BIM-based project progress management system and its application

Abstract

The present invention relates to a project progress management system and application based on BIM. A data model is generated by combining a three-dimensional data model expression method with a method for performing occlusion query from multiple viewpoints to identify hidden parts in an assembly. The progress management model establishes a real-time model and generates a corresponding calculation model based on the analysis of the data model structure. The comparison model overlays and compares the real-time model with the planned model to provide a solution for correcting and adjusting the construction progress. The present invention requires continuous adjustment of goals during the project process and takes appropriate measures to solve problems that arise. Projects often have completion time, total cost or resource allocation deviating from the original planned track. Corresponding measures need to be taken to ensure that project development is consistent with the plan. If the project undergoes major changes or seriously deviates from the project process, it is necessary to rearrange the project schedule and determine the target plan, adjust resource allocation and budget costs, and thus achieve progress balance.

Description

Technical Field

[0001] The present invention relates to the field of management systems, and in particular to a BIM-based project schedule management system and its application. Background Art

[0002] State Grid Infrastructure requires the solid promotion of the standardized application of 3D design in construction management, construction organization, and file transfer. It also requires the establishment of an information transmission and data sharing model based on 3D design applications to meet the management requirements of all stages of engineering construction. The company also explores the use of 3D design results for process standardization management. Traditional construction planning and progress management methods are becoming increasingly difficult to meet existing business needs, primarily due to the lack of visual and intuitive simulation methods for owners and construction managers to remotely control construction site progress.

[0003] Foreign research on 4D simulation of project progress is mainly based on the application of BIM technology. Exploring how to add the time dimension to the three-dimensional building model to achieve 4D simulation of project progress is an important aspect of the application of BIM technology in progress management abroad.

[0004] The existing technology has the following shortcomings: the existing management system still has many limitations. First, the system mainly focuses on the dynamic management of the progress of the construction phase, and does not involve other engineering phases such as engineering design, and cannot be involved in the initial stage of the project; second, the three-dimensional composite model established by the system is only a simulation of the image and size of the building product, and does not include other building product information. It is not a true building information integration model. Therefore, a BIM-based project progress management system is needed to solve the above problems. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the deficiencies of the above-mentioned prior art and provide a BIM-based project progress management system, the system including a data model, a progress management model and a comparison model;

[0006] The data model is generated by combining the three-dimensional data model expression method with occlusion query from multiple viewpoints to identify hidden parts in the assembly. The progress management model establishes a real-time model and generates a corresponding calculation model based on the analysis of the data model structure. The comparison model superimposes and compares the real-time model with the planned model to provide solutions for correcting and adjusting the construction progress.

[0007] Preferably, the data model includes a three-dimensional information model of a power grid, and the three-dimensional information model of the power grid is converted into a surface model by scanning the entity, specifically in the following steps:

[0008] Assume that the points before scanning are A(XA, YA, ZA), B(XB, YB, ZB) ... K(Xk, Yk, Zk), and the scanning vector is V(XV, YV, ZV). After scanning these points along the scanning vector V, we get An, Bn ... Kn. The line-surface network formed by these points is the boundary model of 3DGIS:

[0009] An=A+V, Bn=B+V...Kn=K+V

[0010] For curve scanning, select the curve Curve = (M1 (X1, Y1, Z1), M2 (X2, Y2, Z2) ... Mn (Xn, Yn, Zn)) and change the tangent vector of the curve to Hn = Xn-Xn-1, Yn-Yn-1, Zn-Zn-1. If there is a point A (XA, YA, ZA, 0) after scanning, the point A' is calculated as follows, where N is the surface normal:

[0011]

[0012] The transformation from the entity model to the boundary model is achieved by analogy.

[0013] Preferably, the three-dimensional data model includes a solid model and a mesh model, wherein the solid model is used to describe the geometric and topological information of the component, and the mesh model uses polygonal facets to construct the geometric shape of the object.

[0014] Preferably, the method for performing occlusion query from multiple viewpoints to identify hidden parts in an assembly comprises the following steps:

[0015] (1) Generate a query object ID for the object to be queried;

[0016] (2) Start occlusion query;

[0017] (3) Rendering objects that require occlusion query;

[0018] (4) End the occlusion query;

[0019] (5) extract the number of samples that pass the occlusion query;

[0020] (6) Delete the query object ID and recycle resources.

[0021] Preferably, in the hidden parts identification algorithm, the visibility value of the component at a certain viewpoint is calculated as:

[0022]

[0023] Where: λ1+λ2+λ3=1;

[0024] Among them, λ1, λ2 and λ3 are weights. Initially, the part visibility rate, part area contribution rate and part volume contribution rate are set to have the same weight, that is, λ1, λ2 and λ3 are all 1 / 3;

[0025] After the occlusion query of part i is completed at all the set viewpoints, the maximum visibility value of part i is obtained, that is, Define the critical parameter F T (0 <F T <1), when the F value of a component in the assembly is greater than the critical parameter, the component is visible when observed from the outside of the assembly, otherwise the component is hidden.

[0026] Preferably, the hidden piece identification algorithm evaluates the priority value C(v) of a vertex by its position in the cache and its degree, while the priority value C(R) of a triangle ring is the sum of the priority values of all unoutput triangles in the ring.

[0027] The vertex priority value C(v) can be calculated by the following formula:

[0028] C(v)=C p (v)+C a (v)

[0029] in,

[0030]

[0031]

[0032] In the above formula, p represents the position of vertex v in the cache, s represents the size of the cache space, a represents the number of unoutput triangles adjacent to vertex v, and k1, k2, and k3 are coefficients.

[0033] Preferably, the model metadata structure of the three-dimensional information model of the power grid includes basic data and extended data, wherein the basic data includes geometric data, physical data and functional data, and the extended data includes technical data, economic data and management data.

[0034] Preferably, the real-time model includes a statistical distribution of process duration, which includes normal distribution, beta distribution, uniform distribution, and triangular distribution. In normal distribution, the mean, median, and mode of the process duration coincide, the probability of the process duration being greater than or less than the mean is equal, and the probability of the process being delayed or advanced is equal. In beta distribution, the distribution range is in [0, 1]. The distribution formula of beta distribution is as follows:

[0035] (1) Mean:

[0036] (2) Variance:

[0037] The value range of α and β is 0:0.1:10000, and the value range of the mean is 0.3:0.05:0.5. The simulation results are stored in the matrix, with the first column being the α value, the second column being the β value, and the third column being the variance value.

[0038] Preferably, in the beta distribution, the safety time is the difference between the duration of a process with a 95% completion probability and the duration of a process with a 50% completion probability, and the formula is as follows:

[0039] T k =[F(x)I x＝0.95 -F(x)I x＝0.5 ]×t k

[0040] Where F(x) is the distribution function of the Beta distribution, T k is the safety time of process k, t k is the duration of process k.

[0041] The present invention also provides an application of a BIM-based project progress management system, which is applied to substation construction simulation project management. Based on the D3Station platform, it realizes the analysis and visualization of the model and supports viewing the model's properties.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention analyzes the characteristics of GIM three-dimensional product models and the current data expression methods of three-dimensional models. Based on the characteristics of mobile devices, a three-dimensional data model expression method that can be quickly displayed on desktop devices is proposed. This data model has the advantages of fast rendering speed and memory conservation, meeting the needs of efficient browsing on desktop platforms, and the security of the model is also guaranteed. In addition, the data model contains complete geometric information, product structure information and attributes, meeting the needs of users in actual engineering. The data model of the present invention has a small file size and filters non-geometric information such as parameters and PMI in the original model. The separation of assembly files and part files avoids repeated expression of the same model. Measures such as the reasonable organization and compression of geometric data significantly reduce the amount of model data. The efficiency of model transmission via mobile networks is greatly improved compared to the original model, which is also conducive to storing more three-dimensional models on mobile devices.

[0044] 2. Fast data model reading fully considers the characteristics of mobile device hardware and 3D graphics rendering systems. During the data model generation phase, data is rationally organized and displayed optimized, and time-consuming optimization operations are transferred to desktop computers with stronger computing power. This avoids the weakness of low hardware performance of some desktop devices. The generated data model can fully utilize the performance of mobile terminals, speeding up loading and rendering, making it possible to present and interact with 3D models on low-performance mobile devices.

[0045] 3. A data model generation method is presented, and the lightweight algorithm used is studied. A method for identifying hidden parts in assemblies by performing occlusion queries from multiple viewpoints is proposed. This method can automatically determine invisible components within the assembly model and subtle components that have little impact on the appearance. Combined with lightweight expression methods, the number of primitives drawn can be reduced and rendering efficiency can be improved without destroying the product structure. In addition, a linear-time vertex cache optimization algorithm is proposed. This algorithm uses effective heuristic search rules and can still achieve good cache optimization effects when the cache parameters are unknown, which can effectively speed up the display of the data model.

[0046] 4. By comparing and analyzing the actual progress with the project plan, many deviations can be found, and potential problems in the project can be pointed out. In order to avoid problems caused by deviations, it is necessary to constantly adjust the goals during the project process and take appropriate measures to solve the problems that arise. Project completion time, total cost or resource allocation often deviate from the original planned track. Appropriate measures need to be taken to make the project development consistent with the plan. If the project undergoes major changes or seriously deviates from the project process, it is necessary to rearrange the project schedule and determine the target plan, adjust resource allocation and budget costs, so as to achieve progress balance. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the technical route of the present invention.

[0048] Figure 2 This is a data structure diagram of the power grid information model of the present invention.

[0049] Figure 3 This is a data structure diagram of the data model of the present invention.

[0050] Figure 4 This is a flow chart for generating the data model of the present invention.

[0051] Figure 5 This is a statistical diagram of the occlusion query results of the present invention.

[0052] Figure 6 This is a flow chart of the identification of hidden parts of the present invention.

[0053] Figure 7 It is a simplified schematic diagram of the engine model of the present invention.

[0054] Figure 8 Schematic diagram of the triangle sequence of the present invention.

[0055] Figure 9 This is a flow chart of cache optimization of the present invention.

[0056] Figure 10 This is a data structure diagram of the model element of the present invention.

[0057] Figure 11 This is a system architecture diagram based on MVC of the present invention.

[0058] Figure 12 This is a flowchart of the system architecture implementation based on MVC of the present invention.

[0059] Figure 13 This is a system diagram of the BIM-based progress management application framework of the present invention.

[0060] Figure 14 This is the overall framework diagram of the BIM information platform of the present invention.

[0061] Figure 15 Schematic diagram of 4D simulation analysis of the progress plan of the present invention.

[0062] Figure 16 This is the network diagram operation sequence analysis diagram of the present invention.

[0063] Figure 17 This is the progress monitoring view of the present invention.

[0064] Figure 18 This is a comparison chart between the planned progress model and the actual progress model of the present invention.

[0065] Figure 19 This is a block diagram of the substation 4D construction progress management platform of the present invention. DETAILED DESCRIPTION

[0066] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0067] It should be noted that when an element is referred to as being “fixed to” another element, it may be directly on the other element or there may be an intermediate element. When an element is referred to as being “connected to” another element, it may be directly connected to the other element or there may be an intermediate element.

[0068] Example 1

[0069] See also Figure 1 As shown, the BIM-based project progress management system described in this embodiment includes a data model, a progress management model, and a comparison model, wherein:

[0070] Data model: Combining the advantages of the entity model and the grid model, it fully considers the usage requirements of low-performance desktops. Overall, it has the following features and advantages:

[0071] (1) Small file size: The data model filters the parameters and PMI and other non-geometric information in the original model. The separation of assembly files and part files avoids repeated expression of the same model. The reasonable organization and compression of geometric data greatly reduce the amount of model data. The efficiency of transmitting the model through the mobile network is greatly improved compared with the original model, which is also conducive to storing more 3D models on mobile devices.

[0072] (2) Fast data model reading fully considers the characteristics of mobile device hardware and 3D graphics rendering system. In the data model generation stage, the data is reasonably organized and displayed, and the relatively time-consuming optimization operations are transferred to the desktop computer with stronger computing power. This avoids the weakness of low hardware performance of some desktop devices. The generated data model can fully utilize the performance of mobile terminals, speed up loading and rendering, and make it possible to demonstrate and interact with 3D models on low-performance mobile devices.

[0073] (3) The use of memory-saving references allows only one copy of the same part to be stored in memory, and the others are obtained through the spatial transformation matrix, which can greatly reduce memory usage and meet the more demanding memory requirements of three-dimensional applications;

[0074] (4) Complete model information The data model contains complete geometric information and product structure information, has high usability, and meets the demand for the use of three-dimensional models in mobile network environments;

[0075] (5) Ensure data security. The data model does not contain information such as the model's characteristic parameters and modeling parameters. The model cannot be edited again, which can ensure that the data model is not used for other purposes such as reverse engineering.

[0076] First, the characteristics of GIM 3D product models and the existing data representation methods of 3D models were analyzed. Based on the characteristics of mobile devices, a 3D data model representation method that can be quickly displayed on desktop devices was proposed. This data model has the advantages of fast rendering speed and memory conservation, meeting the needs of efficient browsing on desktop platforms. The model security is also guaranteed. In addition, the data model contains complete geometric information, product structure information and attributes, meeting the needs of users in actual engineering.

[0077] Secondly, a method for generating data models is given, and the lightweight algorithm therein is studied. A method for performing occlusion queries from multiple viewpoints to identify hidden parts in an assembly is proposed. This method can automatically determine the invisible parts inside the assembly model and the subtle parts that have little impact on the appearance. Combined with the lightweight expression method, the number of primitives drawn can be reduced and the drawing efficiency can be improved without destroying the product structure relationship. In addition, a linear-time vertex cache optimization algorithm is proposed. This algorithm uses effective heuristic search rules and can still obtain good cache optimization effects even when the cache parameters are unknown, which can effectively speed up the display speed of the data model.

[0078] Schedule management model: By analyzing the content of engineering project schedule management and commonly used schedule management technical methods, exploring the establishment of a three-dimensional power grid information model and the creation of construction progress data, analyzing the establishment of a four-dimensional power grid information model, and proposing an implementation method for the integration of BIM data and construction progress. Combining the system functional requirements and the analysis of the three-dimensional model data structure, the structural system of the BIM4D system is formed, and its implementation process is designed.

[0079] Comparison Model: This system systematically discusses the process of collecting progress information at the construction site and the method of overlaying and comparing the real-time model with the planned model. This provides a solution for correcting and adjusting the construction progress, solving the problem of comparing the real-time model of the construction site with the planned model with associated information in the platform. Specifically, the system uses a camera to collect construction site information, reconstructs the construction site model through image registration, feature point matching, motion structure, model registration, and other technologies, and overlays it with the planned model, providing a real-time data foundation for the subsequent output of progress reports and 3D progress display.

