A BIM-based construction engineering progress intelligent prediction system

Abstract

The application relates to the field of building information model application technology and engineering project management technology, and discloses a building engineering progress intelligent prediction system based on BIM, which firstly analyzes BIM and progress data, constructs a semantic constraint topological graph containing logical dependence and spatial proximity relationship, then discretizes the building space, constructs an anisotropic diffusion tensor based on component semantics, realizes physical tensorization of geometric constraints, on the basis of which, the system performs multi-physical field coupling calculation, solves the distribution of material flow resistance coefficient field and risk field by using a potential field method and a diffusion equation respectively, finally, the environmental resistance factor in the process operation space is calculated to correct the process prediction duration, and iterative control is performed based on a field-graph closed loop feedback mechanism. The application can quantize the spatial congestion effect and risk non-uniform conduction under a complex building environment, solves the problem of logical plan and physical constraint separation, and significantly improves the accuracy of construction progress prediction.

Description

Technical Field

[0001] This invention relates to the fields of building information modeling application technology and engineering project management technology, specifically a BIM-based intelligent prediction system for building project progress. Background Technology

[0002] Construction engineering is a complex and dynamic process with highly coupled time and space, and its schedule management directly affects the project's economic benefits and delivery risks. Currently, the preparation and control of construction schedules mainly rely on the Critical Path Method (CPM) and the Program Evaluation and Review Technique (PERT). These methods focus on analyzing the temporal logical dependencies between processes, determining the critical path of the project by calculating the earliest and latest start times of each process. However, traditional network planning techniques often implicitly assume an infinitely large workspace or frictionless resource supply, making it difficult to effectively characterize the physical constraints arising from spatial limitations at the construction site.

[0003] With the development of Building Information Modeling (BIM) technology, BIM-based 4D construction simulation has, to some extent, achieved the visualization of construction schedules. However, existing BIM 4D technology mainly focuses on the dynamic presentation of geometric models over time and static component collision detection, lacking the ability to quantitatively analyze the dynamic characteristics of the construction process. In actual construction sites, parallel processes that are not logically related often lead to resource competition, logistical congestion, or environmental interference due to physical proximity, resulting in actual work efficiency being significantly lower than theoretical values. For example, delays in local areas may lead to material accumulation, thereby altering the surrounding traffic environment; this cascading effect is difficult to reflect in existing logical network models.

[0004] Furthermore, while some existing studies have attempted to incorporate spatial distance calculations to assess operational conflicts, most employ isotropic attenuation models based on Euclidean distance. Building interior environments exhibit high levels of non-uniformity and anisotropy; solid walls provide absolute obstruction, while corridors offer significant guidance. Simple distance determinations cannot accurately reflect the actual transmission paths of risks or disturbances within complex building geometries, leading to insufficient prediction accuracy of spatial congestion effects and risk permeation processes. Consequently, the final schedule predictions often deviate significantly from the actual project progress. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a BIM-based intelligent prediction system for building project progress. This system solves the problem in existing building project progress management methods where logical planning and physical space constraints are separated, making it impossible to quantitatively calculate the spatial congestion effect and the non-uniform transmission of delay risks in complex building geometries.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a BIM-based intelligent prediction system for construction project progress, comprising a data acquisition and parsing module, a semantic topology construction module, an anisotropic field modeling module, a multiphysics coupling calculation module, and a schedule correction and iterative control module.

[0007] The data acquisition and parsing module is used to parse building information model files to extract the geometric spatial attributes and type attributes of components, and to parse schedule files to obtain the list of operations and the logical dependencies between operations.

[0008] The semantic topology building module is used to construct a semantic constraint topology graph, which unifies temporal logic and spatial location in the same data structure.

[0009] The semantic constraint topology graph includes a set of nodes representing processes, a set of directed logical edges representing logical dependencies, and a set of undirected spatial edges representing physical spatial proximity relationships.

[0010] In constructing the set of undirected spatial edges, the system calculates the Euclidean distance between the operation spaces of any two process nodes that do not have a direct logical dependency relationship. When the Euclidean distance is less than the preset safe operation distance threshold, an undirected spatial edge is established, thereby explicitly representing the potential physical interference path.

[0011] The anisotropic field modeling module is used to establish a physical field environment that describes the characteristics of the risk transmission medium. This module divides the three-dimensional space of the construction site into discrete voxel meshes and establishes a mapping relationship between the semantic attributes of BIM components and the physical medium attributes, constructing an anisotropic diffusion tensor D(p) for each voxel.

[0012] Specifically, for voxels located inside solid structural components, their diffusion tensors are set to blocking values; for voxels located in open regions, they are set to isotropic scalar matrices; for voxels located in channel regions with directional characteristics, by extracting the dominant channel direction vector and constructing a rotation matrix, a second-order tensor with a diffusion coefficient significantly greater than the lateral diffusion coefficient along the dominant direction is generated, which mathematically represents the geometric constraints of the building space structure on field evolution.

