A method and system for intelligent construction digital management

Abstract

The application relates to the technical field of construction management, in particular to a method and system for intelligent construction digital management, which comprises the following steps: integrating laser radar scanning data, construction machinery Internet of Things sensor data and design drawing parameters according to a building information model platform to construct a three-dimensional digital twin basement model; performing real-time mapping of a construction entity and a digital model through the three-dimensional digital twin basement model according to a point cloud registration algorithm to obtain construction site holographic image data; inputting the construction site holographic image data into a finite element analysis engine, performing dynamic simulation of a structural stress field according to a material mechanics parameter library to obtain structural stress distribution time sequence data under the action of a construction load. The three-dimensional digital twin basement model is constructed by integrating the laser radar scanning data, the construction machinery sensor data and the design drawing parameters, and the holographic image of the construction site can be obtained in real time; the high-precision data integration provides a comprehensive view of the construction site.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of construction management, and in particular to a method and system for intelligent construction digital management. BACKGROUND

[0002] Traditional methods usually rely on manual inspection, traditional measurement tools (such as total station, level, etc.) to collect field data, and these methods are low in efficiency and cannot realize real-time updating. The state of the scene often has a lag, and the accuracy and completeness of information acquisition are poor. Moreover, traditional methods often rely on static and periodic checks, and most use experience to judge the structure performance under load. High-precision simulation techniques such as finite element analysis are rarely used in traditional methods, and the data collection and analysis in the application process usually lack real-time performance. In traditional methods, the influence of temperature changes on materials such as concrete is usually ignored or only evaluated through simple theoretical calculations at the beginning of construction. During the construction process, the dynamic stress evolution of temperature changes cannot be effectively monitored and simulated. Moreover, traditional methods rely on manual detection and periodic checks to evaluate structural deformation, and usually only problems can be found after deformation reaches a certain level. Risk warnings are mostly based on expert experience or static data analysis, lacking intelligent and accurate prediction capabilities. In addition, traditional methods rely on manual decision-making and experience to assess risks, often relying on past experience, drawings, and engineering technical specifications for prevention and correction. This method lacks systematic risk analysis tools and cannot effectively handle complex multidimensional data. SUMMARY

[0003] The technical problem to be solved by the present application is to overcome the shortcomings of the prior art and provide a method and system for intelligent construction digital management.

[0004] The technical solution adopted to solve the above technical problems is: a method for intelligent construction digital management, comprising:

[0005] Integrating laser radar scanning data, construction machinery Internet of Things sensor data, and design drawing parameters based on a building information model platform to construct a three-dimensional digital twin basement model;

[0006] Performing real-time mapping of construction entities and digital models through the three-dimensional digital twin basement model based on a point cloud registration algorithm to obtain construction site holographic image data;

[0007] Inputting the construction site holographic image data into a finite element analysis engine and performing dynamic simulation of structural stress fields based on a material mechanics parameter library to obtain structural stress distribution time series data under construction load;

[0008] simulate the concrete hydration heat effect according to the structural stress distribution time series data to obtain temperature stress synergistic evolution characteristic data;

[0009] predict the structural deformation trend according to the temperature stress synergistic evolution characteristic data through the long short-term memory network to obtain a deformation prediction curve cluster;

[0010] evaluate the construction risk level according to the deformation prediction curve cluster and a preset deformation threshold through the fuzzy logic controller to obtain a structure safety risk grading early warning.

[0011] Preferably, the laser radar scanning data, the construction machinery Internet of Things sensor data and the design drawing parameters are integrated according to the building information model platform to construct a three-dimensional digital twin basement model, including:

[0012] The laser radar point cloud data, the construction machinery Internet of Things sensor data and the design drawing parameters are mapped to the same engineering coordinate system according to the coordinate system conversion engine built in the building information model platform;

[0013] The construction machinery Internet of Things sensor data includes GNSS positioning data, inclination sensor data and load monitoring data; the design drawing parameters include CAD drawing geometric parameters and Revit family parameters;

[0014] The axis intersection and the elevation control point in the design drawing are identified as reference features according to a feature point extraction algorithm; the construction machinery Internet of Things sensor data, the reference features and the design drawing parameters are mounted to the corresponding BIM components in a parameterized manner to obtain the three-dimensional digital twin basement model.

[0015] Preferably, the construction entity and the digital model are mapped in real time through the three-dimensional digital twin basement model according to a point cloud registration algorithm to obtain construction site holographic image data, including:

[0016] The structured features in the three-dimensional digital twin basement model and the geometric features in the laser radar scanning data are identified according to a feature descriptor extraction algorithm;

[0017] The structured features and the geometric features are co-registered according to a RANSAC algorithm to obtain an initial pose deviation between the three-dimensional digital twin basement model and the laser radar scanning data;

[0018] A feature point pair correlation table is determined according to the initial pose deviation, wherein the feature point pair correlation table includes an initial corresponding relationship between each component in the three-dimensional digital twin basement model and each point cloud cluster in the laser radar scanning data;

[0019] The multi-dimensional similarity measure between the three-dimensional digital twin base model and the lidar scanning data is determined based on the feature point association table, wherein the multi-dimensional similarity measure includes geometric distance, normal vector angle and semantic label;

[0020] According to the ICP algorithm, the structured features and geometric features are finely registered and iteratively optimized through the multi-dimensional similarity measure to obtain fine pose deviation. It is then determined whether the fine pose deviation converges to a preset registration threshold to obtain a fine feature point pair association table.

[0021] Based on the detailed feature point pair association table, each point cloud cluster in the lidar scanning data is mapped in real time to the three-dimensional digital twin base model to obtain holographic image data of the construction site.

[0022] Preferably, the holographic image data of the construction site is input into a finite element analysis engine, and dynamic simulation of the structural stress field is performed based on a material mechanics parameter library to obtain time-series data of structural stress distribution under construction loads, including:

[0023] The point cloud and model hybrid data in the holographic image data are divided into grids based on the geometric topology information of the three-dimensional digital twin base model to obtain an unstructured grid.

[0024] The unstructured mesh is hierarchically organized according to the octree spatial index to obtain a hybrid mesh model;

[0025] The hybrid mesh model includes two types of hybrid mesh models with different granularities: coarse-grained hybrid mesh models are used for global stress field overview, and fine-grained hybrid mesh models are used for refined analysis of key parts.

