Intelligent supervision method based on BIM technology

Abstract

The invention discloses an intelligent supervision method based on a BIM technology, and relates to the technical field of constructional engineering, and the method comprises the following steps: obtaining a BIM model corresponding to the constructional engineering, constructing a digital twinning mapping relation, and carrying out the construction of a digital twinning mapping relation based on a project type, construction complexity and historical supervision data; initializing a data entry time limit through an algorithm, and changing a synchronous threshold value and a quality parameter benchmark; and collecting real-time data, wherein the real-time data comprises process progress data, material entering data, design change information and environmental safety data of the construction site. According to the intelligent supervision method based on the BIM technology, digital and intelligent transformation of supervision work is realized through deep fusion of the BIM technology and multi-source data acquisition, the matching degree of data and an actual construction scene is improved, the accuracy of anomaly recognition is enhanced through a dynamic weight distribution strategy, and the accuracy of anomaly recognition is improved. The problem of dependence on fixed threshold judgment is effectively solved, and the manual intervention cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of building engineering technology, specifically to an intelligent supervision method based on BIM technology. Background Technology

[0002] As construction engineering transforms towards digitalization and intelligence, BIM technology, with its advantages of visualization, parametric analysis, and collaboration, has become a core supporting technology in the field of engineering supervision. Currently, engineering supervision work is gradually upgrading from traditional manual inspections and paper records to digital management, with multi-source data collection and real-time comparative analysis becoming key directions for improving supervision efficiency. The industry generally integrates data collection terminals such as visual sensors and RFID devices, combined with component and process information from BIM models, to achieve preliminary correlation of data such as construction progress and material quality. However, how to achieve deep integration of data and models, accurate identification of anomalies, and dynamic optimization of the supervision process remains a key area of ​​exploration for the industry, and also provides broad space for the development of intelligent supervision technology.

[0003] Existing BIM-based supervision technologies have reduced the cost of manual intervention to some extent, but there is still room for improvement in practical applications: some methods lack deep mapping between digital twins and BIM models, and their spatial alignment and cross-validation capabilities for multi-source data are insufficient, resulting in a need to improve the matching degree between data and actual construction scenarios; anomaly identification often relies on fixed threshold judgments, making it difficult to adapt to the construction complexity of different projects, and the analysis of the upstream and downstream impact chains of design changes is not comprehensive enough; supervision parameters are mostly statically set, unable to be dynamically adjusted based on historical data and real-time project status, affecting the accuracy and adaptability of supervision work. These problems limit the efficiency of closed-loop management of the supervision process and make it difficult to fully leverage the core value of digital technology. To address this, we propose an intelligent supervision method based on BIM technology. Summary of the Invention

[0004] To address the aforementioned technical issues, this paper proposes an intelligent supervision method based on BIM technology. This solution resolves the problems of insufficient deep mapping between digital twins and BIM, inadequate spatial alignment and cross-validation of multi-source data, and the need to improve the matching degree between data and construction scenarios. Furthermore, it addresses the issues of anomaly identification relying on fixed thresholds, incomplete analysis of the upstream and downstream impact chains of design changes, and statically set supervision parameters that cannot be adjusted based on historical data and real-time status, thus limiting the efficiency of closed-loop supervision management.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A smart supervision method based on BIM technology includes the following steps: Obtain the BIM model corresponding to the building project, construct the digital twin mapping relationship, calculate the weights based on the project type, construction complexity and historical supervision data, and initialize the data entry time limit, change synchronization threshold and quality parameter benchmark based on the analytic hierarchy process and entropy weight method; Collect real-time data, including: construction site process progress data, material arrival data, design change information, and environmental safety data, and associate them with the corresponding elements of the digital twin model; The collected real-time data is compared with the preset data bound in the BIM model to identify data entry delays, unsynchronized changes, parameter mismatches, and abnormal associations with safety hazards. Based on the abnormal impact weights determined by the analytic hierarchy process and the entropy weight method, early warning information of corresponding levels is generated, along with handling suggestions, and pushed to the relevant responsible persons; Receive the processing results and supplementary data from the responsible person, verify and virtually verify them, and synchronize them to the digital twin model after the verification and validation are passed, and update the relevant parameters. The initial monitoring parameters are dynamically optimized based on the data from each exception handling.

[0006] Preferably, the construction of the digital twin mapping relationship specifically includes binding the construction schedule, material standard parameters, process acceptance specifications, and the components, processes, and spatial coordinates of the BIM model.

