A Data-Based Method and System for Constructing BIM Models for High-Speed ​​Engineering

Abstract

This invention relates to the field of data processing, and more specifically, to a method and system for constructing high-speed engineering BIM models based on data analysis. The method includes: collecting data including point clouds of a BIM design model, point clouds of the construction site, and GIS geographic information of the project area; calculating the dynamic weight of each point cloud at the construction site; constructing a weighted ICP objective function using the dynamic weights of all point clouds at the construction site, and then performing precise registration between the point clouds of the BIM design model and the point clouds of the construction site to obtain a precisely registered BIM design model; and fusing the precisely registered BIM design model with the GIS geographic information to achieve multi-source data collaboration and optimization of the model. The dynamic weighting mechanism constructed in this invention can more accurately measure the importance of each point cloud in the registration process, effectively distinguishing between BIM design entities representing permanent structures and a large number of temporary objects without corresponding BIM semantics.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method and system for constructing high-speed engineering BIM models based on data analysis. Background Technology

[0002] Highway engineering faces challenges such as difficulties in land acquisition and resettlement coordination, numerous technical interfaces, and long management chains. The root cause lies in data barriers and information silos between planning, geology, design, and construction stages. Constructing high-precision BIM (Building Information Modeling) models and integrating GIS (Geographic Information System) technology is a crucial step in breaking down data barriers between different disciplines, achieving information integration and collaborative analysis, and providing a digital foundation for solving these problems.

[0003] Currently, one of the key technologies for achieving BIM and GIS integration is high-precision spatial registration, which involves accurately aligning the BIM design model with the actual geospatial data collected at the construction site. Existing methods mainly rely on geometric coordinate transformation or point cloud-based registration algorithms. However, in real-world scenarios such as highways with complex terrain, linear distribution, and rapidly changing construction dynamics, these traditional methods face significant challenges. For example, the BIM design model cannot encompass temporary objects that dynamically appear at the construction site, making traditional geometric registration algorithms highly susceptible to being misled by these "non-design" features in the GIS environment, resulting in erroneous positioning results. Summary of the Invention

[0004] To address the technical problem of difficulty in distinguishing between BIM design entities representing permanent structures and a large number of temporary objects without corresponding BIM semantics in the point cloud of a construction site, which leads to registration benchmark errors, this invention provides solutions in the following aspects.

[0005] In the first aspect, the data analysis-based method for constructing high-speed engineering BIM models includes:

[0006] Data including point clouds of BIM design models, point clouds of construction sites, and GIS geographic information of the project area are collected, and the collected data is preprocessed.

[0007] For each point cloud at the construction site, calculate the dynamic weight of each point cloud;

[0008] By utilizing the dynamic weights of all point clouds at the construction site, a weighted ICP objective function is constructed, and then the point cloud of the BIM design model is precisely registered with the point cloud of the construction site to obtain the precisely registered BIM design model.

[0009] By integrating the precisely registered BIM design model with GIS geographic information, an integrated model is constructed to achieve multi-source data collaboration and optimization of the model, providing data support for engineering decision-making.

[0010] The process of obtaining the dynamic weights includes:

[0011] The semantic consistency score, terrain adaptability coefficient, and local density index of each point cloud at the construction site are calculated separately. The product of the difference between 1 and the semantic consistency score, the terrain adaptability coefficient, and the local density index is converted into an exponential term to obtain the dynamic weight of each point cloud at the construction site.

[0012] Preferably, the data acquisition further includes generating a design topographic benchmark covering the project area based on the BIM design model and GIS geographic information; the preprocessing includes filtering, downsampling and coarse registration of all acquired point clouds to unify all point clouds to the same coordinate system.

[0013] Preferably, the generation of the design terrain benchmark includes:

[0014] The design terrain elevation of the coordinates is calculated based on the input coordinates, and all the calculated design terrain elevations together constitute the design terrain benchmark. Among them, the elevation design modification amount of the input coordinates is extracted from the BIM design model, and the extracted elevation design modification amount is added to the original terrain elevation of the same coordinate position in the GIS geographic information to obtain the design terrain elevation of the coordinate.