[0080] After the actual progress information is updated in the system, a comparative analysis between the actual progress and the project plan can reveal many deviations and point out potential problems in the project. To avoid problems caused by deviations, it is necessary to constantly adjust the goals during the project process and take appropriate measures to solve the problems that arise. Projects often deviate from the original planned track in terms of completion time, total cost or resource allocation. Appropriate measures need to be taken to make the project development consistent with the plan. If the project undergoes major changes or seriously deviates from the project process, it is necessary to rearrange the project schedule and determine the target plan, adjust resource allocation and budget costs, and achieve progress balance.

[0081] Example 2

[0082] See also Figure 2 As shown in the figure, in the data model, the main idea is to study the data organization structure of general models, product models, power transmission and transformation projects, etc., extract the common characteristics of the data and summarize them to form a data analysis plan for the power grid information model, and establish detection indicators according to the company's three-dimensional design standards to detect the integrity and correctness of the data format and content. Finally, the efficient processing and storage of the power grid information model data are studied. Referring to the organization method of geographic information data, the pyramid scheme is studied according to the display scale and data refinement. The LOD strategy is studied in combination with the data composition structure. The data is simulated from a high altitude to observe the process from blur to clarity of the object. Starting from the data volume, the data geometry texture simplification plan is studied to optimize data storage under the premise of ensuring data clarity.

[0083] See also Figure 3 As shown in Figure 2, the most important factor affecting the technical framework of the fusion of the grid information model (GIM) and GIS is often the data format. Since the grid information model and GIS data have their own data formats, how to make these multi-source heterogeneous data formats compatible on a single platform becomes a key issue.

[0084] Data format conversion is currently the mainstream method for integrating GIS with 3D models. Most data format conversion research focuses on the conversion from IFC to CityGML, with some research focusing on the conversion from CityGML to IFC. The conversion from power grid information model data to GIS data is a process of coarsening data from refined data. Parsing the power grid information model includes two aspects: geometry and semantics:

[0085] 1) Multi-level geometric information extraction and conversion

[0086] Most 3D information models of power grids are based on the solid geometry construction method (CGS), while GIS expression formats are mostly based on the surface boundary method. Therefore, geometric information conversion is required. The scanning method is used to convert a solid to a surface model. After scanning, the scanned model needs to be triangulated to obtain the surface model of 3D GIS.

[0087] Scanning a straight edge is relatively simple. Suppose the points before scanning are A(XA, YA, ZA), B(XB, YB, ZB)...K(Xk, Yk, Zk), and the scanning vector is V(XV, YV, ZV). After scanning these points along the scanning vector V, we get An, Bn...Kn. The line-surface network formed by these points is the boundary model of 3DGIS:

[0088] An=A+V, Bn=B+V...Kn=K+V

[0089] For curve scanning, we select the curve Curve = (M1 (X1, Y1, Z1), M2 (X2, Y2, Z2) ... Mn (Xn, Yn, Zn)) and change the tangent vector of the curve to Hn = Xn - Xn-1, Yn - Yn-1, Zn - Zn-1. If there is a point A (XA, YA, ZA, 0) after scanning, the point A' is calculated as follows, where N is the surface normal:

[0090]

[0091] By analogy, the transformation from the entity model to the boundary model can be achieved;

[0092] 2) Mapping of semantic information

[0093] In addition to simple geometric information conversion, semantic information mapping is also an important part of the integration of GIM and GIS. Simple data format conversion does not achieve the integration of GIM and GIS. The structure of the model changes but the semantics of the model does not change.

[0094] First, there are differences in geometric expression forms. The power grid information model has three expression forms: boundary description, swept body formed by stretching or rotation, and constructed solid geometry. However, the data standard supported by the GIS platform only has one geometric expression form, boundary description. After the GIM model is converted into a format supported by the GIS platform, the geometric information expressed by the swept body formed by stretching or rotation and the constructed solid geometry method can only be expressed using the boundary description method. A large amount of coordinate data is required to express multiple facets, which will inevitably cause the loss of geometric information and the increase of data volume. On the other hand, there is a difference in object semantics. Since the power grid information model and 3DGIS have different expressions and understandings of spatial objects, and there is no relevant object semantic standardization research work, the loss of semantic information is inevitable in the process of converting the GIM model into a data format supported by GIS. Therefore, it is necessary to reconstruct the geometric semantics of the power grid information model into a 3DGIS model.

[0095] The power grid information model is semantically rich and can often be mapped in the 3D GIS LOD4 hierarchical model. However, the semantic information of the LOD3-LOD0 hierarchical models decreases step by step. Therefore, semantic filtering is required during the conversion process to abstract and filter some information and retain the useful information. The semantic information and geometric information of each component are associated with a unique ID value to provide index information for model data management.

[0096] 3) Changes in spatial scale

[0097] After the GIM solid model is triangulated into a GIS-supported triangulated mesh model, since GIM uses a local coordinate system, when the model is imported into the 3D-GIS system, it is necessary to use the model measurement points and the actual measurement point coordinates to establish positioning reference information, obtain the model family coordinate transformation matrix, and transform the local coordinate system into the global coordinate system to complete the correct positioning of the model in the 3D-GIS system coordinate system. The three commonly used models for transformation between different coordinate systems are the Bursa seven-parameter model, the Molodensky model, and the Wuce model.

[0098] To solve the problem of quickly browsing 3D product models on mobile devices with limited performance and resources and to enrich the channels for users to communicate product data, it is necessary to use desktop computers to perform lightweight processing on the original model to obtain a data model that can be directly browsed by mobile devices.

[0099] Based on the current problems of 3D model data sharing, this paper analyzes the characteristics of 3D product models and the shortcomings of existing model expression methods. A data model expression method that can be quickly displayed on mobile devices is proposed. Generally speaking, the expression methods of 3D models can be divided into two categories: solid models and mesh models. The details are as follows:

[0100] 1) Solid Model: Solid models can be divided into three categories: boundary representation (B-Rep, Boundary Representation), construction solid geometry (CSG, Construction Solid Geometry) and scanning methods. Most 3D models created by 3D systems (Pro / E, UG, Catia, Inventor, etc.) are solid models.

[0101] Solid models can accurately describe the geometry and topology of components. However, they have the following problems, which limit their application in data sharing:

[0102] (1) The solid model does not contain any triangular face information. Before displaying, a lot of calculations are usually required to discretize it into polygonal meshes, so the loading time is long. For complex 3D product models, the long loading and display time is even more unacceptable.

[0103] (2) The data structure of the solid model is complex and the file size is huge. The information in it is for various purposes. For users, much of the information is redundant. For example, proofreaders and reviewers mainly use the model to browse and review. They only focus on the structural geometry information of the parts. Analysis engineers only need to use simplified product geometry information during the analysis process. Remote customers only need geometric information to browse the appearance of the model.

[0104] (3) To browse the solid model, one must use the same 3D system as the modeler or general 3D browsing software. However, general 3D browsing software is usually developed by a third party and generally has problems such as lag and lack of stability. The diversity of 3D systems leads to the heterogeneity of 3D product data, which affects data sharing and communication within and between enterprises and reduces production efficiency.

[0105] (4) The confidentiality of the solid model is poor, as it contains sensitive design information such as the modeling process and feature parameters. Once it is used for purposes other than collaborative development, such as reverse engineering, it will cause significant losses to the enterprise;

[0106] 2) Mesh model: A mesh model is an imprecise model expression method that uses a large number of polygonal facets to approximate the geometric shape of an object. Mesh models exist in a variety of file formats, such as STL, VRML, PLY, and OBJ. The most common of these is the STL file format launched and promoted by 3D Systems. STL mesh models are widely used in the manufacturing industry. Most 3D systems can output model files in STL format. STL files describe the geometric shape of a model in the form of triangular facets. Each triangular facet records the normal vector of a triangle and the position coordinates of its vertices. STL mesh models exist in both ASCII and binary storage formats. The file structure of the ASCII format is shown below:

[0107]

[0108]

[0109] Nx, Ny, and Nz in the file represent the components of the triangle normal vector in the xyz direction, V 1x , V 1y , V 1z They represent the coordinate values of the vertex in the xyz direction respectively, and the rest are keywords in the file;

[0110] As can be seen from the above STL example, the mesh model has a simple data structure and the triangular facet representation allows it to utilize hardware-accelerated rendering to quickly display the 3D model. However, the mesh model has the following problems:

[0111] (1) The mesh model only contains the display information of the object, without the assembly structure relationship of the model, and cannot express the relationship between parts. It also lacks key information such as attributes, making it difficult to meet the actual requirements of the project;

[0112] (2) Grid models are generally not compressed or the compression ratio is not high, resulting in a large amount of data, which is not conducive to network transmission and storage of the model;

[0113] Because traditional 3D models have many of the aforementioned deficiencies in conveying information, it is necessary to extract the geometric information of 3D product models, compress and optimize it, and simplify the extraction of required attribute information and product structure information, thereby obtaining a data model that is small in size, highly efficient, and easy to browse quickly. The following will introduce the expression method of this data model.

[0114] Data models are primarily used for quick browsing and interaction on desktop devices. Therefore, the design of data model information storage and expression mechanisms must fully consider the inherent characteristics of desktop devices and optimization strategies for desktop graphics system performance bottlenecks. To ensure that the lightweight 3D product model meets the needs of desktop users, the generated data model must meet the following conditions:

[0115] (1) The data model must first ensure the browsing experience on desktop devices. Due to the embedded architecture, the hardware resources and performance of desktop devices are always not comparable to those of desktop computers. Their computing and graphics rendering capabilities are relatively limited. In order to ensure a good browsing experience, the model must have a fast display speed and be able to meet the needs of real-time interaction. The data model file should be reasonably organized and not too complicated. Under the premise of ensuring the display speed, the compression ratio and other performance can be appropriately sacrificed. At the same time, certain optimization measures that do not rely on the latest hardware support can be taken to ensure that complex models can obtain a relatively smooth experience on various desktop devices.

[0116] (2) The data model must ensure its usability in the engineering field. The lightweight process generally simplifies and compresses the geometric information of the product model, and simplifies and extracts information such as size, attributes, fit, and parameters. The data model generated by this process will lose some information, but this process must not destroy the usability of the model. Specifically, in terms of geometric appearance, hanging edges, hanging surfaces, and holes are not allowed, and key features must be retained. In terms of non-geometric information, due to the limitations of screen size and interaction methods, three-dimensional applications on desktop devices are more display-based. Therefore, modeling processes, engineering constraints, features, and other information can be simplified or exist as optional options, but the assembly and product structure information must be extracted, otherwise the data model will lose its role in data sharing in the engineering field.

[0117] (3) The data model needs to have a small file size. Due to the limitations of the embedded system architecture of desktop devices, their storage and memory capacity are small and difficult to expand. Therefore, it is very necessary to reduce the size of the model. On the one hand, it is convenient to transmit the model through the network, and on the other hand, it can save memory and make it possible to demonstrate data on low-performance desktop devices.

[0118] Based on the above requirements, the following design scheme for the 3D data model file is proposed:

[0119] (1) For each part, the triangle mesh data of the same color are organized together, geometrically compressed, and saved as a separate file. Organizing data based on color rather than face as the basic unit can maximize the amount of data input at one time. It can also reduce the complexity of the shader and the state changes during the rendering process. Only one color value needs to be recorded in the color list, which not only reduces data redundancy but also optimizes the rendering pipeline.

[0120] (2) XML (Extensible Markup Language) format is used to describe the model product structure and attribute data. XML has good scalability and platform independence, and can achieve the separation of content and form. It is very suitable for describing highly structured assembly models. In addition, the XML expression method enables the loading process to use SAX to parse the model file sequentially. The SAX parsing method does not need to save the entire document in memory, so it is very suitable for mobile devices with limited memory resources.

[0121] (3) Use references to reduce redundancy. In a 3D product model, there are usually many parts with the same appearance. By referencing, data redundancy can be avoided, thereby reducing the size of lightweight files and also reducing memory usage.

[0122] (4) Manage data in the form of containers. In lightweight files, all data are included in different container nodes. Lightweight assembly files point to the geometric data files of parts through containers, which is conducive to data expansion and modification.

[0123] (5) Using Levels of Detail (LOD) technology, multi-resolution representation of parts is added to the geometric data, and the LOD model of the part is supported by the bounding box of the part. The bounding box of the part corresponds to the rough model, while the actual part model corresponds to the precise model. By sacrificing display quality, the model drawing time can be reduced, making it easier for users to quickly browse and interact with the 3D model.

[0124] (6) Accelerate the display of geometric data. Display acceleration is mainly reflected in two aspects. On the one hand, the visibility of parts is pre-judged in lightweight design. The visibility of parts is recorded in the assembly file. When drawing the model, invisible parts and subtle parts can be avoided from entering the rendering pipeline based on the pre-provided visibility information. When needed, they can also be manually displayed. A good balance is achieved between drawing speed and usability. On the other hand, the vertex cache of mesh data is optimized. By modifying the triangle sequence of the mesh, the vertex cache hit rate in the rendering process is improved, thereby speeding up the display speed of the data model.

[0125] See also Figure 4 As shown, the data model file adopts a multi-document structure, and the assembly file and the part file are stored separately. The assembly file records the product structure and attribute information, and the part file records the optimized display information.

[0126] Assembly files record relevant information about lower-level subassemblies and parts, including component names, relative paths of lower-level components, component visibility, relative transformation matrices, geometric display properties, etc. Combined with part geometric information, product models can be quickly displayed and interacted with. Assembly files extract attribute information from the original model, including physical properties such as product weight, area, volume, and weight unit, as well as user-defined attributes such as product code and stage mark. In addition, assembly files record the name, format, and relative path of model thumbnails. The existence of thumbnails allows users to roughly understand model information without opening the model file, avoiding the time-consuming loading of complex models. Assembly files do not contain any geometric information, and their data volume is not large, so the file size can be kept small.

[0127] The part file is used to record the display information of the part in its own object coordinate system, including triangular mesh data for fast display and point and edge information used to represent contour points and contour edges. Because most models use face shading, the report organizes mesh data by color. When multiple faces have the same color, only one color value needs to be recorded in the color list. The mesh data is first optimized through the vertex cache and then stored using an index. That is, the part display information is saved as a unique vertex and a triangle sequence arranged by vertex index. For 3D product models, their topological relationships are relatively complex. Considering the low computing performance of mobile devices, the model is not topologically compressed in the part file. Only geometric compression is performed to reduce the accuracy of the vertex data.

[0128] When viewing a data model file, the user can first determine whether it is the model they want to browse with the help of the thumbnail, and then choose whether to open the model file. The loading of the data model file starts with traversing the overall product structure to obtain all lightweight sub-assemblies and parts, and then constructing the entire scene and drawing it. In the lightweight assembly file, the matrices of all components are transformation matrices relative to the position of the upper-level components. Therefore, the transformation matrices of all upper-level nodes of the part can be multiplied to obtain the absolute position of the part in the overall assembly. During the display process of the file, the components marked as invisible by the visibility attribute can be trimmed at the beginning and not sent to the rendering pipeline. When the internal structure of the assembly needs to be viewed, this part of the geometric data will be sent to the rendering pipeline for rendering;

[0129] The main bottleneck of current platform graphics 3D graphics rendering is the contradiction between the huge amount of model data and the limited hardware performance of the terminal. Model simplification and model data layout optimization are two feasible methods to solve this bottleneck. Simplifying the model can reduce the amount of graphics data that needs to be drawn, and optimizing the model data layout can increase the reuse rate of the hardware cache, thereby achieving the purpose of display acceleration. To this end, a hidden parts recognition algorithm for assemblies is proposed. Combined with a lightweight expression method, some parts in the assembly can be simplified to reduce the rendering workload. In addition, in order to improve the vertex cache hit rate, this chapter studies the triangle layout of the mesh model and proposes a new vertex cache optimization algorithm. Both optimization operations are performed on desktop computers, allowing low-performance devices to efficiently browse the data model.