[0026] Preferably, the holographic image data of the construction site is input into a finite element analysis engine, and dynamic simulation of the structural stress field is performed based on a material mechanics parameter library to obtain time-series data of structural stress distribution under construction loads. This also includes:

[0027] Material properties associated with each component in the three-dimensional digital twin substrate model are extracted from the material mechanics parameter library, and the properties are automatically mounted according to the component ID and material ID mapping table in the three-dimensional digital twin substrate model.

[0028] The load spatiotemporal distribution matrix is ​​determined based on the IoT sensor data of the construction machinery. The load spatiotemporal distribution matrix is ​​then applied to the finite element model in the form of equivalent nodal forces. The finite element model is solved step by step according to the construction cycle, and stress field data under the current state is generated at each time step.

[0029] The stress field data of each time step is aggregated according to the spatial position to obtain structural stress distribution time series data evolving over time.

[0030] Preferably, the concrete hydration heat effect simulation is performed according to the structural stress distribution time series data to obtain temperature stress synergistic evolution characteristic data, including:

[0031] The heat conduction calculation grid is determined according to the geometric topology information of the three-dimensional digital twin base model; the thermal force bidirectional coupling relationship is defined according to the heat conduction calculation grid and the structural stress distribution time series data, wherein the thermal force bidirectional coupling relationship includes the driving action of the temperature field on the material thermal expansion and the constraint effect of the structural stress field on the heat conduction path;

[0032] The concrete hydration heat parameters are called from the material mechanics parameter library, including the development curve between the age and the strength and the temperature and the elastic modulus correction coefficient;

[0033] The temperature field distribution is determined according to the heat conduction equation, and the structural stress field is updated through the temperature field distribution according to the thermal force bidirectional coupling relationship.

[0034] Preferably, the concrete hydration heat effect simulation is performed according to the structural stress distribution time series data to obtain temperature stress synergistic evolution characteristic data, further including:

[0035] The three-dimensional temperature field envelope surface data is determined according to the temperature field distribution, wherein the three-dimensional temperature field envelope surface data includes the time history curve of the core temperature, the surface temperature and the internal and external temperature difference of each component;

[0036] The thermal stress distribution nephogram sequence is determined according to the three-dimensional temperature field envelope surface data and the updated structural stress field;

[0037] The temperature stress synergistic evolution characteristic data is determined according to the thermal stress distribution nephogram sequence, wherein the temperature stress synergistic evolution characteristic data includes the time sequence correlation analysis of the core temperature and the first principal stress.

[0038] Preferably, the structure deformation trend prediction is performed through the temperature stress synergistic evolution characteristic data according to the long short-term memory network to obtain a deformation prediction curve cluster, including:

[0039] The feature extraction is performed on the temperature stress synergistic evolution characteristic data to obtain a time series feature matrix, including the concrete component core temperature time history curve, the thermal stress peak sequence and the temperature stress coupling index;

[0040] The geometric information and material properties of each component in the three-dimensional digital twin base model are taken as static features; the multidimensional input feature set is determined according to the time series feature matrix and the static features;

[0041] The structural deformation trend is predicted by a long short-term memory network based on the multi-dimensional input feature set to obtain a deformation prediction curve cluster, wherein the deformation prediction curve cluster includes a displacement time history curve, a strain development curve, and a radius of curvature change curve.

[0042] Preferably, the construction risk level is evaluated by a fuzzy logic controller based on the deformation prediction curve cluster and a preset deformation threshold to obtain a structural safety risk classification warning, including:

[0043] A preset deformation threshold library is obtained, wherein the preset deformation threshold library includes limit thresholds of the displacement time history curve, the strain development curve, and the radius of curvature change curve;

[0044] The deformation prediction curve cluster is defined as an input variable of the fuzzy logic controller, and a membership function is determined based on the deformation prediction curve cluster and the preset deformation threshold library, wherein the membership function is used to divide the displacement time history curve into three displacement levels and the strain development curve into three strain levels;

[0045] A fuzzy rule library is determined based on an IF-THEN rule, the input variable is mapped to the membership function to obtain a fuzzy set, and the fuzzy set is matched with the fuzzy rule library to obtain the structural safety risk classification warning.

[0046] The technical solution adopted to solve the above technical problems is: a system for intelligent construction digital management, which is applicable to the method for intelligent construction digital management, and includes:

[0047] A three-dimensional modeling unit is configured to integrate laser radar scanning data, construction machinery Internet of Things sensor data, and design drawing parameters based on a building information model platform to construct a three-dimensional digital twin base model;

[0048] An entity mapping unit is configured to perform real-time mapping of construction entities and digital models based on the three-dimensional digital twin base model by using a point cloud registration algorithm to obtain construction site holographic image data;

[0049] A stress analysis unit is configured to input the construction site holographic image data into a finite element analysis engine, perform dynamic simulation of a structural stress field based on a material mechanics parameter library, and obtain structural stress distribution time series data under construction load;

[0050] A collaborative evolution unit is configured to simulate a concrete hydration heat effect based on the structural stress distribution time series data to obtain temperature stress collaborative evolution characteristic data;

[0051] a trend prediction unit configured to predict a structural deformation trend according to a long short-term memory network through the temperature stress co-evolution feature data to obtain a deformation prediction curve cluster;

[0052] a hierarchical early warning unit configured to evaluate a construction risk level according to a fuzzy logic controller through the deformation prediction curve cluster and a preset deformation threshold to obtain a structural safety risk hierarchical early warning.

[0053] The beneficial effects of the present application are as follows: (1) The present application can obtain a holographic image of the construction site in real time by integrating laser radar scanning data, construction machinery sensor data and design drawing parameters to construct a three-dimensional digital twin basement model. This high-precision data integration provides a comprehensive view of the construction site. The structural stress distribution during the construction process can be simulated by a finite element analysis engine, which can dynamically evaluate the performance of the structure under different loads during the construction process, which helps to identify potential structural problems such as overload, material fatigue or uneven stress distribution in advance, and avoid safety hazards during construction; (2) The present application simulates the hydration heat effect of concrete and obtains temperature stress co-evolution feature data, which is crucial for stress analysis of materials such as concrete, especially in large-scale construction projects. Temperature changes may cause uneven stress distribution in concrete structures, which may affect structural safety. This method realizes detailed analysis of temperature stress during construction through this step; and the long short-term memory network is used to predict the temperature stress co-evolution feature data to obtain the deformation trend of the structure, which enables the construction team to predict the possible deformation of the structure, such as displacement, strain or bending change, in advance, and take necessary measures to avoid structural damage; (3) The present application uses a fuzzy logic controller to evaluate the risk according to the deformation prediction curve and the preset deformation threshold, which can provide real-time hierarchical early warning of structural safety risk according to different deformation states. This evaluation can accurately distinguish between low, medium and high risk levels to provide timely warning for construction personnel, helping them adjust the construction plan, reinforce or stop work in time to avoid construction accidents; and it can provide timely risk assessment and decision support during construction to help the project management team make more accurate risk judgments before, during and after construction, thereby reducing the probability of accidents and ensuring the safety of construction. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 A step flowchart of the overall method in one embodiment of the present application is shown in the figure;

[0055] Figure 2 A system architecture diagram of the overall system in one embodiment of the present application is shown in the figure.