[0015] The input coordinates include the horizontal coordinates of each point cloud at the construction site.

[0016] Preferably, the calculation process of the semantic consistency score includes:

[0017] Semantic segmentation is performed on the point cloud of the BIM design model to obtain multiple semantic category regions;

[0018] For each point cloud at the construction site, calculate the nearest distance from the point cloud to each semantic category region, and mark the nearest point of the point cloud in each semantic category region; calculate the angle between the normal vector of the point cloud and the normal vector of the corresponding nearest point in each semantic category region;

[0019] Calculate the normalized weight of each semantic category region based on the proportion of point cloud quantity in the BIM design model for each semantic category region;

[0020] The semantic consistency score is calculated by weighted average based on the nearest distance, the included angle, and the normalized weight.

[0021] Preferably, the calculation process of the terrain adaptability coefficient includes:

[0022] Calculate the standard deviation of the original terrain elevation for all coordinate locations within the project area, and obtain the minimum safety value based on this standard deviation; for each point cloud at the construction site, calculate the absolute value of the difference between its original terrain elevation and the design terrain elevation.

[0023] The ratio of the absolute value of the calculated difference to the minimum safe value is used as the terrain adaptability coefficient.

[0024] Preferably, the calculation process of the local density index includes:

[0025] For each point cloud in the pre-defined construction site, a set of neighboring points is set. The sum of the Euclidean distances from any point cloud to all neighboring points in the corresponding set of neighboring points is calculated. The reciprocal of the calculated sum of Euclidean distances is used as the local density index.

[0026] Preferably, the calculation process of the local density index includes:

[0027] Pre-determine the neighboring point set for each point cloud at the construction site, and calculate the centroid of the neighboring point set; use the reciprocal of the Euclidean distance between each point cloud and the centroid of its corresponding neighboring point set as the centroid offset term;

[0028] Calculate the average distance from each point cloud to all its neighboring points in its neighboring point set, and use it as the local average distance of the point cloud. For each neighboring point set of the point cloud, construct a mapping relationship with the point cloud coordinates as the independent variable and the local average distance as the dependent variable, and fit a linear function using the least squares method. Use the coefficient vector of the linear function as the density gradient vector. Use the reciprocal of the sum of the magnitude of the obtained density gradient vector and the preset parameters as the density gradient term.

[0029] The product of the calculated centroid offset term and density gradient term is used as the local density index.

[0030] Preferably, the semantic category region includes at least one of roads, bridges, tunnels, and slopes.

[0031] Secondly, a high-speed engineering BIM model construction system based on data analysis includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the high-speed engineering BIM model construction method based on data analysis described in any one of the claims is implemented.

[0032] The beneficial effects of this invention are:

[0033] This invention calculates dynamic weights by comprehensively considering the semantic consistency score, terrain adaptability coefficient, and local density index of the construction site point cloud, and constructs a weighted ICP objective function for fine registration. This dynamic weighting mechanism can more accurately measure the importance of each point cloud in the registration process, effectively distinguish between BIM design entities representing permanent structures and a large number of temporary objects without BIM semantic correspondence, avoid registration benchmark errors caused by interference from temporary objects, and thus significantly improve the registration accuracy between the BIM design model point cloud and the construction site point cloud. Attached Figure Description

[0034] Figure 1 This is a flowchart of steps S1-S4 in the data analysis-based high-speed engineering BIM model construction method of this invention.

[0035] Figure 2 This is a structural block diagram of the high-speed engineering BIM model construction system based on data analysis, according to an embodiment of the present invention. Detailed Implementation

[0036] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0037] This invention takes highway construction as an application scenario and describes the specific implementation process of a data analysis-based BIM model construction method and system for highway engineering.

[0038] Reference Figure 1 The data analysis-based method for constructing high-speed engineering BIM models includes steps S1-S4, as detailed below:

[0039] S1: Collect data including point clouds of BIM design models, point clouds of construction sites, and GIS geographic information of the project area, and preprocess the collected data.

[0040] In one embodiment, firstly, point cloud data of a highway BIM design model (representing the ideal state of the design) is exported from a computer using software such as Autodesk Revit. Each point has three-dimensional coordinates and contains detailed semantic information such as the road centerline and roadbed slopes, which helps to accurately compare the construction situation later.

