A bridge BIM reverse modeling method, system and device driven by a primitive library
By employing a primitive library-driven reverse modeling method for bridge BIM, and utilizing deep learning networks and primitive library matching technology, the problem of low efficiency in bridge BIM model construction is solved, enabling rapid and efficient bridge BIM model construction and updating.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA STATE RAILWAY GRP CO LTD
- Filing Date
- 2022-11-15
- Publication Date
- 2026-05-22
AI Technical Summary
In existing technologies, bridge BIM model building is inefficient, making it difficult to efficiently handle the various types and quantities of engineering components in large-scale engineering projects. Furthermore, the noise, irregularity, and occlusion properties of point cloud data make model acquisition difficult, and manual modeling makes it hard to achieve efficient updates.
A primitive library-driven bridge BIM reverse modeling method is adopted. Point cloud data is acquired through radar scanning, substructure diagrams are segmented using a deep learning network, and reference models are matched in the primitive library. Parameter values are adjusted to quickly construct a bridge BIM model.
It improves the efficiency of bridge BIM model building, ensures image accuracy while reducing modeling time, and enables rapid model updates and efficient management.
Smart Images
Figure CN115859422B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of model building technology, and in particular to a primitive library-driven method, system and device for reverse modeling bridge BIM. Background Technology
[0002] In recent years, with the development of data acquisition equipment and technology, laser scanning and oblique photogrammetry technologies can directly or indirectly and quickly acquire three-dimensional point cloud data of large areas, and have gradually become important data sources for the construction of digital cities and three-dimensional geographic information systems.
[0003] However, in my country's current large-scale engineering construction, such as high-speed railways and highways, the construction status is mostly statistically analyzed based on on-site construction logs. For the overall project, only manually created component BIM models based on design drawings exist. Due to the diverse types and large quantities of engineering component models, and their variability with the project, although digital twin modeling technology can obtain high-precision x, y, z measurements and color information of objects from oblique photogrammetry or Light Detection and Ranging (LiDAR) point clouds, point cloud data suffers from noise, surface irregularities, and occlusion. In actual engineering, obtaining complete model point clouds is difficult, and the structural differences between components in different areas are also significant. Therefore, actual component models in the construction field are mostly manually created individual models, which are difficult to process efficiently. Summary of the Invention
[0004] In view of this, embodiments of the present invention provide a primitive library-driven bridge BIM reverse modeling method to eliminate or improve one or more defects existing in the prior art.
[0005] One aspect of the present invention provides a primitive library-driven bridge BIM reverse modeling method, the method comprising the following steps:
[0006] The system receives a bridge point cloud structure map acquired by radar scanning, inputs the bridge point cloud structure map into a preset deep learning network model, and the deep learning network model divides the bridge point cloud structure map into multiple sub-structure maps.
[0007] Based on the point cloud computing structure parameters in the substructure graph, the substructure graph is matched in a preset primitive library based on the structure parameters. The reference model corresponding to the substructure graph is obtained by matching in the primitive library. The primitive library includes a variety of components, and each component has multiple preset reference models.
[0008] Obtain the preset parameter type to be modified in the matched reference model, calculate the actual parameter value of the parameter type to be modified based on the position of each point cloud in the substructure diagram, modify the parameter value of the parameter type to be modified in the reference model to the corresponding actual parameter value, and obtain the actual sub-model for each substructure diagram.
[0009] By adopting the above scheme, this scheme uses a pre-defined primitive library containing multiple reference models to match the substructure diagram with the reference models. The reference models are pre-drawn models, and each reference model has a preset parameter type to be modified. By modifying the parameter values of the parameter type to be modified in the reference model to the actual parameter values calculated by the corresponding substructure diagram, the reference model can be quickly modified into the actual sub-model corresponding to the substructure diagram. This scheme improves the efficiency of model construction by only matching the substructure diagram with the reference model and modifying the corresponding parameter values while ensuring image accuracy.
