlusion combination structures corresponding to the maximum occlusion structure confidence scores are selected as multiple matching occlusion combination structures. Based on multiple matching construction structures and multiple matching occlusion combination structures, BIM modeling is performed to obtain the modeling results. Among these, there are multiple duplicate matching construction structures within the multiple matching construction structures and multiple matching occlusion combination structures.

7. A construction site BIM modeling system based on 3D laser scanning, characterized in that, The method for implementing the construction site BIM modeling method based on three-dimensional laser scanning as described in any one of claims 1-6 includes: The laser data clustering module is used to perform 3D laser scanning at the construction site to obtain 3D laser data, and then perform laser data clustering to obtain multiple clustered laser data clusters. The structure fitting and matching module is used to iteratively fit multiple clustered laser data clusters based on various preset construction structures at the construction site, obtain multiple sets of confidence scores for fitted structures, and filter multiple matching construction structures. The low-confidence screening module is used to filter and obtain multiple low-confidence clustered laser data clusters by combining the construction progress information at the construction site with the low-confidence ratio of multiple fitted structure confidence sets. The occlusion modeling integration module is used to combine multiple low-confidence clustered laser data clusters with the nearest clustered laser data clusters to fit various preset occlusion combination structures, obtain multiple occlusion structure confidence sets, filter to obtain multiple matching occlusion combination structures, combine multiple matching construction structures, perform BIM modeling, and obtain modeling results.

Images