[0041] In addition, point cloud data of the construction site (which represents the actual state of the construction site) is collected using a laser scanner or UAV-borne lidar system, with a sampling density of no less than 50 point cloud data per square meter to ensure that the complex terrain features and details of temporary objects are accurately reflected.

[0042] Next, it is necessary to obtain GIS geographic information data of the project area, especially the 1:500 scale digital elevation model. This model can return the original terrain elevation of any given horizontal coordinate, providing basic data for subsequent terrain-related calculations.

[0043] After the aforementioned data collection, since the raw data is coarse and misaligned, all collected point cloud data needs to be preprocessed. Examples include conventional filtering and denoising, downsampling (e.g., using octree spatial segmentation to downsample the point cloud data), and coarse registration (e.g., using a random sampling consistency algorithm based on fast point feature histograms to coarsely register the two sets of point cloud data to transform the point cloud of the BIM design model into an approximate coordinate system of the point cloud at the construction site). In addition, the GIS geographic information is unified to the same coordinate system to ensure consistency in spatial reference for all data.

[0044] It should be noted that direct comparison of the above BIM design model with on-site scans is somewhat difficult because there are various temporary objects and unprocessed original terrain on site.

[0045] Therefore, by generating a design topographic benchmark, a unified standard and reference framework is provided for all data. Secondly, it accurately reflects the expected topographic state to be achieved after construction according to the design drawings, including elevation information for each location. By comparing the actual topographic conditions at the construction site with the design topographic benchmark, it is possible to intuitively determine whether the construction meets the design requirements and promptly identify deviations and problems during the construction process.

[0046] For example, for each point cloud at the construction site, the elevation design modification amounts, such as fill height and cut depth, at the horizontal coordinate position of the point cloud are extracted from the BIM design model. These extracted elevation design modification amounts are then added to the original terrain elevation at the same coordinate position in the GIS geographic information to obtain the design terrain elevation at that point cloud's horizontal coordinate position. This design terrain elevation reflects the expected terrain state after construction according to the drawings and will serve as a geospatial reference for temporary object identification. Furthermore, through the above operations, all calculated design terrain elevations together constitute the design terrain benchmark to be generated.

[0047] It should be noted that the project area mentioned above refers to the geographical scope of the entire highway project, which includes all areas that have been constructed, are under construction, and have not yet been constructed. It is a comprehensive geographical scope definition. The construction site specifically refers to the dynamic area within the project area where construction activities are currently taking place. This area will change continuously as the project progresses. The construction site includes temporary objects such as construction machinery and temporary scaffolding.

[0048] In addition, the construction area refers to the static area within the project area that has been completed according to the design requirements. The elevation and geometry of these areas meet the design standards. Once the construction of a certain area is completed, it will change from a "construction site" to a "construction area".

[0049] S2: For each point cloud at the construction site, calculate the dynamic weight of each point cloud.

[0050] Traditional ICP (Iterative Closest Point) algorithms treat all point clouds equally, assuming each point contributes equally to the registration. However, in the context of highway construction sites, numerous temporary object points have no correspondence with the BIM design model. If these temporary points are included in the registration calculation equally, the algorithm will forcibly distort the transformation matrix to match these non-existent "pseudo-features," leading to a significant deviation from reality in the final registration result. Furthermore, temporary objects can be considered a large-scale, structured noise. Traditional ICP algorithms become unstable and inaccurate in the presence of significant noise.

[0051] To address the numerous interferences caused by temporary objects, a dynamic weight allocation mechanism was established. This mechanism quantifies the spatial proximity of the construction site point cloud to the semantic category of the BIM design model, the standardization deviation between the construction site point cloud and the design terrain elevation, and the spatial clustering of the construction site point cloud with its neighboring points. A comprehensive temporary object probability scoring function was constructed, and the scores were mapped to obtain the dynamic weight of each point cloud on the construction site.

[0052] The process of obtaining the dynamic weights of each point cloud at the construction site is as follows:

[0053] First, the semantic consistency score of each point cloud at the construction site is calculated to provide a semantic basis for subsequent temporary object identification, distinguishing between temporary objects and permanent structures.