[0056] Reference signs: 1, three-dimensional modeling unit; 2, entity mapping unit; 3, stress analysis unit; 4, synergistic evolution unit; 5, trend prediction unit; 6, hierarchical early warning unit. DETAILED DESCRIPTION

[0057] In example one, as shown, the present application proposes a method for intelligent construction digital management, comprising: Figure 1

[0058] S1, integrating laser radar scanning data, construction machinery Internet of Things sensor data and design drawing parameters according to the building information model platform, and constructing a three-dimensional digital twin basement model;

[0059] S2, according to the point cloud registration algorithm, the real-time mapping of the construction entity and the digital model is carried out through the three-dimensional digital twin basement model, so as to obtain the holographic image data of the construction site;

[0060] S3, inputting the holographic image data of the construction site into the finite element analysis engine, and carrying out dynamic simulation of the structural stress field according to the material mechanics parameter library, so as to obtain the structural stress distribution time series data under the action of construction load;

[0061] S4, according to the structural stress distribution time series data, the hydration heat effect of concrete is simulated, so as to obtain the temperature stress synergistic evolution characteristic data;

[0062] S5, according to the long short-term memory network, the structural deformation trend is predicted through the temperature stress synergistic evolution characteristic data, so as to obtain the deformation prediction curve cluster;

[0063] S6, according to the fuzzy logic controller, the construction risk level is evaluated through the deformation prediction curve cluster and the preset deformation threshold, so as to obtain the hierarchical early warning of structural safety risk.

[0064] In example two, the present application proposes a method for intelligent construction digital management, compared with example one, the present embodiment further comprises: integrating laser radar scanning data, construction machinery Internet of Things sensor data and design drawing parameters according to the building information model platform, and constructing a three-dimensional digital twin basement model, comprising:

[0065] A1, according to the coordinate system conversion engine built in the building information model platform, the laser radar point cloud data, the construction machinery Internet of Things sensor data and the design drawing parameters are mapped to the same engineering coordinate system;

[0066] Among them, the construction machinery Internet of Things sensor data includes GNSS positioning data, inclination sensor data and load monitoring data; the design drawing parameters include CAD drawing geometric parameters and Revit family parameters;

[0067] ​A2, the axis intersection points and elevation control points in the design drawing are identified as reference features according to the feature extraction algorithm; the construction machinery Internet of Things sensor data, the reference features and the design drawing parameters are mounted to the corresponding BIM components in a parameterized manner to obtain a three-dimensional digital twin base model.

[0068] In this embodiment, the building information modeling platform typically has its own coordinate system, and all BIM data is based on this coordinate system; different data sources (such as laser radar, Internet of Things sensor data, etc.) usually have different coordinate systems (for example, laser radar point cloud data may have its own local coordinate system); laser radar point cloud data, construction machinery Internet of Things sensor data and design drawing parameters (such as CAD drawing geometric data and Revit family parameters) need to be converted and mapped into a unified engineering coordinate system to ensure the docking and fusion between different data sources; through the positioning information provided by the global satellite positioning system, accurate geographic coordinate conversion can be realized to ensure the accuracy of the model position; tilt sensor and load monitoring data can provide attitude information of construction site machinery (such as the inclination angle of mechanical equipment, load, etc.), and their conversion also needs to be based on a unified coordinate system to ensure their correct matching with other building data; CAD drawings usually provide two-dimensional plane designs, which may include information such as the layout, size and spatial relationship of the building; Revit is a software commonly used in architectural design, and its family parameters contain geometric and attribute data of building components (such as walls, beams, columns, doors and windows, etc.), which are usually 3D model information; this step uses a feature point extraction algorithm to identify important points in the design drawing, such as axis intersection points and elevation control points; these points are key reference points in architectural design and are usually used to locate important information such as the center line of the building structure and the reference level; axis intersection points usually refer to the intersection points of the main axes in architectural design, which determine the layout and symmetry of the building; elevation control points are used to indicate the height difference between different levels or horizontal planes of the building structure; the feature point extraction algorithm automatically identifies these key points from the design drawing, providing accurate data support for subsequent modeling; the reference features (axis intersection points, elevation control points) extracted from the drawing are combined with other data (such as sensor data of construction machinery); sensor data provides real-time information, such as the actual position and state of the machinery; these data are mounted on the corresponding BIM components (such as walls, columns, floors, etc.) through parameterization, ensuring the accurate reflection of the model in the digital space; parameterization means that data and design elements are bound together through variables or parameters, so that each BIM component can be dynamically adjusted or updated when needed; the advantage of parameterized design is that when the data source changes, the model will automatically update; three-dimensional digital twin refers to a digital model created by highly integrated multi-source data (such as laser radar point cloud, sensor data, design drawing, etc.); this model is a virtual copy of the building and can be used for simulation, analysis, prediction, etc.; the base model refers to building a basic three-dimensional framework that covers the basic structure and necessary parameters of the building, which provides a basis for subsequent more detailed digital twin modeling.

[0069] In an optional embodiment, real-time mapping of construction entities and digital models through a three-dimensional digital twin base model according to a point cloud registration algorithm is performed to obtain construction site holographic image data, including:

[0070] B1, identifying structured features in the three-dimensional digital twin base model and geometric features in the laser radar scanning data according to a feature descriptor extraction algorithm;

[0071] B2, performing coarse registration of the structured features and the geometric features according to a RANSAC algorithm to obtain an initial pose deviation between the three-dimensional digital twin base model and the laser radar scanning data;

[0072] B3, determining a feature point pair association table according to the initial pose deviation, wherein the feature point pair association table includes an initial correspondence relationship between each component in the three-dimensional digital twin base model and each point cloud cluster in the laser radar scanning data;

[0073] B4, determining a multi-dimensional similarity measure between the three-dimensional digital twin base model and the laser radar scanning data according to the feature point pair association table, wherein the multi-dimensional similarity measure includes a geometric distance, a normal vector angle, and a semantic label;

[0074] B5, performing fine registration and iterative optimization of the structured features and the geometric features according to an ICP algorithm through the multi-dimensional similarity measure to obtain a fine pose deviation, judging whether the fine pose deviation converges to a preset registration threshold, and obtaining a fine feature point pair association table;

[0075] B6, real-time mapping each point cloud cluster in the laser radar scanning data to the three-dimensional digital twin base model according to the fine feature point pair association table to obtain the construction site holographic image data.