[0007] Preferably, the real-time data is collected through a multi-source data acquisition module; The multi-source data acquisition module includes: a visual sensor array, an RFID reader / writer, an environmental sensor, a drone inspection unit, and a collaborative platform interface; The visual sensor array extracts construction status features, personnel and equipment distribution information, and process compliance identifiers through image recognition; the RFID reader / writer is used to read the model, quantity, arrival time, certificate number, and storage location data of the RFID tags of incoming materials; the environmental sensor is used to collect data on temperature, humidity, dust concentration, and noise at the construction site; the drone inspection unit plans inspection routes based on the construction schedule, collects data on high-altitude operation areas and large-span structure areas, and generates 3D point cloud data for comparison with the BIM model; the collaborative platform interface is used to connect with the digital management platforms of the client, designer, and contractor to capture design change documents, schedule adjustment notices, inspection reports, and instruction information.

[0008] Preferably, the data comparison and analysis process adopts an edge-central fusion architecture, specifically as follows: Edge computing nodes preprocess multi-source data, including: anomaly filtering, data format conversion, timestamp alignment, and cross-device data cross-validation. The abnormal data filtering adopts the Raida criterion, the data format conversion is converted to JSON format, the timestamp alignment is based on UTC time, and the cross-device data cross-validation is achieved by comparing data records of the same event on different devices. The preprocessed multi-source data is used to construct a three-dimensional comparison space by combining the weighted average method with the DS evidence theory. The multi-source data fusion algorithm integrates visual sensor data, RFID data, environmental sensor data and collaborative platform data. When comparing with the preset data in the BIM model, a dynamic weight allocation strategy is used to calculate the data deviation value, synchronization delay time and safety risk correlation. For process progress data, the time difference between actual construction nodes and planned nodes and the deviation between actual construction parameters and design parameters are compared. For material data, the consistency between the information on incoming materials and planned material information, the validity of the certificate of conformity and the compliance of the material storage environment are compared.

[0009] Preferably, the collection and synchronization of the design change information specifically includes: The design change documents are obtained through the collaborative platform interface. The change content, the components or process nodes involved, the new parameter standards, the schedule adjustment requirements are extracted, and the related impact of the change on upstream and downstream processes, material requirements, and cost budget are identified to generate a change impact analysis report. The extracted change information is compared with the original parameters of the corresponding process in the BIM model to determine the model elements and related data that need to be updated. If the model update is not completed within the preset change synchronization threshold time, it is judged as a change synchronization lag anomaly, and the anomaly priority is marked according to the impact chain analysis results.

[0010] Preferably, the data entry is delayed, and its recognition logic is as follows: Based on the construction schedule bound to the BIM model, combined with the construction market distribution of similar historical processes, an LSTM time series prediction model with historical process completion time, environmental factors and resource allocation data as training set is used to optimize network parameters through gradient descent method to generate dynamic completion time intervals for each process. The system monitors in real time whether the multi-source data acquisition module has collected process completion data and quality acceptance data within the data entry time limit of the corresponding dynamic completion time interval. If the relevant data is not collected after the time limit, multi-dimensional verification is performed through visual sensor images, personnel positioning data, and equipment operation records. Specifically: If multiple dimensions of data show that the process has been completed, it is determined that the data entry is delayed, and a supplementary entry template is pushed to the system. If the data shows that the process has not been completed, further analysis using algorithms will be conducted to determine whether there are objective delay factors and schedule adjustment notices. The objective delay factors include: weather factors, equipment failure factors, and material arrival delay factors. If there are no objective delay factors and no schedule adjustment notice, it is judged as an abnormal schedule lag; if there are objective delay factors or schedule adjustment notice, the dynamic completion time range and data entry time limit of the corresponding process in the model will be automatically updated.

[0011] Preferably, the classification of the early warning information is dynamically determined based on the weight of the abnormal impact and the real-time engineering status, specifically as follows: A Level 1 warning corresponds to key process data lagging by more than 8 hours, P0-level design changes not being synchronized, parameter mismatch affecting structural safety, and abnormal correlation of safety hazards. The warning information is pushed to the project manager, supervising engineer, construction manager, and safety administrator through multiple channels. At the same time, the progress update permissions and material entry approval permissions of subsequent related nodes of the process are frozen. Level 2 warning corresponds to a 2-8 hour lag in data for ordinary processes, failure to synchronize P1 level design changes, and mismatch of non-critical parameters. Warning information is pushed through mobile APP, work group message notifications, and BIM platform reminders. A Level 3 early warning is issued when there is a minor delay in data entry that does not affect subsequent procedures, or when a Level 2 design change is not synchronized. In such cases, a message is sent to remind the user on the BIM platform.

[0012] Preferably, the synchronization and update process for the supplementary data is as follows: After receiving the warning information through the human-machine collaborative terminal, the person in charge can supplement the data by choosing to enter it via voice, select a template, or upload an image of a paper record. After the supplementary data is submitted, the data integrity, consistency with preset standards, and correlation of multi-source collected data are verified. If the data is incomplete or does not meet the requirements, a supplementary data prompt will be returned.