[0054] In one embodiment, the BIM design model is semantically segmented into multiple semantic category regions, including categories such as roads, bridges, tunnels, and slopes. This step is to classify the BIM design model according to different structural types so that the spatial relationships between point clouds and these categories can be calculated separately in subsequent steps.

[0055] For each point cloud at the construction site, calculate its nearest distance to each semantic category region. That is, calculate the distance from the point cloud at the construction site to all points in the semantic category region and find the minimum value. This minimum value is the corresponding nearest distance, which reflects the proximity of the point cloud to the semantic category region.

[0056] Furthermore, the geometric center of each semantic category region is obtained, and the average distance from all points in each semantic category region to its geometric center is calculated as the average feature scale of the corresponding semantic category region.

[0057] Furthermore, the ratio of the nearest distance from each point cloud at the construction site to any semantic category region to the average feature scale of that semantic category region is used as a normalized distance, which takes into account the adaptability of structures at different scales.

[0058] Finally, the maximum value of the exponential decay function of the normalized distance from each point cloud at the construction site to all semantic category regions calculated above is used as the semantic consistency score of each point cloud at the construction site.

[0059] Specifically, when the nearest distance from the point cloud at the construction site to any semantic category region is less than the average feature scale of that semantic category region, it indicates that the point cloud is adjacent to the design structure, i.e., the semantic consistency score is close to 1, indicating that the point cloud highly matches a certain permanent structure in the BIM design model, and the point cloud is highly likely to belong to the actual engineering structure. Conversely, when the nearest distance from the point cloud at the construction site to any semantic category region is greater than the average feature scale of that semantic category region, it indicates that the point cloud is far from all design structures, i.e., the semantic consistency score is close to 0, indicating that the point cloud has a significant offset from all semantic category regions, and the point cloud is highly likely to belong to a temporary construction object or an unmodeled area.

[0060] In another embodiment, considering that semantic consistency scoring based on spatial distance cannot fully utilize the essential differences in surface geometry between temporary objects and permanent structures, the surface of temporary objects is usually more regular and flat, and its normal vector direction is significantly different from that of the design structure; while the normal vector of the surface of permanent structures is highly consistent with its BIM design model.

[0061] Following the same steps, after obtaining the normalized distances from each point cloud at the construction site to each semantic category region, the nearest point of each point cloud in each semantic category region is marked. The normal vector of each point cloud and the normal vector of its marked nearest point in each semantic category region are then obtained, and the angle between these two normal vectors is calculated. This angle quantifies the degree of difference in surface geometry between the point cloud at the construction site and its nearest point in the semantic category region. A small angle indicates that the surface orientations of the two are relatively consistent, while a large angle indicates significant surface differences.

[0062] Furthermore, the square of the cosine of the angle between the normal vector of each point cloud and the normal vector of its nearest labeled point in each semantic category region is calculated, mapping all values ​​to the interval [0, 1], so that the quantization of directional differences depends only on the angle itself, not the direction. For example:

[0063] When the included angle is 0°, it means that the direction of the point cloud is completely consistent with that of its nearest point, and the square of the corresponding cosine value is equal to 1; when the included angle is 90°, it means that the direction of the point cloud is perpendicular to that of its nearest point, and the corresponding cosine value is equal to 0.

[0064] Furthermore, considering the importance ratio of each semantic category region in the overall project, the normalized weight of each semantic category region is calculated.

[0065] The number of point clouds in each semantic category region in the BIM design model is counted. Then, the number of point clouds contained in each semantic category is divided by the sum of the number of point clouds contained in all semantic category regions. The result is used as the normalization weight of each semantic category region.

[0066] Furthermore, the semantic consistency score is calculated by combining the factors obtained from the above calculations. The specific calculation formula is as follows:

[0067]

[0068] In the formula, Point clouds at the construction site Semantic consistency score, semantic category region Normalized weights, The total number of semantic category regions. For point clouds To semantic category region Normalized distance, It is an exponential function with the natural number e as its base. For point clouds To semantic category region The angle between the nearest points, It is a cosine function.