[0076] It should be noted that in point cloud data processing, feature descriptors are mathematical tools used to distinguish and identify unique structural features in point clouds. For example, features such as planes, lines, and corners can be identified using feature descriptors. A 3D digital twin base model contains basic structural information about a building, including walls, beams, and columns, while LiDAR scan data contains point cloud information from a real construction site. Feature descriptor extraction algorithms can identify and match the geometric features (such as lines, planes, and corners) and structural features between these two, thus laying the foundation for the subsequent registration process. The RANSAC algorithm is a commonly used robust registration algorithm; it calculates registration parameters by randomly selecting a subset of points, eliminating anomalies. The impact of noise points on the results; in this step, the RANSAC algorithm is used to compare structured features (such as building axes, floor elevations, etc.) and geometric features (such as planes, corner points, etc. in the point cloud) for coarse registration; through this process, the initial pose deviation between the 3D digital twin base model and the LiDAR scanning data can be calculated, that is, the position and orientation difference between the model and the scanning data; the feature point pair association table is a table that records the preliminary correspondence between each component (such as walls, doors, windows, columns, etc.) in the 3D digital twin base model and each point cloud cluster (i.e., the clustered area or specific part in the point cloud) in the LiDAR scanning data; based on the initial coarse registration results, the feature point pair association table helps to establish the relationship between the digital model and the actual points. The correspondence between points in the cloud prepares for fine-grained registration. To determine the degree of matching between the 3D digital twin base model and the LiDAR scanning data, a multi-dimensional similarity assessment is required. Specific metrics include: geometric distance (the geometric distance between points in the point cloud and components in the digital model); normal vector angle (measuring the directional consistency between the point cloud and the model surface; a normal vector is a vector perpendicular to the surface, representing the surface orientation); semantic labels (each point or cluster of points may carry semantic information, such as walls, columns, and ground; semantic labels can determine the consistency of function and type between the point cloud and the digital model); multi-dimensional similarity metrics can comprehensively evaluate the matching accuracy between the digital model and the LiDAR scanning data; and the ICP algorithm. It is a classic point cloud registration algorithm used to accurately align two point clouds. It gradually reduces registration errors by continuously optimizing the matching relationship between each pair of nearest points in the point cloud. In this step, the ICP algorithm performs fine registration of structured and geometric features based on the multi-dimensional similarity measure in the previous step, and continuously adjusts the alignment result between the model and the point cloud through iterative optimization. This process ultimately yields accurate pose deviation, i.e., the fine alignment difference between the 3D digital twin base model and the LiDAR scanning data. It then determines whether a preset registration threshold has been reached to ensure that the registration error is within an acceptable range. Through a fine feature point pair association table, it accurately maps each point cloud cluster in the LiDAR scanning data to a specific component in the 3D digital twin base model.This step realizes the process of real-time mapping of lidar scanning data to the digital model, that is, through the matching of the geometric and structural features of the model with the point cloud information in the scanning data, the virtual representation of the construction site is updated in real time; the final construction site holographic image data is a digital representation of the construction site, combining the lidar scanning data in the actual construction process and the existing three-dimensional digital twin base model; this holographic image not only shows the spatial structure of the building, but also reflects the dynamic changes on site, such as progress tracking, material stacking, mechanical equipment position, etc.

[0077] In an optional embodiment, the construction site holographic image data is input into a finite element analysis engine, and a dynamic simulation of the structural stress field is performed according to a material mechanics parameter library to obtain structural stress distribution time series data under construction load, including:

[0078] C1, grid division is performed on the point cloud and model mixed data in the holographic image data according to the geometric topology information of the three-dimensional digital twin base model to obtain an unstructured grid;

[0079] C2, hierarchical organization is performed on the unstructured grid according to an octree spatial index to obtain a hybrid grid model;

[0080] The hybrid grid model includes a hybrid grid model of two granularities, coarse and fine, wherein the hybrid grid model of coarse granularity is used for global stress field overview, and the hybrid grid model of fine granularity is used for key part detailed analysis.

[0081] It should be noted that the geometric topology information of the three-dimensional digital twin base model refers to the structural information (such as walls, beams, columns, etc.) of the digital building model; these information is used to grid the point cloud and hybrid data in the holographic image data; due to the structure and complexity of the building model, the point cloud data may not form regular grids, so grid division is needed to convert the point cloud data into unstructured grids suitable for finite element analysis; unstructured grids are irregular and flexible, suitable for handling complex geometries, and can more finely represent various structural features of the building; octree spatial index is a high-efficiency data structure for large-scale three-dimensional data management, especially suitable for spatial organization of three-dimensional point cloud data; octree divides space into eight subspaces recursively, creating a hierarchical index, which helps to quickly find and process data; mixed grid model with coarse and fine granularity: this is a model that combines different grid granularities to achieve simulation with different precision; coarse-grained grid is mainly used to describe the general stress field of the entire structure; it is suitable for global analysis and large-scale calculation, and can quickly evaluate the stress distribution of the overall structure; fine-grained grid is used for more detailed analysis of key parts, such as load-bearing columns and important beams, which may bear greater stress concentration during construction, so higher grid resolution is needed to improve simulation accuracy; material mechanics parameter library contains the mechanical properties of various building materials (such as elastic modulus, Poisson's ratio, tensile strength, etc.), which are crucial for stress calculation of the structure.

[0082] In an optional embodiment, the construction site holographic image data is input into the finite element analysis engine, and the structural stress field dynamic simulation is performed according to the material mechanics parameter library to obtain the structural stress distribution time series data under the action of construction load, further comprising:

[0083] C3, extracting the material properties associated with each component in the three-dimensional digital twin base model from the material mechanics parameter library, and automatically mounting the properties according to the mapping table of component ID and material ID in the three-dimensional digital twin base model;

[0084] C4, determining the load space-time distribution matrix according to the construction machinery Internet of Things sensor data, and applying the load space-time distribution matrix to the finite element model in the form of equivalent node force, wherein the finite element model is solved step by step according to the construction rhythm, and the stress field data at the current state is generated at each time step;

[0085] C5, aggregating the stress field data of each time step according to the spatial position to obtain the structural stress distribution time series data evolving with time.