[0013] The multiphysics coupling calculation module is used to perform numerical calculations of time-varying physical fields. At the current prediction time step, this module identifies the distribution of obstacles and the state of process delays, and calculates the distribution of logistics resistance coefficient field and risk field intensity, respectively.

[0014] When calculating the logistics resistance coefficient field, the system constructs an gravitational potential field with the target location as the minimum and a repulsive potential field with the obstacle as the maximum. By superimposing the total potential field and performing normalization mapping, the logistics resistance coefficient reflecting the ease of passage through space is obtained. .

[0015] When calculating the risk field intensity distribution, the system marks the operation space corresponding to the process node in a delayed state as a risk release source, and constructs a system containing source terms. Nonhomogeneous anisotropic diffusion partial differential equation: ; in, For the intensity of the risk field, For Hamiltonian operators, As the dissipation coefficient, the system uses the finite difference method to discretize and solve the equation, simulating the non-uniform infiltration process of schedule delay risk along the principal axis of the tensor in the building space.

[0016] The schedule correction and iterative control module is used to achieve closed-loop feedback of site-map data. This module performs integral calculations on the field quantity data within the process operation space to obtain the environmental resistance factor. The predicted duration of the process is corrected using an exponential decay model.

[0017] in, The corrected predicted duration, For the remaining amount of work, This is the baseline efficiency.

[0018] Subsequently, the system updates the project schedule based on the corrected schedule and determines whether the total schedule deviation meets the convergence condition. If it does not converge, the system updates the time step according to the new schedule status and drives the physical field to perform the next round of evolution calculation until the system reaches dynamic equilibrium.

[0019] A second aspect of this invention provides a BIM-based intelligent prediction method for construction project progress, the method comprising the following steps: The system parses building information model files to extract component attributes, parses schedule files to extract process logic, and constructs a semantic constraint topology graph containing logical edges and spatial edges accordingly. The construction site space is discretized into a voxel mesh, and an anisotropic diffusion tensor is constructed for each voxel based on the semantic properties of the components, thereby realizing the transformation of architectural geometric semantics into physical field medium properties. At the current prediction time step, perform multiphysics coupling calculations: calculate the logistics resistance coefficient based on the static and dynamic obstacle distribution using the artificial potential field method; construct risk source terms based on the process delay status, and solve the risk field intensity distribution using the diffusion equation containing the anisotropic diffusion tensor. The environmental resistance factor is obtained by integrating the distribution of the material resistance coefficient field and the risk field intensity within the process operation space, and the predicted duration of the process is corrected based on the environmental resistance factor. Based on the corrected predicted duration, the critical path algorithm is used to update the project schedule. The total duration deviation between the current iteration step and the previous iteration step is compared to determine whether convergence has occurred. If convergence has not occurred, the above physical field calculation steps are iteratively executed according to the updated schedule to output the final schedule prediction result.

[0020] This invention provides a BIM-based intelligent prediction system for building construction progress. It offers the following advantages: 1. This invention constructs a semantic constraint topology graph containing spatial edges. By calculating the Euclidean distance between non-logically related process nodes and establishing undirected spatial edges, it explicitly represents potential physical proximity relationships. This breaks through the limitation of the traditional critical path method, which only relies on temporal logical relationships. It enables the schedule prediction model to automatically identify and incorporate implicit constraints that are not logically directly dependent but have resource contention or operational interference in physical space, thereby improving the comprehensiveness of identifying cross-operation conflicts on the construction site.

[0021] 2. This invention utilizes anisotropic field modeling technology to transform the geometric semantic attributes of components in the BIM model into the diffusion tensor distribution in physical field theory. By setting blocking tensors for solid walls and tensors with dominant directions for passage areas, the system ensures that the numerical evolution of the risk field and the logistics field strictly follows the physical boundary constraints of the building structure. This avoids the calculation deviations caused by traditional isotropic models in complex geometric environments, accurately reflects the blocking effect of the solid structure and the guiding effect of the passage space, and significantly improves the spatial resolution and simulation accuracy of physical field calculations in the building's interior environment.

[0022] 3. This invention maps the microscopic three-dimensional spatial field distribution into macroscopic schedule correction parameters by integrating the logistics resistance coefficient field and risk field within the process operation space. Based on this, the schedule is dynamically adjusted. The field-map coupled calculation mode ensures that the schedule prediction results can respond in real time to changes in spatial congestion and risk penetration at the construction site. This realizes the transformation from static logical planning to dynamic physical simulation, providing a quantitative prediction basis for project management that conforms to the physical entity. Attached Figure Description

[0023] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention.