[0130] In complex 3D product models, there are a large number of component models, deep hierarchical structures, complex structural relationships, and slow display speed. To meet the need for fast browsing of 3D product models on different hardware, a multi-level data model expression method is proposed. The data model needs to extract geometry, topology, and corresponding assembly information from the original model and organize them reasonably. Therefore, the corresponding data model export interface program is developed using the secondary development tools provided by common 3D systems. This can achieve the desired effect based on the actual project, with high efficiency and flexibility. These development tools are all backward compatible, so the export interface program is relatively stable and does not require frequent changes.

[0131] See also Figure 5 As shown, the lightweight process inputs 3D product models in various file formats. After the following lightweight operations on a desktop computer, a data model that can be quickly displayed directly on a terminal desktop device can be obtained;

[0132] 1) Extraction of product structure and attribute information: Extract the product structure, name and attribute information of all parts of the assembly, as well as non-geometric information such as the transformation matrix of each part relative to the parent assembly. The structure of the assembly is a clear tree structure. During the information extraction process, a recursive approach can be used to obtain the required information. When the model type is an assembly, the information is extracted and the next recursive operation is performed. If it is a part, the recursive operation is exited after the information extraction is completed. Only with the structure and attribute information can the data model be accurately displayed and the attribute query be performed in an environment outside the 3D system;

[0133] 2) Triangulation of geometric data: triangulating each part in the assembly to obtain a triangular mesh that supports fast display;

[0134] 3) Extraction of key geometric information: extracting key edge and point information from the model to support certain special display requirements;

[0135] 4) Hidden parts identification and recording: Perform occlusion queries on the entire assembly model from multiple viewpoints, pre-judge the visibility of parts based on multiple results, and record the results in the lightweight assembly file. This part will be described in detail in the next section;

[0136] 5) Vertex cache optimization: This step modifies the triangle sequence of the 3D model to increase the vertex cache hit rate, reduce memory bandwidth requirements, and speed up model rendering. This will be described in detail below.

[0137] 6) Coding and compression: The geometric information of the part file is geometrically encoded through three steps: vertex position quantization, prediction, and entropy coding. In addition, the file is stored in binary format and the assembly file is compressed using Huffiman coding;

[0138] Since the 3D product model contains a large number of parts and their structural relationships are also very complex, and the entire assembly generally contains invisible parts inside the model and subtle parts that have little impact on the appearance, which are generally referred to as hidden parts. As analyzed in Chapter 2, in order to achieve fast drawing of 3D models on desktop terminal devices, hidden parts in the assembly can be eliminated to reduce the number of primitives sent to the drawing pipeline. Therefore, during the data model export process, an automatic method is needed to determine whether the part is a hidden part and record its visibility in the generated lightweight assembly file. When the data model is displayed, hidden parts can be chosen not to be drawn, and they can be manually displayed if necessary, achieving the purpose of simplification while retaining the complete product structure of the model;

[0139] The basic idea of hidden parts identification is to perform occlusion queries on the assembly from multiple viewpoints to obtain the visible area of the assembly itself and each component. Then, all query results and the volume information of the components are combined to obtain all hidden components.

[0140] In order to judge the visibility of the parts, OpenGL's occlusion query technology is used to query the model to get the number of pixels that pass the depth test. Then, the query result can be used to decide whether to render the object.

[0141] To use occlusion queries, you need to perform the following steps:

[0142] Step 1: Generate a query object ID for the object to be queried;

[0143] Step 2: Start occlusion query;

[0144] Step 3: Render the objects that need to be occluded;

[0145] Step 4: End occlusion query;

[0146] Step 5: Extract the number of samples that pass the occlusion query;

[0147] Step 6: Delete the query object ID and reclaim resources;

[0148] Using occlusion query technology can reduce the number of primitives that need to be drawn and significantly improve the drawing frame rate. However, the overhead of executing occlusion queries is very high, which is mainly manifested in two aspects. First, each query will add an additional draw call, resulting in a large overhead. In addition, waiting for the query results will cause a large delay, affecting the performance of the CPU and GPU. Due to the limited hardware rendering capabilities of ordinary desktop devices and the fact that occlusion queries have only been supported since the OpenGL ES 3.0 standard, there are still significant limitations in directly using this technology to accelerate the drawing of complex 3D graphics on ordinary desktop devices. However, the occlusion query function can be used to comprehensively judge the visibility of all components in the assembly from multiple viewpoints in the data organization stage of the data model. When drawing the model, the visibility information can be used to determine whether to draw and when to draw, which is conducive to fast browsing of the 3D model.

[0149] In order to fully understand the model, users need to observe the model from multiple angles. To this end, multiple viewpoints are set to comprehensively judge whether the parts are visible. In this way, the simplified 3D model can still correctly reflect the appearance of the parts and meet the purpose of 3D model data communication. The viewpoints are evenly distributed on a sphere with a radius of three times the radius of the assembly's bounding sphere, so that the entire assembly is just within the viewing cone. In order to achieve a relatively balanced effect between algorithm complexity and result accuracy, the number of viewpoints is set to 12, and the viewpoint direction is that the vertex points to the center of the 3D model's bounding sphere. Occlusion query tests are performed from these viewpoints to obtain the number of pixels on the screen after the parts are rendered. By comprehensively analyzing the rendering results, it can be finally determined whether the parts are hidden.

[0150] In order to describe the algorithm more clearly, the following variables are defined:

[0151] (1) represents the number of pixels that pass the depth test when performing an occlusion query on part i at viewpoint j when rendering the assembly;

[0152] (2) Indicates the number of pixels that pass the depth test when performing an occlusion query on part i at viewpoint j when rendering part i alone;

[0153] (3) Represents the number of pixels that pass the depth test when performing an occlusion query on the entire assembly at viewpoint j when rendering the assembly;

[0154] (4)V i represents the volume of part i;

[0155] (5)V TRepresents the volume of the entire assembly;

[0156] (6) Represents the visibility value of part i at viewpoint j, which is used to measure whether the part can be considered as a hidden part;

[0157] In the hidden parts identification algorithm, the visibility value of a component at a certain viewpoint is calculated using the following expression:

[0158]

[0159] Where: λ1+λ2+λ3=1;

[0160] In the above formula, λ1, λ2, and λ3 are weights. This expression can effectively identify nearly invisible parts and subtle parts that do not affect the appearance of the assembly model. Initially, the part visibility rate, part area contribution rate, and part volume contribution rate are given the same weight, that is, λ1, λ2, and λ3 are all set to 1 / 3. In actual operation, they can be adjusted accordingly to achieve more satisfactory results based on the specific effects.

[0161] When part i completes the occlusion query at all set viewpoints, the maximum visibility value of the part can be obtained, that is, Define a critical parameter F according to actual needs T (0 <F T <1), when the F value of a component in the assembly is greater than the critical parameter, the component can be considered visible when observed from the outside of the assembly; otherwise, the component is considered hidden;

[0162] See also Figure 6 As shown, for the assembly model, the above-mentioned component visibility judgment algorithm is used to automatically identify the hidden parts in the assembly;

[0163] It should be noted that during the hidden parts identification process, the traversal of parts is a recursive process. When the selected component is a subassembly, on the one hand, it is necessary to perform a multi-viewpoint occlusion query test on the subassembly as a whole. At the same time, it is necessary to read the subassembly model, treat the subassembly as a new whole, and perform hidden parts identification to obtain the visibility information of the components in the subassembly.

[0164] For more complex 3D models, occlusion query is quite expensive. To improve recognition efficiency, the object's bounding box can be used instead of the object for occlusion query. During the lightweight process, the object's bounding box information can be obtained through the 3D system's development interface, eliminating the need to calculate the bounding box separately.

[0165] The input of this process is the mesh model generated during the lightweight process. Once the hidden parts of the components in the assembly are identified, the corresponding visibility information can be recorded in its lightweight assembly file.

[0166] See also Figure 7 As shown in the figure, after simplifying the hidden parts of the assembly model, the number of primitives that need to be drawn can be reduced, and the drawing speed can be accelerated;

[0167] The specific information of the engine model before and after simplification is shown in Table 1:

[0168] Number of parts Number of vertices Number of triangles Original Model 368 404422 424720 After simplification 152 317908 334655

[0169] Table 1

[0170] As can be seen from Table 1, after the hidden parts are identified and simplified, the number of parts in the engine is greatly reduced, resulting in a significant reduction in the number of model vertices and triangles. Of course, the effect of this operation is closely related to the model itself. If the model has a complex shape, a closed shape, and many subtle parts, the number of hidden parts identified is large. On the contrary, the simplified model cannot achieve a significant simplification effect. Since most of the parts identified as hidden parts are inside the assembly or are small parts (such as bolts, nuts, etc.), there is no obvious change in the appearance of the 3D model after removing the hidden parts. In addition, the lightweight assembly file only records the visibility of the parts in the assembly and does not remove these hidden parts from the product structure. Therefore, all parts can be displayed when needed, which will not affect the correct expression of the assembly model product structure and fully ensure the usability of the data model.

[0171] After simplification, many parts still exist inside the assembly. These parts cannot be observed from the outside, but they are not considered hidden parts. Some smaller parts are also identified outside the assembly. The main reason for this phenomenon is that the proposed data model is a multi-level structure. The overall assembly file only records the information of the next-level sub-assembly and parts, while the sub-assembly's components, the location of the part files, and other information are all recorded in the sub-assembly file. This file information organization method can reduce data redundancy and also bring convenience to users. For example, for a 3D car model, if the user wants to view the car engine, they only need to open the engine assembly file without loading the extremely complex car overall assembly file. In order for any level of sub-assembly model to meet the purpose of data exchange, during the hidden parts identification process, the visibility of the parts must be determined with respect to the upper-level assembly, and it is not possible to only focus on whether they are visible in the overall assembly. Therefore, further improvement of the simplification level is limited.

[0172] When rendering a 3D model, the graphics processing unit needs to read the model's vertex data from memory. However, mobile devices have limited memory bandwidth, so accessing vertex data takes a long time. Next, we will start from the model data representation layer. By changing the model triangle arrangement sequence during lightweighting, we can improve the GPU's vertex cache hit rate. This will make the generated data model have a rendering-optimized effect and shorten the time consumed by the mobile device GPU to access vertex data during drawing.

[0173] In order to alleviate the gap between data processing speed and access speed in the graphics rendering process, people added a vertex cache (VertexCache) to the graphics processor. The vertex cache stores vertex data that has been processed by transformation and lighting effects. If a vertex is not in the cache, it is necessary to read the data from the system memory or the graphics processor memory. For the vertex cache content that hits, the data can be used directly in the next stage, avoiding repeated geometry processing and effectively reducing the overhead of data reading. Researchers usually use the average vertex mismatch rate (AverageCacheMissRatio, ACMR) to describe graphics. The processor's vertex cache hit rate (ACMR) is numerically equal to the average number of vertices that each triangle needs to read from memory when drawing a frame. The lower the vertex mismatch rate, the fewer times the graphics processing unit accesses vertex data in memory, and the less time is required for vertex reading and geometric transformation during 3D model rendering. The maximum value of the vertex mismatch rate is 3, which means that all vertices of the model are not in the vertex cache. In a triangular mesh model, the number of triangles is nearly twice the number of vertices, so the theoretical minimum value of the vertex mismatch rate is 0.5. Due to the complexity of the topological relationship and the limitation of the vertex cache capacity, the theoretical minimum value can only be infinitely approached.

[0174] See also Figure 8 As shown in the figure, the vertex cache hit rate is not only related to the graphics processor hardware itself, but also closely related to the arrangement of mesh data. If the triangle sequence of the model is T0T1T2, then when drawing the T1 and T2 triangles, all the vertices of the two triangles need to be read from the memory again. If the triangle sequence of the model is T0T3T4, then when drawing T3, the two vertices of the triangle are already in the vertex cache, and the GPU only needs to read one more vertex. When drawing T4, only one vertex needs to be accessed. Therefore, the cache reuse rate can be improved by modifying the triangle sequence of the model and increasing the model locality. The mesh model converted from the solid model does not consider the vertex hit rate problem. The average vertex mismatch rate can be further reduced by changing the arrangement order of the triangles. Vertex cache optimization for the three-dimensional model can reduce the load of obtaining model data, and also reduce the amount of calculation required for rendering operations, thereby speeding up the drawing process.

[0175] In this regard, we have done a lot of work to achieve mesh rendering optimization. Hoppe used a greedy algorithm to reorder the triangle sequence of a triangular mesh, generating triangle strips with a very high vertex cache hit rate. Using a last-in-first-out data reference method, Hoppe's method was improved to obtain cache-optimized triangle strips with arbitrary cache parameters. The ACMR values generated by these two methods are very close to the minimum value, but the number of triangle strips generated by this method is large. Due to the deep hierarchy and large number of parts of the data model, if this strip generation method is used, the number of triangle strips in the entire assembly model will reach an extremely large order of magnitude, which will lead to excessive drawing function calls, causing the graphics system to overload and affecting the rendering speed. Bogomjakov et al. proposed a universal drawing sequence algorithm, which enables the obtained triangle drawing sequence to obtain a good vertex hit rate in a vertex cache of any capacity, but the speed is slow. Lin et al. performed a global search on the mesh based on a greedy method, which has many heuristic conditions and obtained a significant ACMR optimization effect on the mesh model, but its time complexity is high.

[0176] Sander et al. proposed a cache optimization method based on triangle rings. This method only performs heuristic searches near the previous result, achieving a good vertex cache hit rate while maintaining linear time complexity. However, it is a cache-aware algorithm. When the actual cache is smaller than the ideal cache, the vertex cache hit rate will drop significantly.

[0177] In practice, it is difficult to obtain the vertex cache parameters of the graphics processor. At the same time, in order to ensure the universality of the optimized model, the cache optimization effect should be as independent of the actual vertex cache parameters as possible. Therefore, based on the literature, a new heuristic search rule is adopted. Under the premise of maintaining linear time complexity, it not only ensures a good vertex hit rate in rendering, but also has adaptability to caches of different capacities.