[0069] Among them, when point cloud When adjacent to the design structure and with the same surface orientation Approaching 1 indicates a point cloud Not only does it closely match a permanent structure in the BIM design model in terms of spatial location, but its surface geometry also conforms to the design structure, point cloud. It is highly likely that it belongs to a real engineering structure; conversely, when point clouds... When far from the design structure and the surface orientation is mismatched Approaching 0 indicates a point cloud. Point clouds are either spatially distant from all semantic categories, or their surface geometry is significantly inconsistent with the design structure. It is highly likely that it belongs to a temporary construction object or an area that has not been modeled.

[0070] Secondly, the terrain adaptability of each point cloud at the construction site is calculated to quantify the degree of conformity between each point cloud at the construction site and the designed terrain, providing a key geospatial discrimination basis for the identification of temporary objects.

[0071] In one embodiment, it is first ensured that there is a suitable minimum safety value under various terrain conditions to avoid numerical instability in subsequent calculations. For example, the standard deviation of the original terrain elevation of all coordinate locations within the project area is obtained, this standard deviation is multiplied by a factor such as 1.5, and then compared with a safety lower limit such as 0.1. The larger value between the two is taken as the minimum safety value.

[0072] Then, for each point cloud at the construction site, the absolute value of the difference between its original terrain elevation and the design terrain elevation is calculated, and the ratio of the calculated absolute value of the difference to the minimum safety value is used as the terrain adaptability coefficient of each point cloud at the construction site.

[0073] When the terrain adaptability coefficient approaches 0, it indicates that the point cloud at the construction site matches the design elevation well, and the area is more likely to be a permanent structure area constructed according to the design or a terrain area that meets the design requirements. Conversely, when the terrain adaptability coefficient is much greater than 0, it indicates that the point cloud at the construction site deviates significantly from the design elevation, and the area is highly likely to be a temporary construction object, an unconstructed area, or an area with abnormal construction quality.

[0074] Next, considering the local spatial distribution characteristics of point clouds, the local density index of each point cloud at the construction site is calculated.

[0075] In one embodiment, a neighboring point set is preset for each point cloud at the construction site, for example, obtaining 30 neighboring point clouds of any point cloud as its neighboring point set.

[0076] Calculate the sum of Euclidean distances from any point cloud to all neighboring points in the corresponding neighboring point set, and use the reciprocal of the calculated sum of Euclidean distances as the local density index of the point cloud.

[0077] The larger the local density index, the higher the point cloud belongs to a high-density area, such as the surface of a temporary object; conversely, the smaller the local density index, the lower the point cloud belongs to a low-density or irregular area, such as natural terrain or permanent structures.

[0078] In another embodiment, the same set of neighboring points as described above is still set, and the centroid of the neighboring point set is calculated. The reciprocal of the Euclidean distance between each point cloud at the construction site and the centroid of its corresponding neighboring point set is used as the centroid offset term.

[0079] Furthermore, the mean distance from each point cloud to all neighboring points in its neighboring point set is calculated as the local average distance of the point cloud. For each neighboring point set of a point cloud, a mapping relationship is constructed with the point cloud coordinates as the independent variable and the local average distance as the dependent variable. The least squares method is used to fit a linear function, and the coefficient vector of the linear function is used as the density gradient vector. The reciprocal of the sum of the magnitude of the obtained density gradient vector and a preset parameter (set to 1 for example to prevent the denominator from being 0) is used as the density gradient term.

[0080] The product of the centroid offset term and the density gradient term obtained from the above calculations is used as the local density index of each point cloud at the construction site.

[0081] The calculated local density index is a composite index. When the centroid offset is small (large centroid offset term) and the edge abrupt change is significant (small density gradient term), the product results in a moderate local density index, corresponding to temporary object surfaces. Conversely, when the centroid offset is large or excessively flat, the product results in a local density index approaching 0, corresponding to permanent structures or natural terrain. By balancing the two influences through a product, misjudgment caused by a single feature is avoided.