[0024] The modules include: 101. Data acquisition and parsing module; 102. Semantic topology construction module; 103. Anisotropic field modeling module; 104. Multiphysics coupling calculation module; and 105. Schedule correction and iterative control module. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Example: Please see the appendix Figure 1 This invention provides a BIM-based intelligent prediction system for building construction progress, comprising: a data acquisition and parsing module 101, a semantic topology construction module 102, an anisotropic field modeling module 103, a multiphysics field coupling calculation module 104, and a schedule correction and iterative control module 105.

[0027] The data acquisition and analysis module 101 is an interface for connecting the building information model database and project management software, used to read the geometric information, physical attribute information and initial construction schedule data of building components.

[0028] The semantic topology construction module 102 is connected to the data acquisition and parsing module 101 and is configured to map construction operations as graph nodes according to the parsed schedule, establish logical edges according to the logical dependencies between operations, and establish spatial constraint edges according to the spatial adjacency of the component geometric bounding boxes, thereby generating a semantic constraint topology graph.

[0029] The anisotropic field modeling module 103 is connected to the semantic topology construction module 102 and is configured to discretize the three-dimensional space of the construction site into a voxel mesh, and define anisotropic diffusion tensors for each voxel according to the physical properties of the building components corresponding to the voxels, thereby establishing a physical conduction environment model.

[0030] The multiphysics coupling calculation module 104 is connected to the anisotropic field modeling module 103 and is configured to perform artificial potential field calculations to generate the logistics resistance distribution and to perform anisotropic diffusion equation solving to generate the risk field distribution.

[0031] The schedule correction and iteration control module 105 is connected to the multiphysics coupling calculation module 104 and is configured to quantify the working environment resistance based on the physical field calculation results, correct the schedule attributes of nodes in the semantic topology graph, and recalculate the critical path to determine the system convergence status.

[0032] Please see the appendix Figure 2 This invention provides a BIM-based intelligent prediction method for building construction progress, which includes the following steps: S100, Data Initialization and Semantic Topology Generation. The data acquisition and parsing module 101 parses the IFC format BIM data, extracts the geometric coordinates and type attributes of the components, and simultaneously parses the project schedule file to obtain the process list and logical dependencies; The semantic topology construction module 102 establishes a node set based on the process list, establishes a set of directed logical edges based on logical dependencies, calculates the Euclidean distance between component bounding boxes, and establishes a set of undirected spatial edges between nodes with a distance less than a preset threshold, thereby forming an initial semantic topology graph. S200, Spatial Discretization and Anisotropic Medium Modeling, Anisotropic Field Modeling Module 103 divides the three-dimensional spatial domain of the construction site into a continuous voxel mesh. For each system, the medium type is determined based on the BIM component attributes corresponding to its spatial location; for regions with directional conduction characteristics, a second-order anisotropic diffusion tensor is constructed, which is a symmetric positive definite matrix used to describe the difference in conduction rate of the scalar field along different directions at that location. S300, Dynamic working surface potential field calculation, Multiphysics field coupling calculation module 104 at the current prediction time step, identifies the set of voxels occupied by completed components and temporary facilities as obstacle sources, and identifies the target position of the process to be executed as a gravity source. The total potential energy at any location in space is calculated using the artificial potential field method. Based on the total potential energy distribution, the logistics resistance coefficient is calculated, which characterizes the ease or difficulty of resource flow in space. S400, risk field anisotropic diffusion evolution, multi-physics field coupling calculation module 104 detects process nodes in the semantic topology graph that are in a delayed state or high load state, and sets their corresponding spatial system region as risk source item; By solving the anisotropic diffusion equation containing the source term, the intensity distribution of the risk field is calculated, and the process of schedule risk as a scalar field permeating from the source point to the surrounding area following the physical diffusion law is simulated. S500, Nonlinear correction of schedule parameters, Schedule correction and iterative control module 105 calculates the comprehensive resistance factor in the operating space for each incomplete process node. The comprehensive resistance factor is obtained by weighted integral of logistics resistance coefficient and risk field intensity. Based on the comprehensive resistance factor, the actual operating efficiency of the process is corrected using the exponential decay model, and the predicted duration of the process is updated accordingly. S600, System Iteration and Convergence Judgment, Schedule Correction and Iteration Control Module 105 will backfill the updated predicted duration into the semantic topology graph and recalculate the overall project schedule using the critical path algorithm. Compare the deviation of the total project duration between the current iteration step and the previous iteration step; If the absolute value of the deviation is less than the preset convergence threshold, the current progress prediction result will be output. If the absolute value of the deviation is greater than or equal to the preset convergence threshold, the time step and obstacle distribution status are updated according to the new schedule, and the process returns to step S300 until the system converges.