[0178] The algorithm uses a vertex cache simulator with a capacity of s using a First-In-First-Out (FIFO) to simulate the situation of vertices entering and leaving the cache. In order to speed up the algorithm and maintain a high vertex cache hit rate, the algorithm draws on the idea of triangle ring operation. It continuously searches for the local optimal vertex as the Fanning vertex, and then outputs the adjacent unoutput triangles containing this vertex, and finally obtains a triangle sequence with a high vertex cache hit rate. The search is only performed on the candidate vertex. If it cannot be found there, it searches for other vertices in the vertex cache. If all vertices in the vertex cache have been output, then a non-output vertex will be randomly selected as the Fanning vertex to be output. The search process continues until all triangles in the mesh have been output.

[0179] The reason why the ACMR value increases dramatically when the ideal cache capacity is smaller than the actual vertex cache is that when the actual cache is smaller than the set simulation cache size, the originally assumed hit vertex will be replaced in the actual cache. The GPU must re-access the data in memory, and the later vertices in the triangle sequence will aggravate this phenomenon, causing the triangle sequence to be completely disordered, affecting the final rendering efficiency.

[0180] To overcome this problem, a new heuristic search rule is adopted, that is, the selected Fanning vertex must meet the following two conditions:

[0181] (1) After outputting the triangle ring of this vertex, the vertex must still remain in the vertex cache;

[0182] (2) The Fanning vertex must be the vertex with the highest priority value in its triangle ring. The priority value C(v) of the vertex is jointly evaluated by the position of the vertex in the cache and the degree of the vertex. The priority value C(R) of the triangle ring is the sum of the priority values of all the vertices in the ring that are not output.

[0183] The vertex priority value C(v) can be calculated by the following formula:

[0184] C(v)=C p (v)+C a (v)

[0185] in,

[0186]

[0187]

[0188] In the above formula, p represents the position of vertex v in the cache, s represents the size of the cache space, a represents the number of unoutput triangles adjacent to vertex v, and k1, k2, and k3 are coefficients;

[0189] In the calculation of vertex priority values, vertices that enter the vertex cache later have higher priority values. In this way, even if the capacity of the simulated cache is large compared to the real cache space, the vertex can still hit the vertex cache. However, if only the vertex position is considered, a long and narrow triangle strip will be generated, making it difficult for the vertices in the vertex cache to be reused by subsequent triangles. Therefore, the degree of the vertex is also evaluated in the search and selection process. If the number of adjacent unoutput triangles containing the vertex is small, the vertex is given priority as a Fanning vertex. This can increase the locality of the triangle sequence, so that the subsequently output triangles can fully reuse the vertices of the previously output triangles, reducing the possibility of generating long and narrow triangle strips.

[0190] See also Figure 9 As shown in the figure, the input data of the vertex cache optimization operation is the mesh model converted from the solid model. After the optimization process, the cache-optimized triangle mesh model is obtained. After encoding and compression, the final data model is obtained:

[0191] Step 1: Establish the adjacency relationship between mesh vertices and triangles, and initialize the vertex priority value;

[0192] Step 2: Select any vertex as the Fanning vertex;

[0193] Step 3: Output the triangles adjacent to the Fanning vertex that have not been output, use the vertices of these triangles as candidates for the next selection, update the contents of the vertex cache, and update the priority values of these vertices;

[0194] Step 4: Perform a heuristic search on the vertices in the candidate. First, determine whether the vertex is still in the vertex cache after the triangle ring is output. If multiple vertices meet the conditions, calculate the triangle ring priority values of these vertices and select the vertex with the higher priority value as the next Fanning vertex. If none of the vertices in the candidate meet the conditions, perform the same search operation on the remaining vertices in the vertex cache. If the Fanning vertex is still not found, it means that the triangles containing these vertices in the vertex cache have been output. Therefore, randomly select a vertex as the next Fanning vertex.

[0195] Step 5: When the number of output triangles is less than the total number of triangles in the model, go to Step 3: Otherwise, the algorithm ends.

[0196] The process of establishing adjacency relationships in the algorithm does not require recording edge information. The time consumption of this step is proportional to the number of triangles in the mesh. During the triangle output process, each vertex is a Fanning point at most once, and each triangle is accessed at most three times. The cache update operation is only related to the cache size s, so this operation consumes constant time. The time consumed by calculating the triangle ring priority value is related to the average degree of the model. For an ordinary model, it is a small constant, so the time consumption of this step is a constant. Overall, the time complexity of the algorithm remains at O(t), and the subsequent experimental results will prove this point.

[0197] The test platform used in the experiment is an Intel Core i3 2.13GHz CPU and 2.0GB RAM. The coefficients k1=0.3, k1=0.9, and k1=0.3 are set in the experiment. The vertex cache optimization algorithms compared are the triangle fan method (FanSort) and the general drawing sequence construction algorithm (Bog). The vertex cache size is set to 32 for both the triangle fan method and the algorithm. The reason for choosing these two algorithms for comparison is that the triangle fan method is fast and has good optimization effect. The algorithm is improved to address its difficulty in adapting to caches of different capacities. The general drawing sequence construction algorithm can achieve a good vertex hit rate in vertex caches of any capacity.

[0198] Model a is a gear model with 13,884 triangles. Model b is a sphere generated by modeling software with 20,480 triangles. Models c and d are both from the Stanford University 3D scan database. Model c has 49,954 triangles, while model d has 69,451 triangles. Model e is a motor model with 91,948 triangles. Model f is a plate-shaped part in a car's rear axle with 180,311 triangles.

[0199] From the data in Table 2, we can see that FanSort has the fastest processing speed. The algorithm processing time is proportional to the number of triangles in the model. It takes slightly longer than FanSort. The main reason is that the algorithm needs to calculate the priority value of the triangle ring to ensure good adaptability to vertex caches of different capacities. However, it is still many times faster than the BoG algorithm.

[0200]

[0201]

[0202] Table 2

[0203] Comparing the changes in average vertex mismatch rates when running at different cache capacities in actual rendering, the Fansort algorithm's ACMR value increases dramatically when the actual vertex cache capacity is smaller than the ideal vertex cache capacity. However, the cache mismatch rates of the Fansort algorithm and the BoG algorithm remain stable at different cache capacities, without any drastic changes. However, the optimization effect of the Fansort algorithm is slightly worse than that of the BoG algorithm.

[0204] In summary, the algorithm has a faster processing speed and a better vertex cache hit rate under any cache capacity, which can improve rendering performance.

[0205] Example 3

[0206] In the progress management model, guided by project management, operations research, and systems science, a 4D simulation management system and progress simulation model for the project progress are established. With the management system as the application framework and progress management as the driving force, on the basis of integrating the project progress with time, manpower, and materials, more points of integration between project progress management and BIM visualization are developed to strengthen process control during the construction phase. Based on this, a 4D digital progress management model that integrates BIM three-dimensionally, time, and resources is formed.

[0207] The basic content of project schedule management includes project schedule preparation and project schedule control. Project schedule preparation refers to the development of a reasonable and economical schedule within the specified time. Project schedule control refers to checking whether the actual progress is in accordance with the plan during the execution of the schedule. If there is any deviation, the cause is promptly identified and necessary remedial measures are taken or the original plan is adjusted or modified until the project is completed.

[0208] 1) The project schedule management process mainly includes steps such as activity definition, activity sequencing, activity resource estimation, activity time estimation, schedule preparation, and schedule control. Activity definition is generally based on project scope determination and work breakdown structure, and determines the specific activities included in the deliverables. Activity sequencing requires reference to the project scope description, activity list and attributes, and milestone list, and is completed through methods such as precedence diagrams, arrow diagrams, and condition diagrams to ultimately determine the dependencies and constraints between activities. Activity resource estimation is based on the determination of the project quantity, and the number of working hours and shifts required for each activity is calculated to determine the activity resource requirements;

[0209] Activity time estimation, based on activity resource estimation and logical relationship determination, estimates the duration required for each activity and further determines the start and end times of the activity. After completing the above work, the project schedule can be compiled and the target plan can be used to control the project schedule. The work of each process of project schedule management is carried out after the project team determines the preliminary plan. In the practice of project schedule management, there are no boundaries between different tasks, but rather they overlap and influence each other. Here, the schedule management process is clearly defined to facilitate theoretical analysis.

[0210] 2) After activity definition, activity sequencing, and time estimation are completed, project resources and other constraints must be comprehensively considered to determine the start and end times of project activities, implementation plans, and measures, and complete the preparation of the entire project schedule. The preparation of the project schedule can reasonably arrange the project time and ensure that the project targets are achieved. It is the basis for project progress control during the construction process and can provide a basis for resource allocation and time allocation;

[0211] 3) The project schedule has been compiled to schedule each activity. However, many unforeseen problems arise during the construction process, and deviations often occur during plan execution. Project managers are required to promptly correct these deviations and adjust the plan to ensure that the project is completed according to the contract requirements. Project schedule control refers to the activities of checking, comparing, analyzing, and adjusting the progress of the project after the project schedule is completed to ensure that the overall goals of the project schedule are achieved. Project schedule control is a cyclical, routine activity that mainly includes: determining a fixed reporting period, controlling the entire project execution process, and comparing the actual progress with the planned progress. If the project is delayed, exceeds the budget, or does not meet technical specifications, measures must be taken to get the project back on track. If the plan has been revised to reflect the changes, a new baseline plan must be established.

[0212] Dynamic monitoring of project progress mainly includes tracking and inspecting progress execution, organizing, compiling, and analyzing collected data, and comparing actual progress with planned progress. The difference between the actual and planned progress can be intuitively reflected using Gantt charts, network diagrams, S-curves, schedule tables, and other forms. Through comparison, it can be found whether the actual progress is ahead of, behind, or consistent with the planned progress. The process of project progress adjustment includes: analyzing the causes of progress deviations, analyzing the impact of deviations on subsequent work, determining the constraints that affect the total construction period and subsequent work, taking appropriate progress adjustment measures, and implementing the adjusted schedule.

[0213] With the continuous deepening of project management concepts, schedule planning technology has developed rapidly, supporting project schedule management and bringing great benefits to projects. Currently, commonly used schedule planning technology tools include Gantt charts, critical path method, and plan review and evaluation technology. The development of schedule planning technology has roughly gone through three stages: Gantt charts, network planning technology, and critical chain method.

[0214] Gantt charts are also known as bar charts and are widely used in construction period planning and scheduling. In a Gantt chart, project activities are arranged vertically on the left to show the work content. The horizontal axis represents the progress time, and the horizontal bar represents the project activity time. The bar chart can show the start time, end time, and duration of each work. The Gantt chart is simple, clear, and easy to read. It can clearly express the activity time difference and logical relationship of work tasks. It can be used at any level of the WBS. The time unit can range from years to days or even hours. In addition to being used to make schedules, Gantt charts can also be used as a progress control tool, showing the actual progress status in the form of a bar chart, and intuitively comparing the deviation between the actual progress and the planned progress as a basis for schedule adjustment. In addition, Gantt charts can be used for resource optimization, resource planning, and cost planning. Gantt charts are currently the most commonly used construction progress management method for substation projects.

[0215] The construction progress management system should take time planning as its starting point and integrate network planning technology to achieve comprehensive management of progress and resources. It should include project scope planning and responsibility allocation, comprehensive plan preparation and optimization, target management, plan feedback, analysis and update, reporting and information release functions, etc.

[0216] Project scope planning and responsibility allocation. Project scope definition and control enable project participants to reach a consistent understanding of what the project should and should not do. Scope definition can further decompose deliverables into smaller and more manageable work packages.

[0217] Comprehensive plan preparation and optimization, project schedule preparation, should be based on the construction schedule as the main factor, relevant functional departments formulate their own business plans, thus forming a comprehensive schedule plan including financial plan, design plan, marketing plan, procurement plan, etc., which includes most of the project management content. After the project department or contractor completes the preparation of the main construction plan, it can use the professional schedule management software to prepare a quality certification work plan for the construction work in combination with the project progress. The design review department can also formulate a corresponding drawing demand plan, and the procurement department can match it with the procurement plan. In addition, the quality inspection department or personnel can formulate a quality document submission / inspection work plan in combination with the procurement plan. The quality inspection department or personnel of the contractor can also prepare a quality inspection and related work plan based on the arrangement of the entire construction plan. The safety department can prepare an on-site construction safety environment plan;

[0218] Goal management: The establishment of a project goal plan facilitates the tracking and control of project progress. It can compare progress performance with the target plan, analyze progress status, identify deviations, and take timely corrective measures or update the target plan. Professional progress management software should have the function of establishing and maintaining goals, retaining the current plan as a goal, combining major goals (schedule, resources, costs) with project plans, allowing a project to have multiple target projects for comparison, and having visual project portfolio analysis and comparison functions. It can also set target monitoring values and conduct regular monitoring. It has warning reminders to facilitate the implementation of plan exception management, and key target information can be summarized layer by layer. Goals can be associated at different levels such as EPS, project, and WBS.

[0219] Plan feedback, analysis, and updates. Analysis is the foundation of control decisions and optimized management. Goals and actual feedback are the basis for analysis. The content and depth of progress analysis vary depending on the management level. Construction projects require progress analysis that includes: a summary of responsibilities, recent schedules, a summary of project completion status, including current period completion, cumulative completion, planned completion, the difference from the target plan, and progress trends. Resource analysis includes: current period figures, cumulative values, planned values, differences, labor productivity, resource utilization, and costs. Due to the diversity of goals, the timeframes for operational feedback confirmation vary. Different analysis cycle control measures can be used for different types of goals.

[0220] Report and information publishing functions include standard reports reflecting planned arrangements, standard reports reflecting actual progress, and standard reports comparing planned arrangements with actual progress. Web publishing is also available, allowing standard reports or graphics to be published as web pages.

[0221] The creation of project progress data is an important preparatory work for implementing BIM-based progress management. Project management software such as Project integrates functions such as Gantt charts and network plans. The power grid information model based on the GIM standard carries a large amount of available information about the substation. The creation of construction progress data can be based on a three-dimensional data platform and the application of WBS technology to decompose the project work structure, using project progress management software such as Project as a tool.

[0222] First, basic information about the project should be collected, and then the task details of the project should be determined. After the project documents are established, the input tasks can be prepared. The input task stage is usually called activity definition in the project management process.

[0223] Activity definition is usually completed under the guidance of the scope statement and work breakdown structure WBS. Depending on the project situation, you can create the WBS first and then the task list, or you can create the WBS and task list at the same time. There are many ways to enter tasks into Project software. You can directly add tasks as you think of them without considering whether they conform to the sequence and related task groups, and then adjust and organize them. You can consider the project from beginning to end in a sequential manner. Entering projects in a sequential manner allows you to consider the entire phase of the project first and then add tasks and subtasks.

[0224] You can also consider what work to be done in the project according to milestones and deliverables, and enter these tasks as tasks, thus establishing a task list;

[0225] After developing a task list, sorting the tasks, and establishing a task outline, and possibly customizing a Work Breakdown Structure (WBS) code, the next step is to develop a task schedule. The project schedule is a roadmap for completing tasks, submitting deliverables, passing milestones, and ultimately achieving project goals on time. To develop an accurate, feasible schedule that truly reflects the project's operational status, you need to enter task durations into the software, determine the relationships and dependencies between tasks, and schedule individual tasks as necessary to meet specific time requirements. This will give you an outline of a realistic project schedule. At this point, you can use the software to determine the duration of tasks and the duration of the entire project.