[0082] Finally, the local density index is fused with the semantic consistency score and terrain adaptability coefficient to form a three-dimensional feature space:

[0083] Semantic consistency scoring: Verifies whether an object conforms to the definition of a temporary object;

[0084] Terrain adaptability coefficient: determines whether an object is adapted to the current terrain;

[0085] Local density index: quantifies the geometric regularity (and edge abruptness) of an object's surface.

[0086] In one embodiment, the exponential decay value of the product of the difference between 1 and the semantic consistency score, the terrain adaptability coefficient, and the local density index is used as the dynamic weight of each point cloud at the construction site.

[0087] Only when the three conditions of high density, low semantic consistency, and large terrain deviation are met simultaneously will the result of the above product increase significantly, indicating that a certain point cloud at the construction site is likely to be a temporary object. This causes the exponential decay value of the product to approach 0, that is, the corresponding dynamic weight approaches 0, and the point cloud hardly participates in the registration. Conversely, the larger the dynamic weight, the more the point cloud basically participates in the registration.

[0088] S3: Utilize the dynamic weights of all point clouds at the construction site to construct a weighted ICP objective function, and then perform fine registration between the point cloud of the BIM design model and the point cloud of the construction site to obtain the finely registered BIM design model.

[0089] Using the dynamic weights of all point clouds at the construction site obtained in S2 above, a weighted ICP objective function is constructed, and the distance error of each point pair is differentiated:

[0090]

[0091] In the formula, It is a 3×3 rotation matrix. It is a 3×1 translation vector. This represents the total number of point clouds at the construction site. For point clouds Dynamic weights, To be the minimum value, For point clouds The closest point in the BIM design model (preferably the one with the highest semantic consistency score).

[0092] Then, the optimal rotation matrix and translation vector are obtained through convergence. The optimal rotation matrix and translation vector are applied to the BIM design model to obtain the finely registered BIM design model.

[0093] The BIM design model that has been precisely registered is highly consistent with the actual situation on the construction site, providing construction personnel with accurate construction basis, ensuring that construction operations are carried out strictly in accordance with design requirements, reducing construction errors, and improving construction quality.

[0094] The objective function described above has the same form as the standard least squares function, except that dynamic weights are introduced before the error term.

[0095] S4: Integrate the precisely registered BIM design model with GIS geographic information to build an integrated model, realize multi-source data collaboration and optimization of the model, and provide data support for engineering decision-making.

[0096] The original BIM design model may deviate from the actual situation on the construction site. By using temporary object suppression to perform precise registration processing on the BIM design model, it can more accurately reflect the building's design information, remove interference from temporary objects, and make the model more consistent with reality.

[0097] The precise registered BIM design model is deeply integrated with GIS terrain information. In other words, the design information of the highway building is integrated with the actual terrain and geographic information into a unified model, forming a comprehensive three-dimensional model that includes the building design and the surrounding terrain environment, providing basic data support for subsequent collaborative work.

[0098] Based on the fused data, a three-dimensional collaborative management integrated model is built to achieve multi-source data collaboration and optimization. This integrated model creates a common working environment for multiple roles, including designers, construction companies, and supervisors, enabling them to carry out various related tasks. A unified coordinate system is adopted to ensure that all parties, including designers, construction companies, and supervisors, operate within the same spatial reference frame, effectively avoiding information misalignment and communication barriers caused by inconsistent coordinates, and ensuring that all parties reach a consensus on spatial information such as building location and dimensions.

[0099] This completes the data analysis-based BIM model construction method for high-speed engineering projects. Subsequent engineering decisions can be made based on the constructed model.

[0100] This invention also provides a high-speed engineering BIM model building system based on data analysis. For example... Figure 2 As shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the high-speed engineering BIM model construction method based on data analysis according to the first aspect of the present invention.