[0010] In some embodiments of the present invention, the method further includes the following steps:
[0011] Obtain the actual sub-model for each substructure diagram;
[0012] Based on the position of the substructure diagram in the bridge point cloud structure diagram, multiple actual sub-models are combined to obtain the actual model corresponding to the bridge point cloud structure diagram.
[0013] In some embodiments of the present invention, the reference model is preset with contact constraints, which are used to mark the contact surfaces of the reference model. In the step of combining multiple actual sub-models based on the position of the sub-structure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram, the contact surfaces of the corresponding actual sub-models are marked based on the contact surfaces of the reference model. In the process of combining multiple actual sub-models into the actual model, the contact surfaces of the actual sub-models are made to contact each other.
[0014] In some embodiments of the present invention, the step of the deep learning network model dividing the bridge point cloud structure map into multiple sub-structure maps further includes labeling the component category to which each sub-structure map belongs based on a preset classifier in the deep learning network model.
[0015] In some embodiments of the present invention, the primitive library pre-defines component categories corresponding to each reference model, and the step of matching the substructure diagram in the pre-define primitive library based on structural parameters includes...
[0016] Based on the component category of the substructure graph output by the classifier, the corresponding category is matched in the primitive library, and multiple reference models of the corresponding category are obtained;
[0017] Based on structural parameters, the substructure graph is matched among multiple reference models in the corresponding category of the preset primitive library.
[0018] In some embodiments of the present invention, in the step of the deep learning network model dividing the bridge point cloud structure map into multiple sub-structure maps and labeling the component category to which each sub-structure map belongs, the deep learning network model divides the multiple point clouds in the bridge point cloud structure map into multiple regions, the multiple point clouds in each region constitute a sub-structure map, and determines the category to which the sub-structure map belongs based on a pre-trained classifier.
[0019] In some embodiments of the present invention, the step of receiving the bridge point cloud structure map obtained by radar scanning further includes filtering and noise reduction processing on the received bridge point cloud structure map.
[0020] This invention also provides a primitive library-driven bridge BIM reverse modeling system, the system comprising:
[0021] The point cloud acquisition module is used to receive the bridge point cloud structure map acquired by radar scanning, input the bridge point cloud structure map into a preset deep learning network model, and the deep learning network model divides the bridge point cloud structure map into multiple sub-structure maps.
[0022] The model matching module is used to match the substructure graph with the point cloud computing structure parameters based on the structure parameters in the substructure graph, and to obtain the reference model corresponding to the substructure graph in the primitive library. The primitive library includes a variety of components, and each component has multiple preset reference models.
[0023] The model generation module is used to obtain the preset parameter types to be modified in the matched reference model, calculate the actual parameter values of the parameter types to be modified based on the positions of each point cloud in the substructure diagram, modify the parameter values of the parameter types to be modified in the reference model to the corresponding actual parameter values, and obtain the actual sub-model for each substructure diagram.
[0024] In some embodiments of the present invention, the system further includes a model building module, which is used to obtain the actual sub-model of each substructure diagram; and to combine multiple actual sub-models based on the position of the substructure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram.
[0025] The present invention also provides a primitive library-driven bridge BIM reverse modeling device, which includes a computer device, the computer device including a processor and a memory, the memory storing computer instructions, the processor executing the computer instructions stored in the memory, and when the computer instructions are executed by the processor, the device implements the steps of the method described above.
[0026] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the text, or may be learned by practice of the invention. The objects and other advantages of the invention will become apparent from the description and the accompanying drawings.
[0027] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description
[0028] The accompanying drawings, which are provided to further illustrate the invention and form part of this application, are not intended to limit the scope of the invention.
[0029] Figure 1 This is a schematic diagram of the first embodiment of the bridge BIM reverse modeling method driven by the primitive library of the present invention;
[0030] Figure 2 This is a schematic diagram of the second embodiment of the bridge BIM reverse modeling method driven by the primitive library of the present invention;
[0031] Figure 3 This is a schematic diagram of the third implementation of the primitive library-driven bridge BIM reverse modeling method of the present invention;
[0032] Figure 4 A schematic diagram illustrating one implementation method for constructing a primitive library;
[0033] Figure 5 A schematic diagram illustrating one implementation of the steps for matching substructure diagrams;
[0034] Figure 6 A schematic diagram illustrating one implementation method for constructing an actual model;
[0035] Figure 7 This is a schematic diagram of one implementation of the bridge BIM reverse modeling system driven by the primitive library of the present invention. Detailed Implementation
[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.