[0033] The data acquisition and parsing module 101 performs data initialization processing, converting unstructured multi-source heterogeneous building engineering data into a standardized data format that can be used for subsequent topology construction and physics field calculations. This includes the following processing steps: BIM Model Geometric and Semantic Information Extraction. The data acquisition and parsing module 101 reads the Building Information Model (BIM) file through the standard interface of the Industrial Basic Class. For each entity component object defined in the model file, the module parses its attribute set inherited from the IfcProduct class, extracting the globally unique identifier, component type, and hierarchy information. Simultaneously, it parses the IfcShapeRepresentation field of the component to obtain geometric topology data. For the geometric data, the module extracts the component's three-dimensional vertex set. The component's axis-aligned bounding box is calculated based on this vertex set. The component's spatial properties are parameterized as the bounding box center coordinates. and size vector : ; ; in, , The vertex set is in Minimum and maximum values ​​on the axis, shaft and The same applies to axes. This bounding box defines the basic spatial volume occupied by the component in subsequent physics calculations. For the specific implementation of IFC file parsing and geometric operations, existing geometric kernel libraries can be used, which is a conventional technique in this field.

[0034] The schedule plan is structured and parsed. The data acquisition and parsing module 101 reads the schedule plan file generated by the project management software. The module traverses the task node tree in the file and extracts the task ID, task name, planned duration, planned start time, and planned end time for each process task.

[0035] Specifically, the module parses the dependency description fields between tasks, identifies the index list of predecessor and successor tasks, as well as the dependency type and lag time. The above parsing process transforms linear text or tabular data into a linked list structure containing logical connections.

[0036] The module establishes a mapping table between BIM component GUIDs and schedule task IDs for the association mapping between components and processes. This mapping is achieved by parsing a preset attribute set in the BIM model. If the BIM model already contains a task ID field, index matching is performed directly. If no pre-defined association is used, a rule-based automatic matching algorithm is invoked to associate components based on the string similarity between the component type and the task name. In the case of one-to-many relationships, the module performs a spatial Boolean union operation on the bounding boxes of all associated components to generate the overall operation space set for that process. ; In the case of many-to-one relationships, the module cuts the component geometry according to the construction segment division rules to generate the corresponding subspace set.

[0037] After mapping, the system generates a standardized list of objects containing geometric spatial attributes and temporal logical attributes, which serves as input data for subsequent semantic topology construction.

[0038] The semantic topology construction module 102 constructs a heterogeneous graph structure with dual constraints of logical and spatial dimensions based on a standardized object list, specifically performing the following processing steps: Graph node attribute definition and instantiation, semantic topology construction module 102 initializes an empty directed graph data structure. The module iterates through the generated list of objects and instantiates a graph node for each construction process. And add the node to the node set. For each node The system allocates and stores attribute vectors. The vector contains: a unique identifier Process type coding Quantity of work Baseline construction period based on quota calculation and spatial attribute set .

[0039] Among them, the set of spatial attributes It includes not only the static geometric bounding box of the components, but also the work buffer space based on the work process type. For example, for a work process of type "exterior wall scaffolding erection," its... The spatial volume of a component geometry after expanding outwards by a preset width along the normal direction.

[0040] Topological mapping of logical dependencies. The module iterates through the list of dependencies in the schedule and constructs a set of logical edges. When the process is detected With process When predecessor and successor relationships exist, the module is at the node point to Establish a directed edge in the direction of The directed edge carries a weight attribute. , representing the time lag coefficient between two processes. For different types of dependencies, the module marks them in the edge attributes so that subsequent time parameter calculation algorithms can call them. This step maps the traditional critical path network to a directed acyclic graph substructure in graph theory.

[0041] Geometric computation and edge generation of spatial neighborhood relationships: To capture potential physical interference from non-logical connections, the module constructs a set of spatial edges. The module employs a breadth-first search algorithm based on a bounding box hierarchy to search the node set. The set of spatial attributes corresponding to all processes Perform pairwise intersection tests or distance tests, and define a spatial proximity determination function. .

[0042] ; in, They are respectively spatial volumes and The coordinates of any point in space, For the Euclidean norm, The preset safe working distance threshold, when And nodes and When there is no direct logical edge connection between the two nodes, the module establishes an undirected edge between them. .

[0043] The undirected edge indicates that although the two processes are not directly related in terms of schedule planning logic, they may have the potential to influence each other in physical space, thus forming a potential channel for resource competition or risk transmission.