[0101] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0102] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A high-speed engineering BIM model construction method based on data analysis, characterized in that, The application relates to a method for constructing a BIM (Building Information Modeling) model based on multi-source data. The method comprises the following steps: collecting data including point clouds of a BIM design model, point clouds of a construction site and GIS geographic information of a project area, and pre-processing the collected data; calculating a dynamic weight of each point cloud of the construction site; constructing a weighted ICP (Iterative Closest Point) target function by using the dynamic weights of all point clouds of the construction site, and then performing fine registration on the point clouds of the BIM design model and the point clouds of the construction site to obtain a fine-registered BIM design model; fusing the fine-registered BIM design model and the GIS geographic information to construct an integrated model, realizing multi-source data collaboration and optimization of the model and providing data support for engineering decision-making; wherein the process of obtaining the dynamic weight comprises: calculating a semantic consistency score, a terrain adaptability coefficient and a local density index of each point cloud of the construction site respectively, converting the product of the difference between 1 and the semantic consistency score, the terrain adaptability coefficient and the local density index into an exponential term to obtain the dynamic weight of each point cloud of the construction site; the process of calculating the semantic consistency score comprises: performing semantic segmentation on the point clouds of the BIM design model to obtain a plurality of semantic category regions; for each point cloud of the construction site, calculating the nearest distance of the point cloud to each semantic category region and marking the nearest point of the point cloud in each semantic category region; calculating the included angle between the normal vector of the point cloud and the normal vector of the corresponding nearest point in each semantic category region; calculating the normalized weight of each semantic category region according to the point cloud quantity proportion of each semantic category region in the BIM design model; based on the nearest distance, the included angle and the normalized weight, the semantic consistency score is calculated by weighted average; the process of calculating the terrain adaptability coefficient comprises: calculating the standard deviation of the original terrain elevation of all coordinate positions in the project area, and obtaining a minimum safety value according to the standard deviation; for each point cloud of the construction site, calculating the absolute value of the difference between the original terrain elevation and the design terrain elevation; taking the ratio of the calculated absolute value of the difference and the minimum safety value as the terrain adaptability coefficient; the process of calculating the local density index comprises:

2. The data analysis based high speed engineering BIM model building method according to claim 1, characterized in that, presetting a neighboring point set of each point cloud of the construction site, calculating the sum of the Euclidean distances from any point cloud to all neighboring points in the neighboring point set corresponding to the point cloud, and taking the reciprocal of the calculated sum of the Euclidean distances as the local density index.

3. The data analysis based high speed engineering BIM model building method of claim 2, wherein, The collection further comprises generating a design terrain reference covering the project area based on the BIM design model and the GIS geographic information; the pre-processing comprises filtering, down-sampling and coarse registration processing of all collected point clouds, and unifying all point clouds to the same coordinate system. The generation of the design terrain reference comprises: The design terrain elevation of the input coordinate is calculated, and all the calculated design terrain elevations constitute the design terrain reference; wherein, the elevation design modification amount of the input coordinate is extracted from the BIM design model, the extracted elevation design modification amount is added to the original terrain elevation of the same coordinate position in the GIS geographic information to obtain the design terrain elevation of the coordinate; the input coordinate includes the horizontal coordinate of each point cloud of the construction site.

4. The data analysis based high speed engineering BIM model building method of claim 1, wherein, The calculation process of the local density index includes: A preset adjacent point set of each point cloud of the construction site is calculated, and the centroid of the adjacent point set is calculated; the reciprocal of the Euclidean distance between each point cloud and the centroid of the adjacent point set corresponding to the point cloud is taken as the centroid offset term; The average distance of each point cloud to all adjacent points in the adjacent point set is calculated as the local average distance of the point cloud, for each adjacent point set of the point cloud, a mapping relationship is constructed with the point cloud coordinate as the independent variable and the local average distance as the dependent variable, and a linear function is fitted by using the least square method, and the coefficient vector of the linear function is taken as the density gradient vector; the reciprocal of the sum of the modulus length of the obtained density gradient vector and the preset parameter is taken as the density gradient term; The product of the calculated centroid offset term and the density gradient term is taken as the local density index.

5. The data analysis based high speed engineering BIM model building method according to claim 1, wherein, The semantic category area includes at least one of a road, a bridge, a tunnel, and a slope.

6. A high speed engineering BIM model building system based on data analysis characterized in that, It includes: A processor and a memory, the memory stores computer program instructions, when the computer program instructions are executed by the processor, the method for constructing a high-speed engineering BIM model based on data analysis according to any one of claims 1-5 is realized.

Images