[0037] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.
[0038] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.
[0039] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.
[0040] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.
[0041] To solve the above problems, such as Figure 1 , 3 As shown, this invention proposes a primitive library-driven bridge BIM reverse modeling method, the steps of which include:
[0042] Step S100: Receive the bridge point cloud structure map obtained by radar scanning, input the bridge point cloud structure map into a preset deep learning network model, and the deep learning network model divides the bridge point cloud structure map into multiple sub-structure maps.
[0043] In some embodiments of the present invention, the bridge point cloud structure diagram is composed of multiple three-dimensional point clouds obtained by radar scanning.
[0044] In some embodiments of the present invention, the deep learning network model divides the bridge point cloud structure diagram into multiple parts, and the multiple point clouds of each part constitute a sub-structure diagram.
[0045] In some embodiments of the present invention, the deep learning network model may be a PointNet model, and the deep learning network model may use a multilayer perceptron (MLP) as a classifier to identify the category of the substructure graph.
[0046] Step S200: Based on the point cloud computing structure parameters in the substructure graph, the substructure graph is matched in a preset primitive library based on the structure parameters. The reference model corresponding to the substructure graph is obtained by matching in the primitive library. The primitive library includes a variety of components, and each component has multiple preset reference models.
[0047] In some embodiments of the present invention, in the step of matching the substructure diagram in a preset primitive library based on the point cloud structure parameters in the substructure diagram, the substructure diagram is divided into multiple comparison layers according to the height parameters of the point cloud, and the comparison layers are compared one by one to obtain the reference model corresponding to the substructure diagram.
[0048] In the specific implementation process, if the classifier determines that the substructure diagram belongs to the pier type, then a comparison is made under the category of pier component. If the pier type in the primitive library includes two reference models, hollow pier and solid pier, then the substructure diagram is compared with the two reference models, respectively. If there is a point cloud at the center position in the comparison layer of the substructure diagram, then it is a hollow pier; if there is no point cloud at the center position in the comparison layer of the substructure diagram, then it is a solid pier.
[0049] In the specific implementation process, taking the bridge pier example, the slope of the pier body and the thickness of the left and right bearing plates of most solid piers can be adjusted by parameters. For some bridge piers with maintenance towers, there may be hollow structures in the non-central part of the pier body or pier cap, resulting in discontinuities in the sliced layers of point cloud data at this location. For bearing plates distributed along the center line of the beam joint, there will be no difference in the thickness of the bearing plates on both sides. During the matching process of the primitive library, the outer contour of each layer of data is obtained through the point cloud of each comparison layer using the contour extraction method. The continuity of the contour surface is judged based on the regional density or curvature to identify hollow and solid bridge piers.
[0050] In the specific implementation process, by judging the surface continuity of the outer surface contour of the single-layer point cloud data of the slice layer, it is determined whether there are gaps or discontinuities in the middle of the pier body or the non-center position of the single layer of the pier cap, thereby determining whether it is an ordinary solid pier or a maintenance tower pier, and then matching the pier model in the primitive library according to the distribution of the pad stone.
[0051] In the specific implementation process, for the implementation of bridge beams, the elevation changes of the point cloud data of the slice layer corresponding to the bottom of the beam can be extracted and compared to determine whether the bridge type is a simply supported beam or a continuous beam. The width of the wing plate of the box girder can be determined based on the width of the points on both sides of the bridge, and then the basic model of the bridge can be matched.
[0052] In practice, the bridge includes piers, beams, pile foundations, abutments, power lines, and other components.