[0226] The completion of project tasks requires resource allocation. As tasks become clear and the schedule is established, resources must be actually specified in the project plan. By adding resources to the project, the accuracy of the schedule can be improved. It is also possible to understand in advance whether any resources are overloaded with too much work in the allocated time. The progress of the project can be tracked according to the resource hours, the use, cost and consumption of raw materials in the project, and confirmation that all tasks are assigned to responsible and reliable resources. With tasks and resources in the project plan, it is necessary to match the tasks and resources together and create "assignments". After personnel, equipment and material resources are assigned to tasks, Project can create a project schedule, which not only reflects the project calendar, task durations, dependencies and constraints, but also the calendar and availability of the assigned resources.

[0227] After defining the project scope, arranging the task schedule, and allocating resources, the project planning phase is not over. Usually, we need to review the results to determine whether the project plan meets expectations and requirements. If the project cannot be completed on time according to the plan, the project plan must be adjusted until it meets expectations and requirements. After making adjustments, the project plan still needs to be reviewed. Adding resources to tasks may advance the end time, but it may also increase costs. If more tasks are assigned to existing resources, these resources may be over-allocated. To save time and cost, individual tasks, deliverables, or phases may need to be abandoned. After making a balance based on project needs, the project plan can enter the execution phase.

[0228] See also Figure 10 As shown in the figure, the analytical model primitives are the material basis of the entire 3D information model. All component objects and their properties and operations are defined here. In other words, the core data of the model are concentrated in the model primitives. The model primitives are the ultimate carriers of the 3D information model data. Therefore, studying the data structure of the 3D information model is essentially studying the data structure of the model primitives.

[0229] In the power grid information model, the data structure of the model element includes basic data and extended data;

[0230] Basic data includes geometric data, physical data, functional data, etc. This type of data describes the characteristics and properties of the model components. It is inherent in the components themselves and will not change with time and environment changes. Extended data includes technical data, economic data, management data, etc. This type of data is mostly information or information related to model elements generated during the project management process. It is subject to the specific stage and specific environment of the project and does not describe the characteristics of the model elements themselves. It has a certain degree of independence. The power grid information model can integrate extended data to make BIM technology and function expansion possible. Integrating a large amount of indirect data related to model elements into the information model to form a complete and unique power grid information model can greatly expand BIM functions and applications and maximize the value of BIM technology. The same is true for integrating time data into the power grid information model to achieve progress management based on BIM technology.

[0231] Through the analysis of the 3D model structure based on BIM technology, it can be concluded that the extended data of the model element is the basis for realizing the functions of the power grid information model. If the power grid information model is to integrate other extended functions, such as schedule simulation and cost calculation, the corresponding extended data must be integrated into the model element. The establishment of the 4D model still uses the system structure of the 3D model. The main difference between the two lies in the difference in the model element data structure. On the basis of the 3D power grid information model, the construction plan data is integrated with the 3D geometric data and associated data of the model element to establish a 4D power grid information model and realize functions such as 4D simulation.

[0232] Applications designed with the MVC structure provide the ability to display multiple views of the same database. This structure is often used in the design of distributed application systems and can effectively separate the data layer from the table layer. Applications developed based on MVC are divided into three modules: the model layer, the view layer, and the control layer. The separation of the three allows the same model to be displayed in multiple views. When users change the model data through the control layer of a certain view, other views that depend on this model will be associated with the change. As long as the data changes, the control layer will notify all views of the change, so that the display is updated immediately. Combined with the system functional requirements, the 4D system based on BIM can be designed using the MVC model;

[0233] See also Figure 11As shown, the system model layer can implement the business logic in the system. After parsing the GIM three-dimensional information model through the three-dimensional platform, the spatial data of the model components is exported to the DWF three-dimensional model library, and the construction attribute data of the model components is exported to the SQL database. Through a platform similar to Project, the construction progress data is created and stored in the SQL database. Then, the association between the three-dimensional components and the two-dimensional WBS elements is established through the model component ID number, thereby realizing the integration of spatial data and progress data. The system view layer displays the model information in different views or display forms, displays the construction progress based on DesignReview, performs construction simulation, and uses the attribute table control to realize the display of specific information such as the component construction start time, end time, engineering quantity, and materials. The system control layer handles the interaction between the user and the system, so that the model and view work in coordination, and sets up an independent progress control module to facilitate the comparative control of the component construction start and end time. The system architecture is based on MVC.

[0234] See also Figure 12 As shown, the 4D system based on BIM needs to be established on the basis of project planning technology and 3D power grid information model technology, and be implemented by comprehensively applying database technology and system development technology. After analyzing the 3D platform and Project software, an MVC-based system architecture implementation process can be constructed.

[0235] Currently, the application of related technologies in schedule management is isolated. Although the application of a certain technology alone has brought great benefits to project management, it is far less than the benefits of integrated application of technologies. The emergence of BIM and its related technologies has brought great value and convenience to engineering project management, especially the creation, sharing and transmission of information throughout the project life cycle, which can ensure the effective communication of information. Only by integrating relevant information technologies and building a BIM-based schedule management system can we eliminate the drawbacks of traditional information creation, management and sharing, better realize the informatization of engineering project schedule management, and thus improve the efficiency of project management;

[0236] See also Figure 12 As shown in the figure, the timeliness, accuracy and accessibility of the progress information provided by the system are not high, which cannot meet the information needs of all parties involved in the project at all stages and is inefficient. Through the analysis of relevant theories and technologies of engineering project progress management, combined with the characteristics of BIM technology, its own advantages and derivative functions are integrated into progress management, and a progress management framework system based on the BIM technology platform is constructed to try to make up for the shortcomings of traditional management methods and a progress management application framework system based on BIM is constructed.

[0237] The BIM-based schedule management application framework system can intuitively display the improvement and perfection of schedule management methods and tools after the introduction of BIM technology. BIM-based schedule management is to introduce BIM technology into the existing schedule management system, aiming to comprehensively give play to the value of BIM technology and existing schedule management theories and methods. Since BIM technology models can carry the information required for the management of the entire project life cycle, the BIM information platform and functions generated by BIM technology are conducive to the entire process of project schedule management, and its benefits permeate all aspects of schedule planning and control.

[0238] Based on the existing progress management system, the project should establish the connection between BIM and WBS network plan with BIM information platform as the core, so as to comprehensively utilize various methods and tools to improve the progress management process and increase project benefits;

[0239] See also Figure 13 As shown, the core of the BIM-based progress management system is the BIM information platform. The BIM information platform can be divided into three subsystems: information collection system, information organization system and information processing system. The three subsystems are in a progressive relationship. Only when the work of the previous system is completed can the work of the subsequent system continue. The engineering project information mainly comes from project participants such as owners, designers, contractors, material and equipment suppliers, including all information related to progress management throughout the project life cycle. After the information collection system completes the collection of project information, the information processing system encodes, classifies, stores and models information in accordance with industry standards, specific rules and related requirements. The information processing system can use the system-structured information to support engineering project progress management, provide construction process simulation, construction plan analysis, dynamic resource management and site management functions, and the overall framework of the BIM information platform.

[0240] Example 4

[0241] In the comparative model, using analytical research, experimental verification and other research methods, by analyzing the real-time nature of information collection technology for construction site information collection and the integration of BIM into engineering information, and targeting the management needs and problems faced by the construction side of power transmission and transformation projects, the method of collecting on-site construction progress data and the reconstruction method of real-time model are studied, and a construction progress control method based on BIM is proposed;

[0242] After the critical chain is identified, the size of the buffer needs to be calculated. The size of the buffer is set based on the process safety time. Therefore, the next step is to determine the process safety time determination method used by the model.

[0243] There are many ways to determine the safety time. The following two methods are widely used. Each has its own advantages and disadvantages. The parameters in both methods can be adjusted to suit the actual construction situation:

[0244] Method 1: Establish an uncertainty indicator system for the project construction period, dividing the project into four periods, including the project planning and decision-making stage, the project preparation stage, the project implementation stage, and the project completion acceptance stage. The number of uncertainty indicators in different stages of the project varies, and the impact on the progress varies. A comprehensive assessment is conducted on the uncertainty indicators of different stages throughout the entire construction period of the project. By calculating the indicator system, the uncertainty level of each construction stage is calculated, and safety time is extracted from the processes of each construction stage according to a certain proportion.

[0245] Method 2: Assume that the time required to complete a process follows a Gaussian or lognormal distribution. The longer the time allocated to a process, the higher the probability of completion. If the activity completion time is estimated at a 95% probability, the time will appear to be particularly long. The extra time required to complete an activity at a 95% probability compared to a 50% probability is often used as the safety time for the project task.

[0246] Method 1 is more complicated than Method 2. It requires the construction of an uncertainty index system based on the characteristics of the project. If it is adopted in actual projects, it will inevitably increase management costs and time costs. However, the method of calculating the uncertainty of the actual project based on this system and extracting the safety time is more scientific and more reliable. Method 2 is simpler, easier to operate, and basically conforms to the actual construction. However, due to the differences and mutual influence of each process, this method also has certain limitations.

[0247] Due to the simplicity and ease of use of Method 2, it was decided to select Method 2 for the process safety time extraction method. On this basis, a statistical distribution that meets the characteristics of the project process duration was selected;

[0248] The statistical distribution of process duration can mainly be normal distribution and beta distribution, while uniform distribution and triangular distribution are less commonly used. Although these distributions are not as common as normal and beta distributions, they still exist with a certain possibility.

[0249] When using statistical distribution to simulate the construction period, the construction period generated by different distribution simulations will have a significant impact on the results. Therefore, each process should try to choose a statistical distribution that conforms to the actual situation. MATLAB will be used to simulate and analyze the construction progress. The characteristics of the above four statistical distributions are introduced respectively, as follows:

[0250] (1) Normal distribution. Normal distribution plays a very important role in both theory and reality. It is the basis for analyzing other distributions. When the process duration distribution is normal, its mean, median, and mode coincide. The probability that the process duration is greater than or less than the mean is exactly equal, and the probability that the process is delayed or advanced is exactly equal.

[0251] (2) Beta distribution. Generally speaking, the Beta distribution is an asymmetric unimodal distribution with clear endpoints. Depending on the values of its parameters α and β, the distribution has four variations: left-skewed distribution and right-skewed distribution. Its distribution range is between [0, 1]. The distribution value is relatively stable and is suitable for simulation estimation of process duration.

[0252] (3) Uniform distribution, also called rectangular distribution, is a symmetrical probability distribution. The probability of distribution in intervals of the same length is equally likely. When the duration of a process is uniformly distributed, the time it takes to complete the process is equally likely within a certain range. The probability of occurrence during the construction process is low.

[0253] (4) Triangular distribution. Triangular distribution is determined by three parameters: the lower limit, the mode, and the upper limit. It has both discrete and continuous distributions and is often used in business decision-making, especially in computer simulations.

[0254] In actual on-site construction, estimates of process durations are often very generous. This is because there are multiple layers of hierarchical relationships among the various construction entities. Whenever superiors assign tasks to subordinates, they often take into account the various uncertainties of the tasks and add a certain amount of safety time to the tasks. When the multiple layers of safety time are added together, the actual safety time given is quite considerable. However, despite giving so much safety time, the duration of most processes in actual construction still shows a left-skewed distribution. People often complete the work when the planned time is about to run out, or wait until the planned time is up before reporting the process results. In short, most of the safety time in the process is wasted. According to the concept of critical chain project management, the process time will always automatically expand until it occupies all available time. The more time there is, the more time it takes to complete a task. This view explains the above phenomenon.

[0255] Therefore, ignoring the influence of human psychology and unhealthy behavior, and reducing process safety time, the distribution of process duration should follow a right-skewed distribution. Assuming the total time of a process is 1 day, the actual completion time of the process in construction is roughly 0.7-0.9 days, following a left-skewed beta distribution. Without considering the influence of human factors, the process completion time should be 0.5-0.7 days, following a right-skewed beta distribution. According to the concept of critical chain project management, to eliminate the influence of human factors, the process safety time should be removed. Therefore, when simulating process duration, the statistical distribution of process duration should adopt a right-skewed distribution. The safety time reserved at each level will be treated as a buffer. However, it is impossible for a process to be completed in 0 time, and no matter how much safety time is added, it is difficult to account for half of the time of a task. Therefore, the right-skewed beta distribution graph should be shifted to the right of the coordinate axis, so that the average process completion time is roughly 0.5-0.7 days, and there is a certain possibility that it may take more than 1 day to complete the process.

[0256] Based on the above, it is believed that the process duration should be estimated by using a Beta right-skewed distribution with a non-zero starting value. At the same time, the selection of Beta distribution parameters should have certain requirements. First, it is necessary to ensure that the Beta distribution is right-skewed. Second, since each process has its own characteristics and different degrees of uncertainty, the estimation of its duration can also be differentiated by the selection of Beta distribution parameters.

[0257] As can be seen from the above, there are four types of Beta distribution graphs: descending waveform curve, ascending waveform curve, left-skewed curve, and right-skewed curve.

[0258] In probability theory, the beta distribution, also known as the β distribution, refers to a set of continuous probability distributions defined in the interval, with two parameters α and β>0. Its important properties are shown in the following formula:

[0259] (1) Mean:

[0260] (2) Variance:

[0261] It can be seen that the above two equations are a set of two-variable quadratic equations with infinite solutions. If α and β are known, the mean and variance can be calculated respectively. However, it is difficult to calculate the values of α and β when the mean and variance are known. Through MATLAB design simulation, the value range of α and β is specified as 0:0.1:10000. Since the required distribution graph is a right-skewed graph, the value range of the mean is specified as 0.3:0.05:0.5, and the variance values are calculated respectively. The simulation results are stored in the A11 to A51 matrices, with the first column as the α value, the second column as the β value, and the third column as the variance value. Then, through the data of the A11 to A51 matrices, the appropriate mean and variance can be selected as needed, and the corresponding parameter combination can be selected to simulate the process duration. At the same time, since the conditions for the appearance of the downward waveform curve and the upward waveform curve of the Beta distribution are unclear, the selected parameter combination should be simulated through MATLAB, and its distribution graph should be plotted. It should be applied when it is confirmed to be a right-skewed distribution.

[0262] Through MATLAB design simulation, it can be seen that the downward waveform curve and the upward waveform curve will only appear when the variance is large: when the mean is less than 0.5 and the variance is large, a downward waveform curve will appear, and when the variance is small, a right-skewed waveform curve will appear. When the mean is greater than 0.5 and the variance is large, an upward waveform curve will appear, and when the variance is small, a left-skewed waveform curve will appear. The larger the variance, the greater the curve fluctuation. Therefore, the peaks of the downward waveform curve and the upward waveform curve may appear within the range of curve fluctuation, and it is difficult to see its skewness. Its exact form will no longer be explored. The right-skewed beta distribution with a smaller variance is mainly used. At the same time, if the variance is too large, it is not suitable for the simulation of process duration, and the stability is difficult to meet actual requirements.