[0086] It should be noted that the automatic mounting of material properties automatically mounts the corresponding material properties for each structural member according to the mapping table of the component ID and material ID in the three-dimensional digital twin base model; in this way, in finite element analysis, different materials will be calculated according to their physical properties to ensure the accuracy of the simulation; construction machinery Internet of Things sensor data: mechanical equipment on the construction site is usually equipped with sensors that can collect load data (such as the working state of mechanical equipment, the applied force, etc.) in real time; these data help define the load space-time distribution matrix, i.e., the distribution of load in space and time; equivalent node force application: according to the load space-time distribution matrix, the load is converted into equivalent node force and applied to the finite element model; node force is a basic input in the finite element method, which determines the external load on the structure; step-by-step solution according to construction rhythm: in actual construction, the load is dynamically changing, so the finite element model is solved step by step according to the construction progress; each time step corresponds to a state in the construction process, and the effect of construction load is simulated step by step; time step stress field data aggregation: at each time step, the structure stress data at the current time is calculated through finite element analysis; these stress data are aggregated in space according to the position to form the stress field data that changes with time; structure stress distribution time series data: by time series aggregation of stress data at each time step, the stress evolution process during the entire construction process is obtained; these data can help evaluate the stress distribution of the structure during the construction process and provide a basis for construction safety and design optimization.

[0087] In an optional embodiment, concrete hydration heat effect simulation is performed according to the structure stress distribution time series data to obtain temperature stress co-evolution characteristic data, including:

[0088] D1, determining a heat conduction calculation grid according to the geometric topology information of the three-dimensional digital twin base model; defining a thermal-mechanical bidirectional coupling relationship according to the heat conduction calculation grid and the structure stress distribution time series data, wherein the thermal-mechanical bidirectional coupling relationship includes the driving action of the temperature field on the thermal expansion of the material and the constraint effect of the structure stress field on the heat conduction path;

[0089] D2, calling concrete hydration heat parameters from a material mechanics parameter library, the concrete hydration heat parameters including a development curve between age and strength and a temperature and elastic modulus correction coefficient;

[0090] D3, determining a temperature field distribution according to a heat conduction equation, and updating the structure stress field through the temperature field distribution according to the thermal-mechanical bidirectional coupling relationship.

[0091] It should be noted that the heat conduction calculation grid refers to the structural stress field simulation, and the heat conduction calculation also needs to discretize the space through the grid; here, the heat conduction grid is determined according to the geometric topological information of the three-dimensional digital twin base model, to ensure that the shape and size characteristics of the structure are considered in the calculation of heat conduction; the thermal two-way coupling relationship refers to the mutual influence relationship between the temperature field and the structural stress field; temperature change leads to thermal expansion of the material, thereby generating stress, and these stresses in turn affect the heat conduction path and speed; specifically including: the driving action of the temperature field on the thermal expansion of the material: as the temperature of the concrete rises, the material will undergo thermal expansion, causing additional stress to the structure; the constraint effect of the structural stress field on the heat conduction path: the stress within the structure can change the path or rate of heat conduction; for example, in the areas of the structure under compression, the diffusion of heat may be limited; concrete hydration heat parameters: hydration heat is the heat released during the chemical reaction of concrete hardening, which is usually related to the age and strength of the concrete; here, the concrete hydration heat parameters in the material mechanics parameter library are used, including: age and strength development curve: this describes the growth of concrete strength and the corresponding change of hydration heat over time; generally, there is a certain relationship between hydration heat and strength development, the initial hydration heat is high, and it gradually decreases later; temperature and elastic modulus correction coefficient: the elastic modulus of concrete changes with temperature; under high temperature conditions, the elastic modulus may decrease, thereby affecting the stress state of the structure; simulate the propagation of heat in concrete according to the heat conduction equation; the heat conduction equation will consider factors such as the distribution of the temperature field, boundary conditions, and heat sources, to calculate the distribution of the temperature field at each time step; update the structural stress field through the thermal two-way coupling relationship: the two-way coupling between the temperature field and the stress field means that after calculating the temperature field at each time step, the stress state of the structure needs to be updated; the expansion stress caused by temperature needs to affect the stress field of the structure in turn, thereby affecting the overall stress state of the building.

[0092] In an optional embodiment, the concrete hydration heat effect simulation according to the structural stress distribution time series data to obtain the temperature stress synergistic evolution characteristic data further includes:

[0093] D4, determining three-dimensional temperature field envelope surface data according to the temperature field distribution, wherein the three-dimensional temperature field envelope surface data includes core temperature, surface temperature and internal and external temperature difference time history curves of each component;

[0094] D5, determining a thermal stress distribution nephogram sequence according to the three-dimensional temperature field envelope surface data and the updated structural stress field;

[0095] D6, determining temperature stress synergistic evolution characteristic data according to the thermal stress distribution nephogram sequence, wherein the temperature stress synergistic evolution characteristic data includes time series correlation analysis of core temperature and first principal stress.

[0096] It should be noted that the three-dimensional temperature field envelope surface data: the purpose of this step is to obtain the temperature evolution characteristics of the concrete structure; the temperature field distribution obtained by simulation can extract temperature data at different positions (such as the core, the surface) and different time points, forming time series data of the temperature field; the core temperature represents the temperature inside the structure, the surface temperature represents the temperature outside the structure, and the temperature difference between the inside and outside reflects the heat difference between the inside and outside of the structure; the time history curve represents how these temperature differences evolve over time; the thermal stress distribution cloud diagram can be obtained according to the updated structure stress field and temperature field; these graphics show the stress distribution in the structure due to temperature changes at different time points; thermal stress is usually concentrated in areas with large temperature gradients, such as the core and surface areas of concrete; the thermal stress co-evolution characteristic data is a quantitative analysis of the co-evolution characteristics of temperature and stress, which usually includes: core temperature and first principal stress time series correlation analysis: by analyzing the relationship between the temperature of the core part of the concrete and the principal stress (especially the first principal stress), it reveals how temperature and stress changes affect each other and co-evolve; through time series data, the coupling strength between the two can be analyzed to determine whether the change in temperature will exacerbate stress concentration in the structure, or how the stress distribution changes under different temperature conditions.