[0044] Heterogeneous graph structure fusion and storage: the module will set logical edges. With spatial edge set Merge into edge set Complete the semantic constraint topology graph The construction of this topology graph is a heterogeneous graph, in which logical edges are used to transmit time parameters and critical path constraints, and spatial edges are used to define risk diffusion paths and congestion correlations in subsequent physics field calculations. The generated topology graph data is stored in the system memory or graph database in the form of an adjacency matrix or adjacency list for subsequent modules to call. Through the above steps, the system unifies the temporal logic and spatial location of building construction in the same graph data structure, providing a basic data model for subsequent field-graph coupled calculations.

[0045] The anisotropic field modeling module 103 transforms discrete semantic objects in the building information model into a continuous physical field medium model, establishing a spatial environment foundation for describing logistics flow and risk transmission. Specifically, it performs the following processing steps: The three-dimensional space is discretized using voxelization. The anisotropic field modeling module 103 determines the spatial bounding box of the construction site. This bounding box covers all BIM components and the preset site boundary. The module sets the spatial discretization resolution parameters. The spatial bounding box is divided into a regular set of three-dimensional voxel meshes. .

[0046] For each voxel Calculate the coordinates of its geometric center. This voxelization process transforms continuous Euclidean space into discrete computational units, with each voxel serving as the basic storage unit for subsequent scalar and vector fields.

[0047] Based on semantic medium attribute mapping, the module traverses the voxel set. For any voxel The center coordinates are determined using the point containment test algorithm. Within the geometric enclosure of which type of component in the BIM model is it located? Based on the component's IFC semantic type, the voxels are divided into three media types: Blocking medium collection :like Located inside walls, columns, or other permanent structural components, it is identified as a blocking medium, representing a physically impassable area.

[0048] Confined conductive medium assembly :like Located within what is defined as a corridor, passageway, stairwell, or narrow area, it is considered a confined transmission medium, indicating that the propagation of logistics or risks has a significant directionality.

[0049] Collection of isotropic media :like Located in a hall, square, or open area without specific components, it is determined to be an isotropic medium, which means that the propagation is uniform in all directions.

[0050] Anisotropic diffusion tensor construction: For the different media types mentioned above, the module is used for each system location. Construct a second-order symmetric positive definite diffusion tensor This tensor is used to control the direction and rate of field evolution in subsequent differential equations.

[0051] For the blocker medium set In the voxels, the tensor is set to a zero matrix or a minimum matrix. , to stop the spread of the field.

[0052] For an isotropic medium collection In the voxels, the tensor is set as a scalar matrix. ,in As the reference diffusion coefficient, for Identity matrix.

[0053] For a set of confined conductive media In the voxels, the module first extracts the dominant direction vector of the region. (This vector can be obtained by performing principal component analysis (PCA) on the geometry of the corridor area or by extracting the centerline).

[0054] Subsequently, the eigenvalue diagonal matrix in the local coordinate system is constructed. : ; in, The longitudinal diffusion coefficient is along the dominant direction. Let be the lateral diffusion coefficient perpendicular to the dominant direction, and satisfy . (For example This is to simulate the physical characteristics of risks or logistics that spread rapidly along the channel and are difficult to penetrate the sidewalls.

[0055] Next, the global coordinate system Z-axis is rotated to the dominant direction. rotation matrix Finally, the anisotropic diffusion tensor in the global coordinate system is obtained through similarity transformation: ; Through this step, the module mathematizes the semantic geometric features of the architectural space into tensor field distributions in physical field theory, ensuring that subsequent calculations can accurately reflect the physical constraints of the architectural space.

[0056] The potential field analysis unit in the multiphysics coupling calculation module 104 performs dynamic working face potential field calculation, aiming to quantify the resistance to material flow caused by site layout and congestion at a specific point in time. Specifically, it performs the following processing steps: Time-varying obstacle and target source identification. The module receives the current predicted time step from the schedule correction and iterative control module 105. Traverse the semantically constrained topology graph Based on the planned start time, planned end time, and current status of each node, the voxel set of the construction site is divided into three types of regions: Target gravitational region : The central region of the bounding box associated with the process node that is in the pending or executing state at the current time step.

[0057] Static obstacle area : Corresponds to the component associated with the process node marked as completed in the current time step, and the preset construction boundary line.

[0058] Dynamic semi-permeable barrier area This corresponds to the workspace and temporary storage area occupied by other parallel processes in the current time step. These areas are not absolutely impassable, but they generate high repulsive forces, representing the exclusionary effect caused by resource contention.

[0059] Construction of the gravitational field potential energy function. For each process task to be analyzed. The module is based on the geometric centroid of its operating space. Construct a gravitational field for the global minimum point. Gravitational potential energy function. Defined as about spatial location A quadratic function to drive the virtual logistics to converge toward the work surface: ; in, This is the gravitational gain coefficient. For the current voxel center To the target point The Euclidean distance. This function guarantees that the farther away from the target, the higher the potential energy and the greater the virtual traction force generated.