[0263] According to engineering practice, the longer a process lasts, the greater the uncertainty of the process and the more likely it is to be delayed. By dividing the process duration, three parameter combination options are given, corresponding to the degree of uncertainty of the process, so that the Beta distribution can better fit the actual situation. Details are shown in Table 3.

[0264] Degree of process uncertainty Process duration (α，β) mean Variance value big t>80 days (3，3) 0.5 0.0357 middle 40 days ≥ t > 10 days (2，3) 0.4 0.04 Small t≤10 days. (1.5，3.5) 0.3 0.035

[0265] Table 3

[0266] Corresponding to the above three situations, when simulating the processes in the project network with Beta distributions of different parameters, to avoid the process simulation time being too short and inconsistent with actual on-site construction, the Beta graphs of the above three parameters are shifted right by 0.3 units, and the Beta random distribution number is set to be no less than 0.15. This ensures that the minimum duration of each process is at least 45% of the original duration. In this case, the completion time of the process is between 60% and 80% of the original duration, and there is a certain probability that it will exceed the original duration, which is more in line with the actual situation and has greater uncertainty.

[0267] In summary, it is assumed that all processes in the project follow a right-skewed beta distribution and are independent of each other. The safety time is the difference between the duration of the process under a 95% completion probability and the duration under a 50% completion probability. The formula is as follows:

[0268] T k =[F(x)I x＝0.95 -F(x)I x＝0.5 ]×t k

[0269] Where F(x) is the distribution function of the Beta distribution, T k is the safety time of process k, t k is the duration of process k;

[0270] BIM-based schedule planning and control integrates BIM, WBS, and network planning technologies. The schedule is compiled based on BIM model information and uses the traditional schedule management software interface to establish a work breakdown structure, estimate construction duration, and arrange work logic relationships.

[0271] Before compiling a project schedule, the project scope management and work definition must be completed first. WBS is currently considered one of the most important tools for planning and controlling project work content and scope. Similarly, the first step in BIM-based project scheduling is to establish a work breakdown structure, which is generally completed with the assistance of relevant software or systems.

[0272] WBS establishment and coding is the key to linking BIM model component element information with work progress, resources, costs and other information, thereby realizing the direct application of BIM model information in progress management;

[0273] The project work breakdown structure is to decompose the project goals, tasks, scope of work, and contract requirements into independent, mutually influential, and interconnected project units according to system principles and requirements, and use them as a series of project management work objects such as project planning, implementation, control, and information transmission. Through project management, all project units are combined into a whole work to achieve comprehensive planning and control requirements. In essence, it is a top-down, layer-by-layer decomposition expression of project tasks, so that each task is arranged in the appropriate position of the entire project structure;

[0274] WBS is the basis for compiling time plans in project management. The work breakdown structure method uses system thinking to decompose the project to prevent missing items in the project network plan. It estimates the working hours and duration of each decomposed unit, allocates resources, establishes overlapping relationships, and optimizes time to achieve optimal management of the project time plan. Time control during project implementation is also based on the decomposed units.

[0275] WBS creates conditions for computer-aided project management. The application of computers in project management is a hallmark of modern project management. During the project implementation and control process, WBS coding system is used to collect and classify resources and progress information, ensuring that the information is annotated using a unified and consistent annotation method. By inputting the names and overlapping relationships of work units at each level into the computer, bar charts, network diagrams, and 4D models can be automatically generated. During the project implementation process, the project progress status can be checked at any time, and the computer automatically generates periodic progress reports.

[0276] From the creation process, the work breakdown structure (WBS) is to decompose a functional entity (project) into sub-projects according to a tree diagram, and then decompose it into several relatively independent work units step by step, and determine the tasks of each work unit and its subordinate work (or activities) in order to organize the project more effectively.

[0277] In the BIM-based schedule management system, WBS compilation is realized by computer. For obvious and necessary work for the project, the information in the BIM model can be directly used to view the model or export the detailed list for confirmation;

[0278] For tasks that are somewhat hidden, it is necessary to list the tasks required for a complete project based on experience or by referring to previous project practices. Through the computer software system, the WBS elements should be input and coded, and the connection between project tasks and BIM model components should be established.

[0279] A BIM-based scheduling system allows you to add an unlimited number of WBS milestones, which can also be used to calculate earned value. Milestones are assigned at the WBS level, and each milestone is assigned a weight to indicate its importance to the project schedule. When a WBS milestone is marked as completed, the module uses its weight to calculate the completion percentage of all activities contained in the WBS level. The completion percentage is applied to all activities under the WBS level and then aggregated to the WBS. For example, if a specific WBS level contains 10 activities, and actual completion dates are entered for 5 of the activities, and 4 WBS milestones with the same weight are assigned to the WBS level, but only one of them is marked as completed, the module will use the completed WBS milestone to calculate the completion percentage of the WBS level, which is 25%, even if the activities contained in the WBS level are already 50% complete.

[0280] If a WBS element has a total of 4 weighted milestones, and all milestones have a weight of 1.0, marking one of them as "Completed" means that the WBS element is 25% complete. If the milestone has a weight of 9.0 and the other 3 have weights of 1.0, marking the milestone as "Completed" means that the WBS element is 75% complete. The module uses the following formula to calculate the completion percentage based on the weighted milestones: Actual weight of the completed milestone / Total weight of all milestones. Applying this formula to the above example, the weight of the completed milestone is 9.0, divided by the total weight of all milestones, 12.0, the result is 75% completion percentage.

[0281] WBS Earned Value BIM-based systems can define earned value settings for specific WBS elements, thereby implementing WBS earned value analysis. Earned value is a method of measuring project performance based on project costs and schedule. This method compares the budgeted cost of work with the actual cost. Earned value analysis is commonly applied to WBS elements, but it can also be performed on activities or groups of activities.

[0282] Using the earned value feature of the work breakdown structure, you can specify the settings for calculating earned value for selected WBS elements. Earned value refers to the total budgeted cost of activities actually completed since the start of the project data date. It is calculated as follows: Earned value = Budget at completion × Execution completion percentage. The calculation method for completion percentage depends on the earned value method selected for the WBS for the activity.

[0283] A project schedule is a roadmap for completing tasks, submitting deliverables, achieving milestones, and ultimately achieving project goals on time. Developing a project schedule is closely related to time management. To develop an accurate, feasible schedule that truly reflects the project's operational status, you need to determine activity durations, the logical relationships between activities, allocate resources, estimate costs, and set a budget.

[0284] After the work breakdown structure defines the project activities, it is necessary to estimate the time for each activity. Activity time refers to the duration of the activity. Estimating activity time is the core of time planning. Estimating activity time is not simply based on mathematical calculations, but also needs to be adjusted according to the capabilities of the project team and the available professionals, equipment, and funds. Activity time estimation needs to consider internal and external factors that affect the construction period. It can be done by combining experience, historical data, research, Delphi method, modeling, etc.

[0285] In BIM-based scheduling, when completing the work breakdown structure, the WBS code is associated with the model component ID number. By selecting a task, the basic data information of the corresponding model component can be viewed. Therefore, the task time can be estimated by using the geometry, function and other data of the model component, through the acquisition of engineering quantity information, combined with specific calculation methods. The task duration can be estimated by referring to the experience and historical information of other project models in the system database;

[0286] The BIM-based progress system allows users to view and edit detailed progress information for selected tasks, including planned start and finish dates, actual start and finish dates, free float, total float, constraints, planned duration, and actual duration. It also allows users to view and edit labor and non-labor unit costs, as well as material costs. Furthermore, the system provides PERT analysis to aid in duration estimation, allowing users to set optimistic, expected, and pessimistic durations. The system automatically calculates the duration using a weighted average. Editing detailed progress information for tasks in the system also directly reflects changes in the extended data of the model element.

[0287] Once the activity durations are determined, the next step in creating the project schedule is to establish logical relationships between activities to indicate whether one activity can only begin after another has started or completed. After assigning these logical relationships, the earliest and latest dates for each activity are calculated through the project schedule.

[0288] There are usually four types of logical relationships between predecessor activities and successor activities:

[0289] Finish-Start (FS): The successor activity can only start when the predecessor activity is completed.

[0290] Finish-to-Finish (FF): The completion of the successor activity depends on the completion of the predecessor activity;

[0291] Start-Start (SS): The start of the successor activity depends on the start of the predecessor activity;

[0292] Start-Finish (SF): The successor activity can be completed only when the predecessor activity starts.

[0293] In cases where the successor activity cannot start or finish at the same time as the predecessor activity starts or finishes, a delay can be defined for the relationship. The delay is the amount of time between the start or finish of one activity and the start or finish of the successor activity. The delay can be positive or negative. For example, a start-start relationship with a three-day delay means that the successor activity can only start three days after the predecessor activity starts.

[0294] There are many ways to assign logical relationships in a BIM-based schedule. You can use an activity network diagram to visually display the logical flow of connected activities, or use a bar chart to view the logical relationship based on time. You can also directly select a WBS activity to assign the relationship to other activities in the project. After completing the setting of the logical relationship, the network diagram and bar chart are completed. You can select multiple activities through the system to apply the network plan or bar chart for plan analysis, and view the four-dimensional dynamic simulation of the selected activity.

[0295] As project activities become clear and the schedule is established, resources must be actually identified in the project plan. By adding resources to the project, the accuracy of the schedule can be improved. The progress of the project can be tracked based on resource hours. The use, cost, and consumption of materials in the project can be tracked to ensure that all tasks are assigned to responsible and reliable resources. Resources include both manual and non-manual resources required to perform all project activities, such as engineering dollars and equipment. These are usually calculated by time and are often allocated to other activities. Material resources, such as supplies and other consumables, are calculated by unit price rather than by hour.

[0296] The BIM-based schedule management system can add resources and create a resource hierarchy to reflect the organization's resource structure, and support the allocation of these resources to activities. It can set resource classification codes without hierarchical restrictions for grouping and summarizing resources, and generate resource reports and overviews to analyze resource allocation and adjust project plans to avoid over-allocation of resources and peaks and troughs in resource usage. In addition, the system can compile resource plans to integrate resources, costs, and schedules for effective project control. After the system allocates resources and specifies the budgeted amount expected to be used in activities, you can use "activity analysis tables," "resource analysis tables," "activity histograms," and "resource histograms" to track their usage during the project.

[0297] After the initial completion of the schedule, the plan needs to be analyzed to confirm its rationality. The analysis of the high-level plan mainly includes: whether the project plan content is comprehensive, whether the activities are detailed at the minimum level of the WBS, whether it is easy to control the plan, and other contents such as key resources, cost allocation, activity duration, logical relationship between construction processes, construction process constraints, construction process time difference, major process handover points, and whether work products and documents are allocated to the corresponding WBS and activities. The BIM-based schedule management system provides network optimization, schedule target comparison analysis, earned value analysis, 4D simulation analysis and other functions. The results of various analyses are very helpful in reasonably determining project control objectives.

[0298] See also Figure 14 As shown, the schedule duration analysis should first review the consistency of major milestones to ensure that the overall goal is consistent with the contract date. Secondly, identify the critical path and analyze whether the duration of work on the critical path is acceptable. In addition, under comprehensive consideration of the constraints, whether the start milestone date and completion milestone date meet the project requirements. Through the hierarchical plan developed by multiple parties, analyze and check whether there is a significant deviation between the work duration of the summarized plan and the original plan. Project participants should communicate with each other and reach an agreement.

[0299] See also Figure 15 and Figure 16 As shown, the BIM-based progress management system provides WBS work list view, bar view, network view, and 4D view to analyze the rationality of work duration and logical relationships. Through the bar view or network view, you can easily find the key paths and key tasks and display them in different colors. If you need to adjust the key paths and key tasks, you can modify and edit them directly in the view. In addition, through the 4D view, you can directly simulate the work, which can more vividly show the progress of the work and find out whether the logical relationship between the tasks is rational.

[0300] For time-constrained projects, completion time should take precedence over other factors, but budget constraints and project scope must still be balanced. To advance the project's completion date, the schedule itself should be reviewed to confirm the accuracy and necessity of all pre-set schedule control factors. In the activity list or Gantt chart, date constraints, durations, activity dependencies, and activity calendars should be reviewed and updated. To influence the completion date, only critical activities need to be adjusted. This can shorten the critical path activity sequence, making another activity sequence the current critical path and ensuring that this path ends before the target completion date.

[0301] The use of resources is the most important constraint in a project. After resources are assigned to work, it is necessary to check the workload and work resource allocation to ensure that all resources are well utilized and allocated to appropriate tasks, avoiding under-allocation or over-allocation. When resources are under-allocated, resources do not directly perform project work, resulting in missed deadlines, reduced quality, and increased costs. When resources are over-allocated, resources need to continuously overwork, working hours beyond the available time, and will be charged more for the hours worked.

[0302] Before the work begins, you can review the schedule and analyze resource allocation to identify under-allocation or over-allocation, and make necessary adjustments to maximize resource contribution and reduce the risks brought by resource allocation. The purpose of the review is to maximize the workload, reduce the waste of resource costs, and make the best use of key resources. The BIM-based schedule management system provides resource allocation analysis functions. By viewing the detailed information of the selected resources in the resource worksheet, you can check the resources to determine whether the actual demand for resources in the project can be reflected by modifying the availability. You can balance the resource load by adjusting the project scope, adding or reducing WBS work items, using a limited delay to start a resource, or by adjusting the duration, splitting or delaying the work until the resource has time to execute. You can also adjust the budget and invest more money to increase resources. Under-allocated resources can be added to the work to reduce or replace over-allocated resources, provided that they have the same skills and availability. Resource allocation analysis;

[0303] By placing cameras at various locations on the construction site, real-time progress information can be collected and integrated into images used to generate a real-time model. These images can then be used for progress management. This requires addressing the following issues: First, 3D reconstruction of existing buildings on the construction site to form a real-time model. This model is then compared with the planned model, and finally, progress statistics are calculated. 3D reconstruction requires the following process:

[0304] 1) Image registration is a process in which a camera takes a large number of photos from various directions of the construction site, selects pictures taken from different angles at a certain point in time, forms a group, and performs feature point matching on the pictures in this group. The so-called feature points are points that can reflect the initial structure of the scene. Feature point detection can use SIFT feature detectors. Feature detectors are usually called corner detectors, but they do not only select corner points, but select any image position with a large gradient in a predetermined ratio in all directions. The SIFT feature detector is used because it has good invariance and can scale, view and illumination changes, and has a wide range of applications in the field of computer vision. SIFT is translated as scale-invariant feature transform, which can convert image data into coordinates that are invariant to the scale of local features. The steps for forming SIFT image feature points are as follows:

[0305] 1. Scale space extrema detection, searching all scales and image locations, and identifying potential points of interest for scale and rotation by using Gaussian derivative functions;

[0306] 2. Keypoint positioning: At each candidate location, a fine-tuned model needs to be fitted to determine the position and scale. Keypoints should be selected based on their stability.