[0097] In an optional embodiment, the structure deformation trend prediction based on the long short-term memory network through the temperature stress co-evolution characteristic data is performed to obtain a deformation prediction curve cluster, including:

[0098] E1, feature extraction is performed on the temperature stress co-evolution characteristic data to obtain a time series feature matrix, the time series feature matrix including a concrete member core temperature time history curve, a thermal stress peak value sequence, and a temperature stress coupling index;

[0099] E2, the geometric information and material properties of each member in the three-dimensional digital twin base model are taken as static features; a multi-dimensional input feature set is determined according to the time series feature matrix and the static features;

[0100] E3, the structure deformation trend prediction is performed based on the long short-term memory network through the multi-dimensional input feature set to obtain a deformation prediction curve cluster, wherein the deformation prediction curve cluster includes a displacement time history curve, a strain development curve, and a curvature radius change curve.

[0101] It is necessary to extract meaningful time series features from the temperature stress co-evolution characteristic data before performing time series prediction; these features are the basis for the patterns and rules that the LSTM model can capture; the concrete core temperature time history curve describes the temperature changes of the concrete core at different time points; this is a direct manifestation of the concrete hydration heat effect and one of the main factors affecting the generation and development of thermal stress; the thermal stress peak sequence represents the peak value of thermal stress generated due to temperature changes at each time step; these peaks reflect the highest point of temperature-induced stress in the structure, usually occurring when the temperature gradient is the largest; the temperature stress coupling index measures the coupling relationship between temperature changes and stress changes; for example, the interaction between temperature-induced expansion stress and stress-induced heat conduction effect, this index can reveal the degree of coupling between the two; through these time series features, a time series feature matrix can be constructed, where each row represents the feature data at a time point, which will be input into the LSTM model for further learning and prediction; In addition to time series data, the static characteristics of the structure are also crucial for deformation trend prediction; static characteristics include: geometric information of each component in the three-dimensional digital twin base model: refers to the size, shape and geometric properties of the concrete structure; the geometry of the structure will affect the distribution of temperature field and stress field, thereby affecting the deformation trend; the material properties of concrete (such as elastic modulus, tensile strength, thermal expansion coefficient, etc.) directly affect the stress response caused by temperature changes; the LSTM model needs to consider both time series data (such as temperature, stress, etc.) and static data (such as geometric information, material properties, etc.); Therefore, it is necessary to combine the time series feature matrix and static features to construct a multi-dimensional input feature set for the LSTM model to learn; Long Short-Term Memory Network (LSTM) is a type of Recurrent Neural Network (RNN) specifically designed to handle time series data; it can maintain and pass historical information through its gating mechanism, effectively solving the gradient vanishing problem that ordinary RNNs face when dealing with long sequence data; LSTM is particularly suitable for long-term dependent data prediction tasks such as structure deformation prediction; LSTM learns the complex relationship between temperature, stress and structural properties by inputting the multi-dimensional input feature set (time series feature matrix + static features) into the network for training, and predicts future deformation trends; the trained LSTM model can predict the structure deformation trend in the future, outputting a series of deformation prediction curve clusters; these curves describe the future deformation state of the structure, including: displacement time history curve: shows the displacement changes of nodes (such as concrete components) in the structure at different time points; displacement reflects the overall movement trend of the structure under external forces (such as temperature, load, etc.); strain development curve: reflects the change process of strain in different parts of concrete or structure over time;Strain is usually directly related to the stress state of the structure, and can reveal the deformation degree of the structure under the influence of temperature change, etc.; curvature radius change curve: curvature radius describes the bending degree of the structure, and with the accumulation of stress and temperature change, the structure may be bent and deformed; the curvature radius change curve can describe the evolution of the bending degree of the structure with time.

[0102] In an optional embodiment, construction risk level assessment is performed according to the fuzzy logic controller through the deformation prediction curve cluster and the preset deformation threshold value, to obtain a structure safety risk grading early warning, including:

[0103] F1, obtaining a preset deformation threshold value library, wherein the preset deformation threshold value library includes limit threshold values of the displacement time history curve, the strain development curve and the curvature radius change curve;

[0104] F2, defining the deformation prediction curve cluster as an input variable of the fuzzy logic controller, and determining a membership function according to the deformation prediction curve cluster and the preset deformation threshold value library, wherein the membership function is used to divide the displacement time history curve into three displacement levels and divide the strain development curve into three strain levels;

[0105] F3, determining a fuzzy rule library according to the IF-THEN rule, mapping the input variable to the membership function to obtain a fuzzy set, and matching the fuzzy set with the fuzzy rule library to obtain the structure safety risk grading early warning.