[0060] Multi-source repulsive field superposition calculation: The module calculates the repulsive potential energy generated by static and dynamic obstacles in space, for any spatial location. repulsive field It is composed of the superposition of potential energy generated by all obstacle sources. To avoid the local minima problem of the traditional artificial potential field method, an improved repulsive force function is adopted: ; single obstacle The generated repulsive component Defined as:

[0061] in, This is the repulsive force gain coefficient; For position To the obstacle The shortest distance to the nearest point on the surface; The threshold for the repulsive force's influence range is set only when the resource enters the vicinity of the obstacle. It is only within the range that it experiences a repulsive force; For the introduction of adjustment terms ( (A positive number) is used to ensure that the repulsive force near the target point is zero, thus guaranteeing the target's reachability.

[0062] For dynamic semi-permeable barrier areas, set a smaller The value allows resources to pass through at a higher energy cost, simulating the congestion and passage behavior on site.

[0063] The module for synthesizing the total potential field and calculating the gradient linearly superimposes the gravitational and repulsive fields to generate the total potential field distribution. .

[0064] Furthermore, the negative gradient vector of the total potential field is calculated. This vector indicates the ideal direction and magnitude of the resultant force of resource flow at that location.

[0065] The logistics resistance coefficient mapping module establishes a potential energy resistance mapping model to transform the potential energy meaning of the physical field into the damping meaning required for schedule calculation, defining spatial location. At any moment Logistics resistance coefficient : ; in, The maximum damping ratio constant (e.g., This indicates that efficiency is reduced to 1 / 3 under the most congested conditions. The hyperbolic tangent function is used to normalize the potential energy value and map it to an interval. ; For reference potential energy constant.

[0066] The output of this step The scalar field directly reflects the ease of passage at various points on site. The higher the potential energy of an area, the greater the resistance coefficient, which means that it will take more time to transport materials or carry out operations through that area.

[0067] The risk diffusion unit in the multiphysics coupling calculation module 104 performs non-logical interference modeling and calculates the dynamic permeation process of schedule delay risk in physical space through numerical simulation methods, specifically executing the following processing steps: Logical delays are mapped to source terms in the physical space; the module traverses the semantic constraint topology graph to identify the current moment. Work processes in an abnormal state include those experiencing delays and those facing rush orders. For each abnormal work process... The module sets its corresponding operation space set Marked as a risk release source, define a spatiotemporal risk source term function. This function characterizes the rate at which risk is injected into the physical field: ; in, For process At any moment The cumulative delay time; This is the resource intensity coefficient for this process, used to distinguish the difference in environmental impact between labor-intensive operations and general operations; This is the source strength normalization constant. This step transforms the time bias in the project management dimension into a scalar source strength in the physical field dimension.

[0068] The anisotropic diffusion control equations are constructed, and the module uses partial differential equations to describe the risk scalar field. The spatiotemporal evolution law of this field quantity The physical meaning of is defined as the spatial disorder or randomness at a specific location. To accurately reflect the constraint effect of building structures on risk propagation, a non-homogeneous anisotropic diffusion equation containing a source term is established: ; In the formula, For Hamiltonian operators; Represents divergence operations; This refers to the second-order anisotropic diffusion tensor related to the spatial location constructed above; Input the source item for the step; The natural dissipation coefficient represents the ability of a construction site to restore order through management measures.

[0069] The first term on the right side of the equation It describes the non-uniform propagation of risk along the principal axis of the tensor; Second item It describes the continuous injection of risk due to delays; Third item It describes the exponential decay of risk over time.

[0070] For three-dimensional finite difference numerical solutions, given the complexity of the architectural space boundary conditions, the module employs an explicit finite difference method to discretize and solve the aforementioned partial differential equations, discretizing the time domain into steps with a step size of... The sequence, spatially discretized into steps with a step size of A voxel grid. For time... Arbitrary voxel nodes The next moment field value The calculation is as follows: ; in, For the discrete difference operator of the diffusion term, due to the diffusion tensor Discrete operators containing off-diagonal elements include not only second-order central differences along the principal axes but also mixed difference terms with cross derivatives. Specifically, for the cross term... The central difference scheme is used for approximation.

[0071] During the solution process, the module is for blocking media. The voxel boundary is subject to a zero-flux boundary condition, i.e. To ensure that risks do not penetrate the solid wall and propagate, the system obtains the steady-state or transient risk field distribution of the entire construction site at the current time step through multi-step iterative calculations. This distribution visually shows which non-logically adjacent areas were affected by the delayed processes.