[0307] 3. Determination of direction: one or more directions are assigned to the position of each key point according to the local gradient direction of the image. All subsequent operations on the image data are transformations relative to the direction, scale, and position of each feature, thus providing invariance to these transformations.

[0308] 4. Keypoint description, in which the local gradient of the image is measured at a selected scale in the neighborhood around each keypoint. These gradients are transformed into a representation that allows for relatively large local shape changes and lighting changes;

[0309] The acquisition of feature points by SIFT will form a large number of features, which will densely cover the image in the entire scale and position range. Usually, the number of feature points acquired depends on the content of the image and the selection of various parameters, but there is a typical size, that is, an image of 500*500 pixels will produce about 2000 stable feature points. In order to detect small objects in a cluttered background, at least 3 features need to be correctly matched from each object for reliable recognition.

[0310] 2) After the feature points are detected, image matching and recognition begins. The method is as follows: First, SIFT feature points are extracted from a set of reference images and stored in a database. The new image is matched by comparing each feature point in the new image with the previous database and then finding candidate matching features based on the Euclidean distance of the feature vectors;

[0311] After detecting feature points on the collected image dataset, these feature points need to be matched. The matching method is SIFT descriptor, which captures the intensity gradient on the pixel window centered on the detected key point. Then, these pixels are classified into 4*4 sampling windows, where each sampling window and intensity gradient histogram are stored in 8 basic directions. Then the descriptor becomes a feature vector of 4*4*8-128 dimensions. SIFT descriptor matches feature points by calculating the distance function of two matching intensity gradient histograms. Feature point matching is done by using the nearest neighbor matching method. If the number of features is large enough, the KD tree matching scheme can be used instead. In order to minimize the amount of calculation, the ANN priority search algorithm can also be used.

[0312] False matches may occur when matching features. To eliminate these errors, a ratio test is used. The specific method is as follows: for a feature descriptor in image i, two nearest neighbors are found in image j, with distances d1 and d2, respectively. If d1 / d2 < 0.6, then it is the match we are looking for. If multiple features in image i match the same feature in image j, since one of them is a false match, these false matches will be deleted.

[0313] Due to the sensitivity of the reconstruction algorithm to false matches, the debugging method needs to be further improved. That is, once matching features are detected in an image pair, the fundamental matrix of the image pair is robustly estimated using RAXSAC. The fundamental matrix enforces that the corresponding features must be consistent under viewpoint transformation, that is: P,'FP,-0 (P, and P, are point coordinates, and F is the fundamental matrix). In our model, the eight-point algorithm is used to calculate the fundamental matrix in each iteration of RANSAC, and then the problem is normalized to improve robustness to noise;

[0314] The RANSAC outlier threshold is set to 0.6% of the maximum image size, which is 0.006 times the maximum width or height of the image (approximately 12 and 9 pixels for a 2144*1424 two-dimensional image size). The Levenber-Marquardt algorithm is run on the eight parameters of the basis matrix to refine the basis matrix returned by RANSAC. The above recommended threshold is used to remove false matches.

[0315] If there are few matches in the image pair, there is still a high false match rate even after the basic matrix is fitted through RANSAC iterations. Therefore, a threshold of 20 matches is set. If the number of matches in the image pair is less than the threshold, the match has no reference value and is deleted. After finding consistent matches between all image pairs, these matches are organized into tracks. The tracks connect the key points across multiple images for the next stage of 3D reconstruction. In the next stage of 3D reconstruction, at least two key points are tracked.

[0316] 3) SFM, which stands for Structure from Motion, aims to reconstruct unknown 3D scene structures and estimate the position and orientation of unknown cameras based on a set of feature correspondences between images. Bundle adjustment is a key tool for obtaining powerful 3D reconstruction from a large number of sparse images. It is robust to image resolution, time, focal length variability, and illumination changes.

[0317] First, estimate the external parameters and intrinsic parameters of an image pair. Since bundle adjustment is prone to falling into bad local minima, many researchers suggest that the initial image and camera parameters should be well selected, and the SPM initial image pair should not only have a large number of matches but also a large baseline so that the initial as-built scene can be reliably reconstructed. Image pairs that are difficult to describe with homography transformations meet this condition. 2D image homography is a projective transformation that maps points from one plane to another. Use RANSAC to find homographies between all image pairs with an outlier threshold of 0.4% of the maximum image width and height, and store the percentage of feature matches that are outliers from the estimated homography.

[0318] Select an initial image pair that has the lowest percentage of pixels with the recovered homography image, with at least 100 matches. Estimate the extrinsic parameters of the camera pair using Niste's five-point algorithm, then triangulate the tracks visible in the image pair. After two-frame bundle adjustment on this initial pair, add another camera to the optimization and select the camera that checks the maximum number of estimated tracks. Then use the direct linear transform (DLT) technique to initialize the new camera extrinsic parameters. For this RANSAC step, use an outlier threshold of 0.4% of the maximum image width or height. For the estimation of the focal length of the new camera and the intrinsic camera matrix, the focal length of the Exchangeable Image File Format tag of the EXIF-JPEG image can be used for initialization.

[0319] Starting from this set of initial parameters, while the model remains unchanged, a bundle adjustment algorithm is run, allowing only the new camera and the key points it observes to change. Finally, the points observed by the new camera are added to the optimization algorithm. When at least one existing restored camera observes the key point, the key point is triangulated. After triangulation, the position of the camera can be well estimated. The condition is estimated by considering all ray pairs that can be used to triangulate the key point and finding the ray pair with the maximum separation angle. If this maximum separation angle is greater than a threshold, the point is triangulated. Once the new key point is added, another global bundle adjustment is run to reconstruct the new scene. Finally, the solution error is minimized through the sparse bundle adjustment library. This process is performed for all cameras until no camera can observe more reliable 3D reconstructed points. In this way, the construction site model is reconstructed.

[0320] 4) There are three ways to display the progress model generated by the progress management system. One is to display the as-built model on the planned model with color difference, or to display the as-built model in the same way as the planned model in the traditional BIM platform. Another is to superimpose the virtual model on the construction site, also using color to distinguish the progress, so that managers can see the construction progress very intuitively. The third is to display the project progress by the actual appearance of the completed building. The specific implementation process is described below.

[0321] The reconstructed model is aligned with the planned model to determine the absolute geographic coordinates of each camera. The absolute coordinates are estimated by translation, rotation and uniform scale transformation of the image. The estimation method uses the closed form solution of the absolute orientation of the unit quaternion, and then the reconstructed model can be registered with the planned model. The planned model is registered as the built scene. The sparse 3D point set reconstructed in real time is registered with the control point set of the planned 3D model to minimize the residual sum of squares between the reconstructed point set and the model. After alignment, the model reconstructed by the camera will be superimposed on the site under construction. In this way, from the perspective of the model reflecting the progress:

[0322] The first method is that on the day of the inspection, the manager can see a virtual 3D image of the planned model superimposed on the corresponding location of the construction project on the construction site. If the progress plan is fully completed on the inspection day, it will be displayed in dark green. If any progress plan is under construction but not completed on the inspection day, then this part will be displayed in light green. If any work has not started on the inspection day, then this part will be displayed in red.

[0323] The second way is to display the real appearance of the building. The BIM progress management system can display the planned model in its real-world appearance. To achieve this effect, the following data are required: a set of key points, each of which contains a 3D position and a color composition averaged from all on-site images where the key point is located; a set of cameras with known external parameters (translation and rotation) and intrinsic parameters (focal length and distortion in height and width); and a mapping between each key point and all cameras that observe the key point. The mapping can store a list of cameras that obtain key points, the position of the point in the local coordinates of the image, and the SIFT key point index. While the system stores these data, the camera will be rendered into a cone of view. By accessing the camera in the reconstructed scene, the camera cone of view will be texture mapped into a resolution image. After the planned model is aligned with the reconstructed scene, the planned model will cover the reconstructed sparse scene and display the covered reconstructed sparse image with the image of the planned model. In this way, what is presented to the manager is the real appearance of the completed building.

[0324] The third method is the three-dimensional display method of traditional BIM progress management, that is, on the BIM platform, the progress is reflected by displaying color differences on the plan model.

[0325] No matter how detailed the plan is, it is impossible to foresee all possibilities. Deviations will still occur during the implementation of the project plan. Tracking project progress and controlling project changes are the main tasks of the implementation phase. After the BIM-based schedule is completed, the project enters the implementation phase. The implementation phase mainly includes tracking, analysis and control. Tracking the progress of the work, actually understanding when the allocated resources will complete the tasks, checking the deviations between the original plan and the actual progress of the project, and predicting potential problems: taking necessary corrective actions to ensure that the project moves forward steadily within the constraints of the completion deadline and budget;

[0326] During the schedule planning phase, a BIM-based schedule management system uses a combination of technologies such as WBS, Gantt charts, network planning, and BIM to complete schedule arrangements, allocate resources, and budget costs. During the implementation phase, the BIM-based schedule management system provides features such as progress curves, Gantt charts, and 4D simulations to track and control project progress.

[0327] After analysis and adjustment, the project plan achieves a balance between scope, schedule, and cost and can be used as a target plan. Project activities all define progress information such as the earliest start time and the latest start time, so the system can provide multiple target plans to facilitate progress analysis. The project target plan cannot remain unchanged and needs to change as the project progresses. After tracking the project progress for a certain period of time, the deviation between the target progress and the actual progress will gradually increase. At this time, the original target plan will lose its value and need to be recalculated and adjusted. After entering the corresponding progress information in the system, the project plan will automatically calculate and adjust to form a new target plan.

[0328] The BIM-based schedule management system provides the creation and update of target plans and can also assign target plans to each job. When updating the target plan, you can choose to update all jobs or use filters to update jobs that meet the filter conditions. You can also specify the data type to be updated. After the target plan is updated, the system will automatically calculate the project schedule and balance resource allocation to ensure that resource requirements do not exceed resource availability. During the balancing process, the system uses the resource requirements of all calculated jobs as the maximum available quantity during the balancing process. If there are too few available resources during the job duration, the job will be delayed. After selecting the resources to be balanced and adding the balancing priority, you can specify the project or job that will be balanced first in the event of a conflict. In addition, after changing the resource information, the cost needs to be recalculated based on the engineering quantities provided by the BIM model to obtain the correct job cost value;

[0329] After the project plan is created, it is necessary to continue tracking the project progress. The BIM-based schedule management system provides multiple tracking views, including project tables, Gantt charts, network diagrams, schedule curves, four-dimensional models, resource curves, and histograms. The project table displays project data in a tabular format. The project bar chart displays project data in a horizontal "bar chart" format. The project bar chart / histogram displays project information in a column and "bar chart" format, and displays time-proportioned project data in a spreadsheet or histogram format. The four-dimensional view dynamically displays the building construction process in the form of a three-dimensional model. The resource analysis view displays resource / project usage information in a column and "bar chart" format, and displays time-proportioned resource allocation data in a spreadsheet or histogram format.

[0330] All tracking views can be used to review the project, first in a comprehensive manner and then in more detail based on the work breakdown structure, phases, and specific WBS data elements. You can also use filtering and grouping features to customize the format and level of information included in the tracking view.

[0331] During the project implementation phase, progress information such as the actual start time of the activity, the percentage of completion of the visual progress, the actual completion time, the calculated actual duration, and the actual number of resources consumed need to be regularly entered into the system. Sometimes, it is necessary to adjust the work breakdown structure, delete or add activities, and adjust the logical relationships between activities. Updating the progress is very important as the project progresses. The actual duration may differ from the original estimate, and the order of activities may change as soon as work begins. In addition, new activities may need to be added and unnecessary activities deleted. Regularly updating the progress and comparing it with the target planned schedule ensures efficient resource utilization, monitors project costs against the budget, and obtains the actual duration and costs in a timely manner so that contingency plans can be implemented when necessary.

[0332] During the implementation phase, while maintaining the target plan and updating the progress information, it is necessary to continuously track the project progress, compare the plan with the actual progress, analyze the progress information, find deviations and problems, and take appropriate control measures to solve existing problems and prevent potential problems. The BIM-based progress management system provides a variety of analysis methods at different levels to achieve a comprehensive analysis of project progress. During the implementation phase, it is necessary to review the progress, resource allocation and cost situation to ensure that the project development is consistent with the plan.

[0333] ① Progress analysis mainly includes milestone control point impact analysis, critical path analysis, and comparison analysis between planned and actual progress. By reviewing the milestone plan and critical path, combined with the actual completion time of the task, it is possible to check and predict whether the project progress is completed according to the planned time. Critical path analysis can be performed using the bar chart or network chart in the system;

[0334] See also Figure 17 As shown, the comparison between planned progress and time progress is generally completed by combining bar chart comparison, progress curve comparison, and model comparison. The system can display three views simultaneously to achieve the comparison between planned progress and actual progress.

[0335] See also Figure 18 As shown, you can set the color of the view to compare the planned progress with the actual progress. In addition, by comparing the project planned progress model, the actual progress model, and the on-site conditions, you can clearly see the growth process of the building and discover the progress and other problems during the construction process.

[0336] ② Resource situation analysis During the project progress, the analysis of resource situation is mainly based on the review of working hours differences to check whether there is over-allocation or under-allocation of resources. The progress management system based on BIM can provide resource analysis tables, resource histograms or resource curves in the system to analyze resource allocation. The resource view can be combined with the Gantt chart tracking view to display the allocation and usage status of resources in the selected time period, and timely discover resource allocation problems and resource allocation;

[0337] ③ Cost analysis For most projects, especially those with budget constraints, analysis of budget costs during the implementation phase is essential. If actual progress information indicates that the project may exceed the budget, adjustments to the project plan are required. The BIM-based progress management system can use cost analysis tables, histograms, and cost control reports to monitor expenditures.

[0338] After entering actual job information into the system, the system automatically uses planned values and actual costs to calculate earned value to evaluate current cost and schedule performance. These values can be tracked over time, and the project's past expenditure and schedule trends, as well as future cost forecasts and earned value analysis, can be viewed.

[0339] After entering actual progress information into the system, a comparative analysis between the actual progress and the project plan can reveal many deviations and point out potential problems in the project. To avoid problems caused by deviations, it is necessary to continuously adjust goals during the project process and take appropriate measures to solve problems that arise. Project completion time, total cost or resource allocation often deviate from the original planned track. Appropriate measures need to be taken to align project development with the plan. If a project undergoes major changes or seriously deviates from the project process, it is necessary to rearrange the project schedule and determine the target plan, adjust resource allocation and budget costs, and achieve progress balance.

[0340] Correcting the project schedule can be achieved by changing the duration of implementation work, such as by rushing the work, but this usually requires increasing resource inputs such as man-hours. It is necessary to use duration-resource or duration-cost optimization to select solutions that shorten the schedule, reduce resource input, and minimize cost increases. Another approach is to change the logical relationship or overlap between project implementation tasks, without changing the duration of the tasks, but only changing the start and end times. If these two approaches fail to achieve the goal of shortening the schedule and the schedule is seriously delayed, it is necessary to readjust the project schedule and update the target plan.