[0106] It should be noted that the preset deformation threshold library contains the limiting threshold of each deformation prediction curve, which is used to judge whether the structural deformation exceeds the safe range; the limiting threshold of the displacement time history curve represents the displacement variation range of the structure at different time points; if the displacement of the structure exceeds a certain threshold, it may mean that the structure has undergone abnormal deformation and needs to be alerted; the limiting threshold of the strain development curve represents the upper limit value of the strain development of the structure; if the strain development exceeds the safe threshold, it may cause permanent damage or instability of the structure; the limiting threshold of the curvature radius change curve describes the bending degree of the structure, and exceeding a certain curvature radius change range may mean that the structure has undergone excessive bending, and even structural failure may occur; these limiting thresholds are pre-set, usually based on engineering experience, standard specifications or historical data, to judge whether the structure is in a dangerous state; the deformation prediction curve cluster is obtained by predicting the displacement time history curve, strain development curve and curvature radius change curve using the LSTM model, which reflects the deformation trend of the structure in the future period of time; these deformation prediction curve clusters (i.e. displacement, strain and curvature change) will be used as input variables of the fuzzy logic controller; each curve can be represented as a time series data or a set of numerical values, reflecting the deformation state at different time steps; in order to combine continuous prediction curves with fuzzy logic, membership functions need to be defined to convert the numerical values of prediction curves into fuzzy sets; the membership function is used to represent the degree to which a numerical value belongs to a certain fuzzy set, usually with a value range of [0, 1]; for the displacement time history curve, the membership function divides the displacement value into three displacement levels (such as normal, warning, and overrun), each level representing a different safety state; for the strain development curve, the membership function divides the strain value into three strain levels (such as low, medium, and high), each level representing a different strain intensity, reflecting the threat of deformation to the safety of the structure; for the curvature radius change curve, the membership function also divides the curvature change into different levels to evaluate the bending degree of the structure; these membership functions provide a basis for the construction of the subsequent fuzzy rule base, converting continuous numerical data into fuzzy level form; the fuzzy rule base in the fuzzy logic controller uses IF-THEN rules to describe the relationship between input variables and output; here, the fuzzy rule base will contain multiple rules for inferring the risk level of the structure according to the input deformation prediction curves (such as displacement, strain, and curvature change); IF-THEN rule example: IF displacement level is "overrun" AND strain level is "high" THEN structure risk is "high"; IF displacement level is "normal" AND strain level is "low" THEN structure risk is "low"; IF curvature level is "high" THEN structure risk is "medium"; the right side of the rule (THEN part) represents the safety risk level of the structure, which may be "low", "medium" or "high";Fuzzy set and rule base matching: after mapping the membership functions of input variables (displacement, strain, curvature) to fuzzy sets, the fuzzy controller uses IF-THEN rules in the rule base to perform reasoning; this reasoning outputs a fuzzy set representing the risk level of the structure; this fuzzy set represents the degree of fuzziness of different risk levels, for example, it may contain a combination of "low risk" (50%), "medium risk" (30%), and "high risk" (20%); structure safety risk classification warning: finally, the fuzzy controller generates a structure safety risk classification warning according to the reasoning result, that is, it assigns a risk level to the structure according to the fuzzy reasoning result and provides a warning to the corresponding construction personnel or management personnel; the risk level can be: low risk: the structure is within the normal deformation range, and the construction can continue; medium risk: the structure deformation is close to the preset threshold, which needs to be monitored and further analyzed; high risk: the structure has exceeded the safe deformation range, and immediate measures such as stopping work, reinforcing or repairing need to be taken.

[0107] In one embodiment, as shown in Figure 2 The system is applicable to the method and comprises:

[0108] A three-dimensional modeling unit 1 is configured to integrate laser radar scanning data, construction machinery Internet of Things sensor data, and design drawing parameters based on a building information model platform to construct a three-dimensional digital twin base model.

[0109] An entity mapping unit 2 is configured to perform real-time mapping of construction entities and digital models based on a point cloud registration algorithm through the three-dimensional digital twin base model to obtain construction site holographic image data.

[0110] A stress analysis unit 3 is configured to input the construction site holographic image data into a finite element analysis engine, perform dynamic simulation of a structure stress field based on a material mechanics parameter library, and obtain structure stress distribution time series data under construction load.

[0111] A co-evolution unit 4 is configured to simulate a concrete hydration heat effect based on the structure stress distribution time series data to obtain temperature stress co-evolution characteristic data.

[0112] A trend prediction unit 5 is configured to perform structure deformation trend prediction based on a long short-term memory network through the temperature stress co-evolution characteristic data to obtain a deformation prediction curve cluster.

[0113] A graded warning unit 6 is configured to perform construction risk level assessment based on a fuzzy logic controller through the deformation prediction curve cluster and a preset deformation threshold to obtain a structure safety risk classification warning.

[0114] The embodiments of the present application are described in detail above with reference to the accompanying drawings, but the present application is not limited to the embodiments. Various changes that one skilled in the art can make within the scope of the knowledge of the art without departing from the spirit of the present application are also possible.

Claims

1. A method for intelligent construction digital management, characterized in that, include: A three-dimensional digital twin base model is constructed by integrating LiDAR scanning data, construction machinery IoT sensor data, and design drawing parameters from the building information modeling platform. Based on the point cloud registration algorithm, the construction entity and the digital model are mapped in real time through the three-dimensional digital twin base model to obtain holographic image data of the construction site; The holographic image data of the construction site is input into the finite element analysis engine, and the dynamic simulation of the structural stress field is performed according to the material mechanics parameter library to obtain the time series data of structural stress distribution under construction load. Based on the time series data of the structural stress distribution, the hydration heat effect of concrete is simulated to obtain the temperature stress co-evolution characteristic data. Based on the long short-term memory network, structural deformation trend is predicted using the temperature stress co-evolution characteristic data to obtain a cluster of deformation prediction curves; The fuzzy logic controller performs a construction risk level assessment based on the deformation prediction curve cluster and a preset deformation threshold to obtain a structural safety risk classification and early warning.

2. The method for intelligent construction digital management according to claim 1, characterized in that, Based on the integration of LiDAR scanning data, construction machinery IoT sensor data, and design drawing parameters from the Building Information Modeling (BIM) platform, a three-dimensional digital twin base model is constructed, including: Based on the coordinate system transformation engine built into the building information modeling platform, the lidar point cloud data, construction machinery IoT sensor data and design drawing parameters are mapped to the same engineering coordinate system; The construction machinery IoT sensor data includes GNSS positioning data, tilt sensor data, and load monitoring data; the design drawing parameters include CAD drawing geometric parameters and Revit family parameters. The intersection points of axes and elevation control points in the design drawings are identified as reference features based on the feature point extraction algorithm. The sensor data of the construction machinery IoT sensor, the reference features and the parameters of the design drawings are parametrically mounted to the corresponding BIM components to obtain a three-dimensional digital twin base model.

3. The method for intelligent construction digital management according to claim 2, characterized in that, Based on the point cloud registration algorithm, the construction entity and the digital model are mapped in real time through the three-dimensional digital twin base model to obtain holographic image data of the construction site, including: The structured features in the 3D digital twin substrate model and the geometric features in the lidar scanning data are identified based on the feature descriptor extraction algorithm. The structured features and geometric features are coarsely registered using the RANSAC algorithm to obtain the initial pose deviation between the 3D digital twin substrate model and the lidar scanning data. A feature point pair association table is determined based on the initial pose deviation, wherein the feature point pair association table includes the initial correspondence between each component in the three-dimensional digital twin base model and each point cloud cluster in the lidar scanning data; The multi-dimensional similarity measure between the three-dimensional digital twin base model and the lidar scanning data is determined based on the feature point association table, wherein the multi-dimensional similarity measure includes geometric distance, normal vector angle and semantic label; According to the ICP algorithm, the structured features and geometric features are finely registered and iteratively optimized through the multi-dimensional similarity measure to obtain fine pose deviation. It is then determined whether the fine pose deviation converges to a preset registration threshold to obtain a fine feature point pair association table. Based on the detailed feature point pair association table, each point cloud cluster in the lidar scanning data is mapped in real time to the three-dimensional digital twin base model to obtain holographic image data of the construction site.