[0072] The schedule correction and iterative control module 105 serves as the feedback interface connecting physical field calculation and logical topology update. It performs the mapping from continuous field data to discrete schedule parameters and controls the iterative convergence process of the prediction system. Specifically, it executes the following processing steps: The calculation of the field quantity integral and environmental resistance factor in the workspace, and the traversal of the semantic topology graph in the 105-module module for schedule correction and iterative control. All process nodes in the process are either not started or in progress. For any process node... The module retrieves its corresponding operation space set. It identifies a subset of voxels falling within this spatial range. The module reads the logistics resistance coefficient field output by the multiphysics coupling calculation module 104. and risk field Calculate the comprehensive environmental resistance factor faced by this process. This factor is obtained by weighted volume integrals of the field quantities within the operating space volume: ; in, For operating space The total number of voxels included, used to normalize the effect of space size; voxels The logistics resistance coefficient at the location; voxels The intensity of the risk field at that location; This is the weighting coefficient for physical congestion resistance; These are the risk interference weighting coefficients. These two weighting coefficients are preset based on the type of engineering project and management preferences. This step compresses the complex, non-uniform physical field in three-dimensional space into a scalar index characterizing the severity of the working environment for a specific process.

[0073] Based on exponential decay-based dynamic efficiency correction, the module establishes a nonlinear mapping model between environmental resistance and construction efficiency. Considering that the impact of congestion and interference at the construction site on efficiency typically exhibits a nonlinear, marginally increasing characteristic, an exponential decay model is used to calculate the corrected actual work efficiency. : ; in, For process The baseline operational efficiency under ideal conditions is obtained by consulting the construction quota database or historical project data. For the natural constant An exponential function with base 0.5. The module updates the process based on the corrected work efficiency. Remaining forecast duration .

[0074] ; in For process The remaining amount of work.

[0075] The module will calculate the topology graph parameter backfilling and critical path reconstruction. Numerical write-back to semantic topology graph Corresponding node The module replaces the original planned duration with the updated attribute fields. Then, it calls the critical path algorithm engine to perform forward and backward recursive calculations on the entire graph based on the updated duration data.

[0076] Calculate the earliest start time of each node using a forward recursive approach. and earliest completion time : ; ; in For nodes The set of immediate predecessor nodes, For logical edge weights.

[0077] Calculate the latest start time of each node using reverse recursion. and latest completion time This determines the total float time. The sequence of nodes with zero float time constitutes the new critical path, and the total project duration... Update to the earliest completion time of the endpoint node.

[0078] The system iterative closed-loop and convergence determination are based on the fact that changes in the project duration lead to shifts in the process time window, which in turn alters the spatiotemporal distribution of the defined obstacles and target sources. Therefore, the system forms a strongly coupled closed loop. The module calculates the current iteration step. Compared to the previous iteration step Relative deviation of the total project duration : ; Set convergence threshold .like This indicates that the disturbance of the physical field significantly altered the schedule, and this alteration, in turn, drastically affected the distribution of the physical field, preventing the system from reaching a steady state.

[0079] At this point, the module feeds back the updated time parameters to steps S300 and S400, advances the time step or resets the obstacle state, and starts the next round of physics calculation.

[0080] like This indicates that the system has reached dynamic equilibrium, the iteration has terminated, and the module outputs the final converged schedule, critical path diagram, and risk heatmap as the prediction results. The specific programming implementation of critical path calculation and iterative control is a standard application of data structures and algorithms, and will not be elaborated upon here.

[0081] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A BIM-based intelligent prediction system for building construction progress, characterized in that, include: The data acquisition and parsing module is used to parse building information model files to extract the geometric spatial attributes and type attributes of components, and to parse schedule files to obtain the list of processes and the logical dependencies between processes. The semantic topology construction module is used to construct a semantic constraint topology graph, which includes a set of nodes representing processes, a set of directed logical edges representing logical dependencies, and a set of undirected spatial edges representing physical spatial proximity relationships. The anisotropic field modeling module is used to divide the three-dimensional space of the construction site into discrete voxel meshes, and construct an anisotropic diffusion tensor representing the medium conduction properties for each voxel based on the component semantic attributes corresponding to the voxel position. The multiphysics coupling calculation module is used to calculate the logistics resistance coefficient field based on the obstacle distribution and the risk field intensity distribution based on the process delay status at the current prediction time step. The schedule correction and iteration control module is used to calculate the environmental resistance factor in the process operation space based on the logistics resistance coefficient field and the risk field intensity distribution, correct the predicted duration of the process based on the environmental resistance factor, and determine whether the system has converged based on the corrected predicted duration. If it has not converged, the time step is updated and the next round of calculation is triggered.

2. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When constructing the undirected space edge set, the semantic topology construction module performs the following operations: Traverse the set of nodes to obtain the component geometric bounding box and operation buffer space corresponding to each process node; Calculate the Euclidean distance between the operation spaces of any two process nodes that do not have a direct logical dependency relationship; When the Euclidean distance is less than the preset safe working distance threshold, an undirected spatial edge is established between the two process nodes to characterize potential physical interference paths.

3. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When constructing the anisotropic field modeling module, the following operations are performed: For a system located inside a solid structural component, its anisotropic diffusion tensor is set to a blocking value to stop the propagation of the field; For voxels located within an open region, their anisotropic diffusion tensor is set as a scalar matrix to characterize isotropic uniform propagation. For a voxel located within a channel region with directional diffusion characteristics, the dominant direction vector of the channel region is extracted, a diagonal matrix containing longitudinal and transverse diffusion coefficients is constructed, and the diagonal matrix is ​​transformed into an anisotropic diffusion tensor in the global coordinate system by a rotation transformation based on the dominant direction vector.

4. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When calculating the logistics resistance coefficient field, the multiphysics coupling calculation module performs the following operations: Identify the voxel regions where completed components are located in the current prediction time step as static obstacle sources, and identify the voxel regions occupied by parallel construction processes as dynamic obstacle sources. Construct an gravitational potential field with the target position of the process to be executed as the minimum value, and a repulsive potential field with the static obstacle source and the dynamic obstacle source as the maximum values; The gravitational potential field and the repulsive potential field are superimposed to obtain the total potential field, and the potential energy value of the total potential field is mapped to the normalized logistics resistance coefficient through the hyperbolic tangent function.

5. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When calculating the risk field intensity distribution, the multiphysics coupling calculation module performs the following operations: Detect process nodes in the delayed state in the semantic constraint topology graph; The operating space area corresponding to the process node that is in a delayed state is marked as a risk release source; Based on the cumulative delay time of the process and the resource intensity coefficient, the source term intensity of the risk release source in the physical space is calculated, and the schedule deviation in the time dimension is transformed into a scalar source term input in the spatial dimension.

6. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When calculating the risk field intensity distribution, the multiphysics coupling calculation module also performs the following operations: Establish a nonhomogeneous diffusion partial differential equation that includes the source phase intensity, the natural dissipation coefficient, and the anisotropic diffusion tensor; The non-homogeneous diffusion partial differential equation is solved by discretization using the finite difference method. The spatiotemporal evolution of the risk scalar field from the risk release source to the surrounding area is calculated, and the risk field intensity distribution at the current prediction time step is obtained.

7. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When calculating the environmental resistance factor, the schedule correction and iterative control module performs the following operations: Retrieve the set of operation space systems corresponding to the process node to be corrected; Read the material resistance coefficient and risk field strength values ​​at each voxel position in the voxel set; The weighted integral and normalized values ​​of the logistics resistance coefficient and the risk field intensity are used to obtain the environmental resistance factor, which characterizes the severity of the working environment of the process.

8. The intelligent prediction system for building construction progress based on BIM according to claim 7, characterized in that, When correcting the predicted duration of a process, the schedule correction and iteration control module uses an exponential decay model to calculate the efficiency decay ratio based on the environmental resistance factor, and then uses the decayed work efficiency to update the predicted duration required for the remaining work volume of the process.

9. The intelligent prediction system for building construction progress based on BIM according to claim 1, characterized in that, When determining whether the system has converged, the schedule correction and iteration control module performs the following operations: The corrected predicted duration is backfilled into the semantic constraint topology graph, and the overall project schedule and total duration are recalculated using the critical path algorithm. Calculate the relative deviation between the total project duration in the current iteration step and the total project duration in the previous iteration step; If the relative deviation is less than the preset convergence threshold, the system is determined to have converged and the prediction result is output. If the relative deviation is greater than or equal to the preset convergence threshold, the time step and obstacle state are updated according to the new schedule, and the multiphysics coupling calculation is returned to be performed.

10. A BIM-based intelligent prediction method for building construction progress. A BIM-based intelligent prediction system for building construction progress according to any one of claims 1-9, characterized in that, Includes the following steps: The system parses building information model files to extract component attributes, parses schedule files to extract process logic, and constructs a semantic constraint topology graph containing logical edges and spatial edges accordingly. The construction site space is discretized into a voxel mesh, and an anisotropic diffusion tensor is constructed for each voxel based on the semantic properties of the components. At the current prediction time step, the logistics resistance coefficient field is calculated based on the obstacle distribution, and the risk source term is constructed based on the process delay status to solve the risk field intensity distribution. The environmental resistance factor is obtained by integrating the distribution of the material resistance coefficient field and the risk field intensity within the process operation space, and the predicted duration of the process is corrected based on the environmental resistance factor. The project schedule is updated based on the revised predicted duration. The total duration deviation is compared to determine whether convergence has occurred. If convergence has not occurred, the physics calculation steps are iteratively executed according to the updated schedule.

Images