[0053] While beam bridges employing the above approach exhibit diverse structures, they all consist of piers and beams. However, piers and beams possess various structural forms, such as continuous beams, simply supported beams, and hyperbolic piers with slopes. Furthermore, beams or piers of the same type can exhibit different shape characteristics, dimensions, and edge structures. Therefore, it is necessary to differentiate between these different structures, identify shapes that cannot be represented by the same type, and establish different primitive models. This approach aims to reduce workload while accommodating model structures across various scenarios. For instance, piers with and without slopes can both be represented by the same pier body primitive model; the only difference lies in whether the dimensions of the two cross-sections constituting the pier are identical.
[0054] Step S300: Obtain the preset parameter type to be modified in the matched reference model, calculate the actual parameter value of the parameter type to be modified based on the position of each point cloud in the substructure diagram, modify the parameter value of the parameter type to be modified in the reference model to the corresponding actual parameter value, and obtain the actual sub-model corresponding to each substructure diagram.
[0055] In practice, the parameters to be modified can be height, slope, and diameter of the cylinder cross-section, etc.
[0056] In the specific implementation process, if the preset parameter type to be modified for the matched solid bridge pier is slope angle, then the slope angle is calculated based on the points in the substructure diagram, and the original slope angle value in the reference model is replaced with the calculated slope angle value.
[0057] Using the above scheme and a model-driven approach, a cross-sectional model with BIM semantics is created based on the stable cross-section. Parameters are established as constraints to control the size of the cross-section. The cross-sections are arranged according to certain rules and characteristics in one direction. By adjusting the parameters of the cross-sections, the shape of the component primitive cross-sections is constrained, and longitudinal parameters are added to control the longitudinal dimensions of the primitive model, thereby fitting a single BIM structural primitive. One or more structural primitive models are combined to obtain a parametric BIM component primitive model with structural consistency. The structure of the component model is decomposed, and each decomposed single structure can be composed of a specific cross-section arranged and combined in a certain direction. The cross-sections are linearly distributed in the vertical direction, and the basic structure of the cross-sections is consistent, differing only in parameter dimensions. This specific cross-section that can form a structural model can be used as a basic structural unit to establish components or component primitives.
[0058] If the above solution is adopted, such as Figure 4As shown, this solution uses a pre-defined primitive library containing multiple reference models to match the substructure diagram with the reference models. The reference models are pre-drawn models, and each reference model has a preset parameter type to be modified. By modifying the parameter values of the parameter type to be modified in the reference model to the actual parameter values calculated by the corresponding substructure diagram, the reference model can be quickly modified into the actual sub-model corresponding to the substructure diagram. This solution improves model construction efficiency by only matching the substructure diagram with the reference model and modifying the corresponding parameter values while ensuring image accuracy.
[0059] like Figure 2 As shown, in some embodiments of the present invention, the method further includes the following steps:
[0060] Obtain the actual sub-model for each substructure diagram;
[0061] Based on the position of the substructure diagram in the bridge point cloud structure diagram, multiple actual sub-models are combined to obtain the actual model corresponding to the bridge point cloud structure diagram.
[0062] like Figure 6 As shown, in some embodiments of the present invention, the position of the substructure diagram in the bridge point cloud structure diagram is determined based on the position of each point cloud in the substructure diagram. The point clouds in the substructure diagram are sorted according to height, and point clouds with heights ranging from the lowest height to a preset height range above the lowest height are extracted. The average coordinate value of the point clouds within the range is calculated, and the point corresponding to the average value of the horizontal, vertical, and longitudinal coordinates is taken as the position point of the substructure diagram.
[0063] In other embodiments of the present invention, a reference model that matches the actual sub-model has a preset installation position, and the actual sub-model is combined based on the installation position of the reference model to obtain the actual model.
[0064] In some embodiments of the present invention, the reference model is preset with contact constraints, which are used to mark the contact surfaces of the reference model. In the step of combining multiple actual sub-models based on the position of the sub-structure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram, the contact surfaces of the corresponding actual sub-models are marked based on the contact surfaces of the reference model. In the process of combining multiple actual sub-models into the actual model, the contact surfaces of the actual sub-models are made to contact each other.