[0341] During the project, the main corrective measures for resource allocation are: adjusting resource availability, adjusting allocations, such as adding resources, replacing resources, delaying work or assignments, splitting work to balance workload, and adjusting project scope. The main corrective measures for cost are: re-examining budget expense settings, such as the cost per use of resources, fixed costs of activities, etc., shortening activity durations or adjusting activity dependencies to reduce costs, appropriately adding, deleting, or replacing resources to reduce costs, and reducing project scope to reduce costs.

[0342] Adjustments to schedule deviations and updates to target plans must take into account factors such as resources and costs, and adopt appropriate organizational, management, technical, and economic measures. Only in this way can a multi-party balance be achieved and the ultimate goal of schedule management be realized.

[0343] Example 5

[0344] See also Figure 19 As shown, the application of the BIM-based project progress management system described in this embodiment has the following platform functions:

[0345] (1) Realize substation construction simulation project management and support project creation, deletion, import and export;

[0346] (2) Support data import in GIM format for substation design results and IFC format for building results;

[0347] (3) Support the import of common 3D model formats, including STL, Dgn, Dwg, etc.

[0348] (4) Support operations such as query, modification and deletion of data to achieve management and adjustment of terrain data. Basic geographic information data includes image data, digital elevation model data and basic vector data;

[0349] (5) Image data includes satellite images, aerial images, scanned topographic maps, etc.;

[0350] (6) Digital elevation model data includes contour lines, discrete points, and raster data, where raster data is obtained by converting contour lines or discrete points;

[0351] (7) Basic vector data include administrative divisions, place names, settlements, transportation, water systems, vegetation distribution, agricultural and forestry land, etc.;

[0352] Supports the import of individual and overall models of GIM models, IFC models, and conventional 3D models (stl, dgn, dwg, rvt, 3ds, dae). Based on the D3Station platform, it enables rapid parsing and visualization of models, and supports viewing of model properties.

[0353] It supports the import of terrain data such as Dem high-level data, Dom image data, and oblique photography. Based on the D3Station platform, it realizes the analysis and visualization of geographic information data. It supports the display of basic geographic information data including image data, digital elevation model data, and basic vector data. Image data includes satellite images, aerial images, scanned topographic maps, etc. Digital elevation model data includes contour lines, discrete points, and raster data. Raster data is obtained by converting contour lines or discrete points. Basic vector data includes administrative divisions, place names, residential areas, transportation, water systems, vegetation distribution, agricultural and forestry land, etc. Power grid thematic data includes wind zones, ice-covered areas, pollution areas, earthquake zoning, dancing areas, lightning-damaged areas, bird-damaged areas, etc. Transmission line channel data includes important industrial planning areas, environmental protection and water conservation, mineral plant areas and other areas within the line channel, as well as cross-span data. Survey data includes professional data such as surveying, hydrology, meteorology, geology, and geophysical prospecting.

[0354] Model structure, parse GIM, IFC model structure, support model structure function panel, view the hierarchical structure of the model, and highlight the model according to the model structure. Not only the imported model will be included in the model tree structure, but also the model arranged through the site will be included in the tree structure.

[0355] Model management includes:

[0356] Character model: A rich variety of character models are built in, including designer, supervisor, chief engineer, etc., with attributes including character height, animation movement, and material.

[0357] Vehicle model: A rich variety of vehicle models are built in, including ordinary cars, maintenance vehicles, etc., and attributes include vehicle length, vehicle width, vehicle height, and vehicle material;

[0358] Vegetation model: A rich variety of vegetation models are built in, including common trees, arbor trees, shrub trees, etc. Attributes include tree height and tree material;

[0359] Earthwork component model: Built-in cross-sectional information of foundation pit, earthwork, etc., with properties including: cross-sectional dimensions of corresponding components, corresponding component materials, and components are placed by linear drawing;

[0360] Building component model: Built-in cross-sectional information attributes of the ground, wall, road, and barrier include: cross-sectional dimensions and materials of the corresponding components. Components are placed using linear drawing. Built-in models of doors and windows include: size information and material information of the corresponding models.

[0361] Safety protection model: A rich variety of built-in safety protection models, including safety warning lights, safety warning signs, etc., and attributes include the material of the safety protection model;

[0362] Green civilization model: A rich variety of green civilization models are built in, including LED screens, nine cards and one map, etc. Attributes include: the material of the green civilization model;

[0363] Temporary power equipment model: A rich variety of temporary power equipment models are built in, including lighthouses, switch boxes, electrical boxes, lights, cameras, etc. Attributes include: the material of the temporary power equipment model;

[0364] Mechanical equipment model: A wide variety of built-in construction machinery equipment models, including cranes, tower cranes, welding machines, bulldozers, rollers, etc., with attributes including height, length, width, and material of the equipment.

[0365] FBX Editor: This allows you to edit model actions. By adding keyframes, you can animate imported models in the form of displacement and rotation. You can animate all imported models and reflect them in the construction simulation process. You can animate different types of models to achieve the corresponding action. The FBX Editor is an important component of construction simulation.

[0366] Site layout: Provides a substation site layout area where various models can be placed in the 3D viewport. It supports calling and placing models in the model management module, including: human models, vehicle models, vegetation models, land component models, building component models, safety protection models, green civilization models, temporary power equipment models, and mechanical equipment models. Select the corresponding model, fill in the relevant parameters of the model, and quickly place the site layout model in the substation. After the model is placed, the model structure will also be quickly refreshed and linked.

[0367] Construction management personnel import the project plan list and fill in the actual progress time as the project is under construction, realizing unified management of project plans and actual progress, truly reflecting actual differences, and facilitating the control of construction progress of different people. The substation construction project tree structure is constructed, and the creation, deletion, import, and export of the overall construction project of the substation project are supported. The planned progress information and the actual progress information are filled in. The creation of the project and the creation of sub-subsets are supported, and the deletion of the project and the deletion of sub-subsets are supported. The import supports the import of regular EXCEL, ACCESS, and MPP format files, and the export supports the export of EXCEL, ACCESS, and MPP format files. The planned progress information can be filled in manually or automatically by importing files, and the actual progress information can be filled in manually or automatically by importing files.

[0368] Obtain tree structure data from the model structure, analyze the correspondence between the construction plan and the 3D model, establish relationships between the construction milestone plan, the first-level network plan, and the 2D network plan tasks, assign time attributes to the model, construct 4D display basic conditions, and bind the model to each project in the project list. Repeatable binding is supported, and multiple bindings are supported to achieve the binding relationship between the tree structure model and the project plan list, providing data support for 4D simulation of construction progress.

[0369] Through the construction project tree structure list, the timeline information of each project is linked to display, allowing you to quickly view the weekly, monthly, and quarterly project progress, intuitively view the timeline of planned progress and actual progress, and highlight the difference between the planned progress time and the actual progress time in red on the timeline;

[0370] Generate an overall growth animation of the substation through the list of planned and actual progress information, the binding relationship of the model, and the project timeline. The animation can be controlled to play, pause, accelerate, and decelerate. Based on the start and end time of the overall construction plan of the project, the overall construction progress of the project at different stages is dynamically displayed through the growth of the model. At the same time, it supports the visualization of the progress of a single construction plan, realizing dynamic progress display of different dimensions and granularity, and supporting the animation display of the simulation of planned and actual progress at the same time.

[0371] You can customize the installation sequence between equipment components, obtain the current equipment model, disassemble the equipment model through the FBX editor, add keyframes using FBX, and simulate the equipment installation process and sequence from beginning to end through the combination of displacement and rotation of each component, and generate equipment installation simulation animation;

[0372] The model path setting is realized, and the model displacement position, time, speed can be set. The substation construction simulation simulates the construction vehicle entry under different scenarios and different processes by calling different construction vehicle models (transport vehicles, lifting vehicles, etc.), setting vehicle operation trajectory, channel distance, output safety distance and other parameters, verifying vehicle transportation safety distance, and optimizing the construction vehicle entry sequence;

[0373] Use the FBX editor to combine construction machinery and lifting equipment. By adding keyframes, the overall model can be split and the lifting simulation process of the equipment can be restored by combining the displacement and rotation of each component.

[0374] Hard collision verification: During the equipment installation simulation, mechanical path simulation, equipment hoisting simulation, and installation process simulation, the system detects whether collisions between entities will occur in the substation model under the simulation state. If collisions occur, the verification fails and the colliding parts of the model are highlighted.

[0375] Soft collision verification: During equipment installation simulation, mechanical path simulation, equipment hoisting simulation, and installation process simulation, in the substation model under the simulation state, the distance between two objects is checked to see if it meets the electrical clearance requirements by setting the electrical clearance parameter distance. If the distance verification fails, a prompt will be given, and the safety value and actual value between the equipment or buildings will be marked;

[0376] By directly importing the project construction schedule file, the project schedule is automatically generated, including the schedule name, schedule content, person in charge, schedule start time, and schedule completion time. Based on this, the contents of each construction schedule can be refined to meet the actual work needs of construction management personnel and avoid the increase in workload caused by repeated preparation of schedules in the system.

[0377] The progress management module can edit the personnel deployment plan for each sub-project. By accessing the personnel entry and exit data from the smart construction site personnel gate, the module compares the actual number of personnel entering the site with the planned number of personnel entering the site, provides reminders for insufficient manpower arrangements, analyzes the possible impact on the construction progress of the sub-project, and assists management personnel in adjusting the construction personnel deployment to ensure on-time completion.

[0378] By associating the GIM hierarchical structure with the project schedule tasks, a 3D model of the project progress is simulated to display the project construction progress in real time. Delayed, normal, and unfinished items are marked on the 3D model using color, providing auxiliary decision support for on-site personnel in progress management. A realistic model of the construction site is created using drones and oblique photography technology. This can be used as a process record for project progress management and as a basis for construction management summaries and progress analysis, providing data support for progress management based on the actual scene.

[0379] Through BIM construction simulation technology, the three-dimensional design model is disassembled and reassembled, and construction simulation is performed on construction projects with complex construction processes and strict construction sequence requirements. Construction simulation animations are generated and provided to construction teams and technical personnel for technical briefings, avoiding the problem of unclear process flow caused by obscure on-site written plan briefings.

[0380] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0381] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. BIM-based project progress management system, characterized by: The system includes a data model, a progress management model, and a comparison model. The data model includes a three-dimensional information model of the power grid. The three-dimensional information model of the power grid is converted into a surface model by scanning the entity. The three-dimensional data model includes a solid model and a mesh model. The solid model is used to describe the geometry and topology information of the component, and the mesh model uses polygonal patches to form the geometric shape of the object. The data model is generated by combining a 3D data model expression method with occlusion queries from multiple viewpoints to identify hidden parts in assemblies. The progress management model establishes a real-time model and generates a corresponding calculation model based on the analysis of the data model structure. The comparison model overlays and compares the real-time model with the planned model to provide solutions for correcting and adjusting the construction progress. In the hidden parts identification algorithm, the visibility value of the part at a certain viewpoint is calculated as: ； in: ; in, represents the number of pixels that pass the depth test when performing an occlusion query on part i at viewpoint j when rendering the assembly; Indicates the number of pixels that pass the depth test when performing an occlusion query on part i at viewpoint j when rendering part i alone; Represents the number of pixels that pass the depth test when performing an occlusion query on the entire assembly at viewpoint j when rendering the assembly; represents the volume of part i; Represents the volume of the entire assembly; in, 、 and As weight, the initial setting is that the part visibility rate, part area contribution rate and part volume contribution rate have the same weight, that is, 、 as well as Both are 1 / 3; After the occlusion query of part i is completed at all the set viewpoints, the maximum visibility value of part i is obtained, that is, , defining the critical parameters , When the F value of a part in the assembly is greater than the critical parameter, the part is visible when observed from the outside of the assembly, otherwise the part is hidden.

2. The BIM-based project progress management system according to claim 1, characterized in that: The specific steps for converting the three-dimensional information model of the power grid into a surface model by scanning the entity are as follows: Let the point before scanning be A (X A 、Y A , Z A )、B(X B 、Y B , Z B )……K(X K 、Y K , Z K ), scan vector V(X V 、Y V , Z V ), after scanning these points along the scanning vector V, we can get An, Bn...Kn. The line-surface network formed by these points is the boundary model of 3DGIS: An=A+V,Bn=B+V……Kn=K+V For curve scanning, select the curve Curve=(M1(X1, Y1, Z1), M2(X2, Y2, Z2)…Mn(X n 、Y n , Z n )), the tangent vector of the curve is Hn=(X n -X n-1 , Y n -Y n-1 , Z n -Z n-1 ), then there is a point A (X A 、Y A , Z A ,0) The scanned point A' is calculated as follows, where N is the surface normal: ； The transformation from the entity model to the boundary model is achieved by analogy.

3. The BIM-based project progress management system according to claim 1, characterized in that: The method for performing occlusion query from multiple viewpoints to identify hidden parts in an assembly comprises the following steps: (1) Generate a query object ID for the object to be queried; (2) Start occlusion query; (3) Rendering objects that require occlusion queries; (4) End the occlusion query; (5) Extract the number of samples that pass the occlusion query; (6) Delete the query object ID and recycle resources.

4. The BIM-based project progress management system according to claim 1, characterized in that: The hidden piece identification algorithm evaluates the priority value of the vertex by the position of the vertex in the cache and the degree of the vertex. , and the priority value of the triangle ring is the sum of the vertex priority values of all unoutput triangles in the ring ; The vertex priority value C(v) can be calculated by the following formula: ； in, ； ； In the above formula, p represents the position of vertex v in the cache, s represents the size of the cache space, and a represents the number of unoutput triangles adjacent to vertex v. 、 、 is the coefficient.

5. The BIM-based project progress management system according to claim 4, characterized in that: The model metadata structure of the power grid three-dimensional information model includes basic data and extended data, wherein the basic data includes geometric data, physical data and functional data, and the extended data includes technical data, economic data and management data.

6. The BIM-based project progress management system according to claim 5, characterized in that: The real-time model includes the statistical distribution of process duration, which includes normal distribution, beta distribution, uniform distribution and triangular distribution. In normal distribution, the mean, median and mode of process duration coincide with each other, the probability of process duration being greater than or less than the mean is equal, and the probability of process delay or advance is equal. In beta distribution, the distribution range is in [0, 1]. The distribution formula of beta distribution is as follows: (1) Mean: ; (2) Variance: ; The value range of is 0:0.1:10000, the value range of mean is 0.3:0.05:0.5, and the simulation results are stored in the matrix respectively. The first column is Value, the second column is The third column is the variance value.

7. An application of a BIM-based project progress management system according to any one of claims 1 to 6, characterized in that: The progress management system is applied to substation construction simulation project management. Based on the D3Station platform, it realizes model analysis and visualization, and supports viewing model properties.

Images