4. The method for intelligent construction digital management according to claim 3, characterized in that, The holographic image data of the construction site is input into the finite element analysis engine, and dynamic simulation of the structural stress field is performed based on the material mechanics parameter library to obtain the time series data of structural stress distribution under construction loads, including: The point cloud and model hybrid data in the holographic image data are divided into meshes based on the geometric topology information of the three-dimensional digital twin base model to obtain an unstructured mesh. The unstructured mesh is hierarchically organized according to the octree spatial index to obtain a hybrid mesh model; The hybrid mesh model includes two types of hybrid mesh models with different granularities: coarse-grained hybrid mesh models are used for global stress field overview, and fine-grained hybrid mesh models are used for refined analysis of key parts.

5. The method for intelligent construction digital management according to claim 4, characterized in that, The holographic image data of the construction site is input into the finite element analysis engine, and dynamic simulation of the structural stress field is performed based on the material mechanics parameter library to obtain the time series data of structural stress distribution under construction loads. This also includes: Material properties associated with each component in the three-dimensional digital twin substrate model are extracted from the material mechanics parameter library, and the properties are automatically mounted according to the component ID and material ID mapping table in the three-dimensional digital twin substrate model. The load spatiotemporal distribution matrix is ​​determined based on the IoT sensor data of the construction machinery. The load spatiotemporal distribution matrix is ​​then applied to the finite element model in the form of equivalent nodal forces. The finite element model is solved step by step according to the construction cycle, and stress field data under the current state is generated at each time step. The stress field data at each time step are aggregated according to spatial location to obtain time-series data of structural stress distribution evolving over time.

6. The method for intelligent construction digital management according to claim 5, characterized in that, Based on the time-series data of the structural stress distribution, the hydration heat effect of concrete is simulated to obtain temperature and stress co-evolution characteristic data, including: The heat conduction calculation grid is determined based on the geometric topology information of the three-dimensional digital twin substrate model; a two-way thermo-mechanical coupling relationship is defined based on the heat conduction calculation grid and the time series data of structural stress distribution, wherein the two-way thermo-mechanical coupling relationship includes the driving effect of the temperature field on the thermal expansion of the material and the constraint effect of the structural stress field on the heat conduction path; The concrete hydration heat parameters are retrieved from the material mechanics parameter library. These parameters include the development curve between age and strength, as well as temperature and elastic modulus correction coefficients. The temperature field distribution is determined based on the heat conduction equation, and the structural stress field is updated based on the temperature field distribution according to the thermo-mechanical bidirectional coupling relationship.

7. The method for intelligent construction digital management according to claim 6, characterized in that, Based on the time-series data of the structural stress distribution, a simulation of the heat of hydration effect of concrete is performed to obtain temperature-stress co-evolution characteristic data, which also includes: The three-dimensional temperature field envelope data is determined based on the temperature field distribution, wherein the three-dimensional temperature field envelope data includes the core temperature, surface temperature and internal and external temperature difference time history curves of each component; The thermal stress distribution cloud map sequence is determined based on the three-dimensional temperature field envelope data and the updated structural stress field. Based on the thermal stress distribution cloud map sequence, temperature stress co-evolution characteristic data are determined, wherein the temperature stress co-evolution characteristic data includes the temporal correlation analysis of core temperature and first principal stress.

8. The method for intelligent construction digital management according to claim 7, characterized in that, Based on the long short-term memory network, structural deformation trend is predicted using the temperature stress co-evolution characteristic data to obtain a cluster of deformation prediction curves, including: Feature extraction is performed on the temperature stress co-evolution characteristic data to obtain a time-series feature matrix, which includes the core temperature time history curve of the concrete component, the thermal stress peak sequence, and the temperature stress coupling index. The geometric information and material properties of each component in the three-dimensional digital twin substrate model are used as static features; a multi-dimensional input feature set is determined based on the temporal feature matrix and the static features; Based on the Long Short-Term Memory Network, structural deformation trend is predicted through the multi-dimensional input feature set to obtain a deformation prediction curve cluster, wherein the deformation prediction curve cluster includes displacement time history curve, strain development curve and curvature radius change curve.

9. The method for intelligent construction digital management according to claim 8, characterized in that, Based on the fuzzy logic controller, a construction risk level assessment is performed using the deformation prediction curve cluster and a preset deformation threshold to obtain a structural safety risk classification and early warning, including: Obtain a preset deformation threshold library, wherein the preset deformation threshold library includes limit thresholds for displacement time history curves, strain development curves, and radius of curvature change curves; The deformation prediction curve cluster is defined as the input variable of the fuzzy logic controller. The membership function is determined according to the deformation prediction curve cluster and the preset deformation threshold library. The membership function is used to divide the displacement time history curve into three displacement levels and the strain development curve into three strain levels. A fuzzy rule base is determined based on the IF-THEN rule, and the input variables are mapped to membership functions to obtain a fuzzy set. The fuzzy set is then matched with the fuzzy rule base to obtain a structural safety risk classification and early warning.

10. A system for intelligent construction digital management, applicable to the method for intelligent construction digital management as described in any one of claims 1-9, characterized in that, include: The three-dimensional modeling unit (1) is used to construct a three-dimensional digital twin base model by integrating LiDAR scanning data, construction machinery IoT sensor data and design drawing parameters according to the building information modeling platform. Entity mapping unit (2), the entity mapping unit (2) is used to perform real-time mapping between construction entities and digital models through the three-dimensional digital twin base model according to the point cloud registration algorithm, so as to obtain holographic image data of the construction site; The stress analysis unit (3) is used to input the holographic image data of the construction site into the finite element analysis engine and perform dynamic simulation of the structural stress field according to the material mechanics parameter library to obtain the time series data of structural stress distribution under construction load. Co-evolution unit (4), the co-evolution unit (4) is used to simulate the hydration heat effect of concrete based on the time series data of the structural stress distribution, so as to obtain temperature stress co-evolution characteristic data; Trend prediction unit (5), the trend prediction unit (5) is used to predict the structural deformation trend based on the temperature stress co-evolution characteristic data through the long short-term memory network, so as to obtain a cluster of deformation prediction curves; The graded early warning unit (6) is used to assess the construction risk level based on the deformation prediction curve cluster and the preset deformation threshold by the fuzzy logic controller, so as to obtain a graded early warning of structural safety risk.

Images