[0065] In the specific implementation process, if the reference model is a solid bridge pier, the corresponding preset contact surface includes a contact surface for contacting the bridge. The bridge reference model is also correspondingly provided with a contact surface for contacting the solid bridge pier. Preferably, each contact surface is assigned a number, and the numbers of the corresponding contact surfaces are preset to have a corresponding relationship.
[0066] In some embodiments of the present invention, the step of the deep learning network model dividing the bridge point cloud structure map into multiple sub-structure maps further includes labeling the component category to which each sub-structure map belongs based on a preset classifier in the deep learning network model.
[0067] In some embodiments of the present invention, the deep learning network model can be a model obtained by a semi-supervised training method using a variable kernel convolution model, and the classifier can be a multilayer perceptron.
[0068] In some embodiments of the present invention, the primitive library pre-defines component categories corresponding to each reference model, and the step of matching the substructure diagram in the pre-define primitive library based on structural parameters includes...
[0069] Based on the component category of the substructure graph output by the classifier, the corresponding category is matched in the primitive library, and multiple reference models of the corresponding category are obtained;
[0070] Based on structural parameters, the substructure graph is matched among multiple reference models in the corresponding category of the preset primitive library.
[0071] like Figure 5 As shown, in the specific implementation process, the shape features of the key structures of the component point cloud are obtained through the point cloud of the substructure diagram;
[0072] Based on the shape characteristics, the structural features of the component point cloud can be determined, such as whether the bridge flange is symmetrical, the parameters of the flange size, the thickness of the beam, the shape and type of the beam, the height of the pier component and the cross-sectional dimensions of the pier component, etc.
[0073] Based on the characteristics of the types mentioned above, primitive models with corresponding structures are matched in the primitive library according to the encoding type.
[0074] In the specific implementation process, different cross-sectional structures are drawn during the creation of the primitive library to create BIM reference models with different detailed structures under each major category;
[0075] In the specific implementation process, the main parameters of the bridge pier section are length, width, and radius. In addition, the bridge pier also includes detailed structures such as the pad stone and the top plate.
[0076] The beam section of a bridge is divided into an inner contour and an outer contour. The outer contour must take into account the box girder size, the dimensions of the two side flanges, and the road surface slope.
[0077] Based on the cross-sectional profile and reference surface constraining the component position, the function model is used to control the cross-sectional size and position of the intermediate layer of the curve, thereby establishing a parametric structural element model that relies on the endpoint position to constrain the overall model shape;
[0078] Depending on the specific structural requirements, adjust the details and repeat the modeling process to establish primitive models for various structural types.
[0079] The method of using a function model to control the cross-sectional dimensions and positions of the intermediate layer of the curve, thereby establishing a parametric structural primitive model whose overall shape is constrained by the endpoint positions, includes:
[0080] For hyperbolic piers, the pier cap section is a straight-sided ellipse, and the sides are two circular arcs with different radii. Because the radii of the two circular arcs are different, it is not possible to directly stretch and model them. Instead, volume fitting is required.
[0081] Therefore, the cross section used for fitting the curve in the middle is constrained by the shapes of the upper and lower cross sections, thereby enabling the control of the pier cap size using parameters;
[0082] Treat the length from the endpoints of the upper and lower sections to the center as the horizontal axis coordinates in the Cartesian coordinate system of the side view, calculate the center of the circle in reverse, and thus find the length of the middle section on this elevation.
[0083] By using the method of solving the center of the circle from two points on the circle, the parameters of the middle layer section of the pier cap can be calculated, thus realizing the pier cap model that is driven by the parameters of the upper and lower layers.
[0084] For a continuous beam, the hyperbolic function of the base surface of the variable-height section is: ,in c It is the length of the sorghum. b It refers to the thickness of the middle section of the variable sorghum beam. a It is a constant. a + b Indicates the thickness at the end of the variable-height beam; x It is a constant, and its value range is (0, ...). c ); This represents the value of a hyperbolic function.
[0085] The shape of a beam can be determined by its length and the thickness at both ends using the hyperbolic equation.
[0086] In the specific implementation process, the various primitive models established will be divided according to structural type, and components of the same type will be coded hierarchically according to category, starting with the same type number, including:
[0087] The primitive models are classified by type, and then categorized layer by layer according to structural and type differences. After refining the structure, the models are coded hierarchically from major categories to minor categories.
[0088] Subcategories within the same major category should have the same prefix number. When adding a subset of subcategories, the number of digits should be increased.
[0089] For the encoding of primitive items at the same level, the preceding major category number remains unchanged, and the encoding of the corresponding number of bits in the current subset is directly modified.
[0090] The various primitive models are divided according to structural type, and components of the same type are coded hierarchically according to category starting with the same type number, finally resulting in a parametric bridge BIM primitive model library containing multiple types of structures.
[0091] In some embodiments of the present invention, the classifier outputs a category number, and the corresponding category is obtained through the category number.
[0092] In the specific implementation process, the category includes a pier category, which includes reference models of solid piers and hollow piers. If a substructure diagram is identified as a pier category, it needs to be matched among multiple reference models including solid piers and hollow piers.
[0093] In some embodiments of the present invention, in the step of the deep learning network model dividing the bridge point cloud structure map into multiple sub-structure maps and labeling the component category to which each sub-structure map belongs, the deep learning network model divides the multiple point clouds in the bridge point cloud structure map into multiple regions, the multiple point clouds in each region constitute a sub-structure map, and determines the category to which the sub-structure map belongs based on a pre-trained classifier.
[0094] In some embodiments of the present invention, the step of receiving the bridge point cloud structure map obtained by radar scanning further includes filtering and noise reduction processing on the received bridge point cloud structure map.
[0095] In specific implementation, the filtering method can be bilateral filtering or Gaussian filtering, etc.; the noise reduction method can be bin-based noise reduction.
[0096] In practical implementation, the collected point cloud data contains a lot of noise and other useless ground features. This invention achieves reverse modeling of point cloud data, requiring the acquisition of feature parameters of the actual targets within the point cloud data. Therefore, it is necessary to separate the target components from the point cloud data. This invention utilizes a deformable kernel-point convolutional neural network to process the point cloud data. Semantic labels are added to the target components in the 3D point cloud scene data. Based on the predicted weights of the points, spherical sampling is performed to extract center point features layer by layer. Resolution is restored through nearest neighbor upsampling, ultimately obtaining the point features of each network layer. The point cloud data of the target components (piers, beams) is extracted based on the weights of the point features.
[0097] like Figure 7 As shown, the present invention also provides a primitive library-driven bridge BIM reverse modeling system, the system comprising:
[0098] The point cloud acquisition module is used to receive the bridge point cloud structure map acquired by radar scanning, input the bridge point cloud structure map into a preset deep learning network model, and the deep learning network model divides the bridge point cloud structure map into multiple sub-structure maps.
[0099] The model matching module is used to match the substructure graph with the point cloud computing structure parameters based on the structure parameters in the substructure graph, and to obtain the reference model corresponding to the substructure graph in the primitive library. The primitive library includes a variety of components, and each component has multiple preset reference models.
[0100] The model generation module is used to obtain the preset parameter types to be modified in the matched reference model, calculate the actual parameter values of the parameter types to be modified based on the positions of each point cloud in the substructure diagram, modify the parameter values of the parameter types to be modified in the reference model to the corresponding actual parameter values, and obtain the actual sub-model for each substructure diagram.
[0101] In some embodiments of the present invention, the system further includes a model building module, which is used to obtain the actual sub-model of each substructure diagram; and to combine multiple actual sub-models based on the position of the substructure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram.
[0102] In practical implementation, digital twins acquire and process multi-source spatial data to achieve a mapping from the real world to the virtual world. During engineering construction management, a clear understanding of the on-site construction situation and progress is necessary. In a digital context, digital engineering management is an indispensable means to achieve rapid and efficient engineering management. Combining the concept of digital twins to achieve the digitization and modeling of engineering construction brings more efficient and intuitive analysis and control methods to engineering management. A virtual construction platform for components is built based on digital twins to construct accurate three-dimensional models of the components.
[0103] However, in construction, due to the complexity of actual engineering structures, the construction speed of 3D BIM models is slow, inefficient, and may not be consistent with the actual construction progress, leading to difficulties in updating, low efficiency, and poor timeliness. If a 3D model could be quickly established, combined with digital twin technology, and based on point cloud data for rapid modeling, the efficiency and timeliness of project management would be greatly improved. The inventors discovered that if a primitive library is created in advance during the twin modeling process, and primitive models containing BIM semantics are pre-stored in the library and arranged using a specific encoding method for easy indexing, the imported point cloud data can be matched with the primitive library model to select the most suitable primitive as a reference. The structural parameters extracted from the point cloud are used to constrain the primitive, and after adjusting the parameters, a 3D model can be obtained, achieving rapid twin modeling with the assistance of point cloud data.
[0104] Existing technologies face challenges due to the noise, surface irregularities, and occlusion inherent in point cloud data. In practical engineering, obtaining a complete model point cloud is difficult, and the structural differences between components in different areas are slight. Furthermore, the large workload and low efficiency of modeling engineering components prevent dynamic updates and real-time rendering. Therefore, accurately and efficiently reconstructing 3D engineering models with BIM semantics from point cloud data automatically remains a challenging issue and a pressing problem that needs to be solved.
[0105] This application achieves reverse engineering of the BIM model from the point cloud by separating independent structures from the scanned point cloud and matching them with primitives in a pre-established bridge BIM primitive library. A matching BIM model is created at the spatial location of the point cloud data, and the dimensional parameters of this BIM model are adjusted according to the actual dimensions of the point cloud data. Furthermore, the individual primitive models that are in contact are connected, ultimately achieving reverse engineering of the point cloud data into a BIM model. In other words, before performing reverse engineering of the point cloud, a parametric model library is pre-established. The size of this model library is controlled by parameters. By identifying the structural parameters of the point cloud data and passing these parameters to the pre-established BIM primitive model, the efficiency of reverse modeling is improved. This avoids the problems of large workload, long modeling cycles, high effort consumption, and difficulty in real-time updates caused by creating a large number of BIM models individually.
[0106] This application achieves twin modeling by creating a parametric model primitive library and matching it with point cloud data, extracting parameters to modify primitive dimensions, and thus rapidly generating 3D component models with BIM semantics based on point cloud data. In other words, before matching with point cloud data for twin modeling, a parametric BIM primitive library for rapid model generation needs to be created first. This effectively avoids problems such as low efficiency, poor timeliness, and update delays caused by creating a large number of models when performing rapid twin modeling of point cloud data.
[0107] The beneficial effects of this invention include:
[0108] 1. By pre-establishing a parameterized primitive model library through prior knowledge, the parameterized primitive models containing coding rules are arranged in the primitive library. Primitive models with homogeneous structures are expressed in the form of parameters, and the size and structure of the model can be adjusted by parameters, which greatly reduces the time spent building the model.
[0109] 2. By using actual collected point cloud data, deep learning methods are used to extract component point cloud data, which greatly improves the segmentation efficiency of point cloud data and facilitates feature extraction of targets based on component point cloud data.
[0110] 3. Using parameters extracted from point cloud data for reverse modeling not only improves the reliability and accuracy of bridge BIM twin modeling, but also speeds up the modeling process, highlighting its high efficiency and real-time characteristics.
[0111] The present invention also provides a primitive library-driven bridge BIM reverse modeling device, which includes a computer device, the computer device including a processor and a memory, the memory storing computer instructions, the processor executing the computer instructions stored in the memory, and when the computer instructions are executed by the processor, the device implements the steps of the method described above.
[0112] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned primitive library-driven bridge BIM reverse modeling method. The computer-readable storage medium can be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disks, removable storage disks, CD-ROMs, or any other form of storage medium known in the art.
[0113] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.
[0114] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.
[0115] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.
[0116] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A primitive library-driven bridge BIM reverse modeling method, characterized in that, The steps of the method include: The system receives a bridge point cloud structure map acquired by radar scanning, inputs the bridge point cloud structure map into a preset deep learning network model, and the deep learning network model divides the bridge point cloud structure map into multiple sub-structure maps. Based on the point cloud computing structure parameters in the substructure graph, the substructure graph is matched in a preset primitive library based on the structure parameters. The reference model corresponding to the substructure graph is obtained by matching in the primitive library. The primitive library includes a variety of components, and each component has multiple preset reference models. Obtain the preset parameter type to be modified in the matched reference model, calculate the actual parameter value of the parameter type to be modified based on the position of each point cloud in the substructure diagram, modify the parameter value of the parameter type to be modified in the reference model to the corresponding actual parameter value, and obtain the actual sub-model corresponding to each substructure diagram. Obtain the actual sub-model of each substructure diagram, and combine multiple actual sub-models based on the position of the substructure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram.
2. The primitive library-driven bridge BIM reverse modeling method according to claim 1, characterized in that, The reference model is preset with contact constraints, which are used to mark the contact surfaces of the reference model. In the step of combining multiple actual sub-models based on the position of the sub-structure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram, the contact surfaces of the corresponding actual sub-models are marked based on the contact surfaces of the reference model. In the process of combining multiple actual sub-models into the actual model, the contact surfaces of the actual sub-models are made to contact each other.
3. The primitive library-driven bridge BIM reverse modeling method according to claim 1, characterized in that, The step of the deep learning network model dividing the bridge point cloud structure map into multiple sub-structure maps further includes labeling the component category of each sub-structure map based on a preset classifier in the deep learning network model.
4. The primitive library-driven bridge BIM reverse modeling method according to claim 3, characterized in that, The primitive library contains pre-defined component categories for each reference model. The step of matching the substructure diagram in the pre-defined primitive library based on structural parameters includes... Based on the component category of the substructure graph output by the classifier, the corresponding category is matched in the primitive library, and multiple reference models of the corresponding category are obtained; Based on structural parameters, the substructure graph is matched among multiple reference models in the corresponding category of the preset primitive library.
5. The primitive library-driven bridge BIM reverse modeling method according to claim 3, characterized in that, In the step of the deep learning network model segmenting the bridge point cloud structure map into multiple sub-structure maps and labeling the component category of each sub-structure map, the deep learning network model divides the multiple point clouds in the bridge point cloud structure map into multiple regions, the multiple point clouds in each region constitute a sub-structure map, and determines the category to which the sub-structure map belongs based on a pre-trained classifier.
6. The primitive library-driven bridge BIM reverse modeling method according to claim 1, characterized in that, The step of receiving the bridge point cloud structure map acquired by radar scanning further includes filtering and noise reduction processing on the received bridge point cloud structure map.
7. A primitive library-driven bridge BIM reverse modeling system, characterized in that, The system includes: The point cloud acquisition module is used to receive the bridge point cloud structure map acquired by radar scanning, input the bridge point cloud structure map into a preset deep learning network model, and the deep learning network model divides the bridge point cloud structure map into multiple sub-structure maps. The model matching module is used to match the substructure graph with the point cloud computing structure parameters based on the structure parameters in the substructure graph, and to obtain the reference model corresponding to the substructure graph in the primitive library. The primitive library includes a variety of components, and each component has multiple preset reference models. The model generation module is used to obtain the preset parameter types to be modified in the matched reference model, calculate the actual parameter values of the parameter types to be modified based on the position of each point cloud in the substructure diagram, modify the parameter values of the parameter types to be modified in the reference model to the corresponding actual parameter values, and obtain the actual sub-model corresponding to each substructure diagram. The model building module is used to obtain the actual sub-model of each substructure diagram, and combine multiple actual sub-models based on the position of the substructure diagram in the bridge point cloud structure diagram to obtain the actual model corresponding to the bridge point cloud structure diagram.
8. A primitive library-driven bridge BIM reverse modeling device, characterized in that, The device includes a computer device, which includes a processor and a memory, wherein the memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the device implements the steps of the method as described in any one of claims 1-6.