[0013] Preferably, the method for binding the construction schedule, material standard parameters, process acceptance specifications, and components, processes, and spatial coordinates of the BIM model comprises the following steps: The elements in the BIM model are parsed and represented, and all components, construction process nodes and their spatial coordinate information are extracted from the model. A unique identifier is assigned to each element. Associating and mapping external data with model elements includes linking task nodes in the construction schedule with corresponding model components or processes using unique identifiers, matching specifications and model information in the material standard parameter library with the material properties of components in the model, and binding specific clauses of process acceptance specifications to corresponding process nodes as data rules. Establish a mapping relationship database and store the data of the established relationship mapping in the database.

[0014] Preferably, the method for constructing the three-dimensional comparison space comprises the following steps: A unified coordinate system is established based on the BIM model coordinate system, and all collected real-time data is converted to this unified spatial framework. Visual sensor data is matched with BIM components through feature point recognition to establish coordinate transformation from two-dimensional image to three-dimensional space. RFID data locates material information to specific component coordinates through the correspondence between physical tags and model tags. Environmental sensor data is bound to preset monitoring points to form a dynamic parameter field with spatial distribution. UAV point cloud data is geometrically aligned with the BIM model through registration algorithms. Collaborative platform documents are associated with the affected spatial areas through content parsing to complete the spatial mapping of multi-source data. A unified timestamp is added to all spatial mapping data to establish a four-dimensional spatiotemporal coordinate system. Based on this framework, the system automatically associates data from different sources according to the principle of spatial proximity, and finally generates a three-dimensional comparison space.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The intelligent supervision method based on BIM technology proposed in this invention achieves the digital and intelligent transformation of supervision work by deeply integrating BIM technology with multi-source data acquisition. It not only improves the matching degree between data and actual construction scenarios, but also enhances the accuracy of anomaly identification through dynamic weight allocation strategy, effectively solving the problem of relying on fixed threshold judgment in traditional methods. It can analyze the upstream and downstream impact chain of design changes, ensure the timely synchronization and accurate transmission of change information, and dynamically adjust supervision parameters based on historical data and real-time project status, significantly improving the adaptability and accuracy of supervision work, optimizing the closed-loop management of the supervision process, and reducing the cost of manual intervention. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0018] Reference Figure 1 As shown, an intelligent supervision method based on BIM technology includes the following steps: Obtain the BIM model corresponding to the building project. The model sources include the BIM model delivered by the design party during the design phase, the model generated by the digital transformation of existing projects, and the model exported from the third-party platform in compliance with regulations. The format must comply with IFC4.0, Revit family library standards or relevant specifications of GB / T 51235, and ensure that the model component information is complete and the coordinate system is consistent.

[0019] When constructing the digital twin mapping relationship, the elements in the BIM model are first comprehensively analyzed and standardized, and the attributes such as type, size, and material of all components, the logical relationship and sequence of construction process nodes, and the spatial coordinate information corresponding to each element are extracted. A unique global identifier is assigned to each component, process node, and spatial location. The identifier adopts the combination rule of project number + component type code + sequence number to ensure uniqueness.

[0020] External data is precisely mapped to model elements. Information such as start and end times, responsible persons, and resource allocation of each task node in the construction schedule is linked to the corresponding model components or process nodes through unique identifiers. Information such as specifications, performance indicators, and quality requirements in the material standard parameter library is precisely matched with the material properties of the components in the model. Specific clauses, testing methods, and qualification standards of the process acceptance specifications are bound to the corresponding process nodes as data rules.

[0021] A mapping relationship database is established, and a distributed storage architecture is adopted. The established associated mapping data is classified and stored according to component type, process stage, and data type. At the same time, data indexes are set to improve query efficiency, so as to realize the deep binding of BIM model with construction schedule, material standard parameters, and process acceptance specifications, forming the foundation of digital twin mapping.

[0022] Based on project type, construction complexity, and historical supervision data, the algorithm initializes data entry time limit, change synchronization threshold, and quality parameter benchmark. Real-time data is collected, including: construction site process progress data, material arrival data, design change information, and environmental safety data, and is associated with the corresponding elements of the digital twin model; the real-time data is collected through a multi-source data acquisition module, which includes: a visual sensor array, an RFID reader / writer, an environmental sensor, a drone inspection unit, and a collaborative platform interface.

[0023] The visual sensor array is arranged in a grid pattern according to the construction area. It collects continuous image data through high-definition cameras, uses convolutional neural networks for image recognition, and extracts construction status features, including the completion of component installation and the execution of construction processes. It identifies the number of personnel, equipment types and distribution locations through target detection algorithms, and generates a heat map of personnel and equipment distribution. It identifies compliance marks of procedures, such as hidden works acceptance marks and welding quality inspection marks, through template matching technology.

[0024] RFID readers are installed at the construction site entrance, material storage area, and component installation station. The reading distance is set to 3-5 meters. They read the model, quantity, arrival time, certificate number, and storage location data stored in the RFID tags of incoming materials in real time. The read data is uploaded to the data processing center in real time via a wireless transmission module. At the same time, the system automatically retries and provides alarm prompts for tag reading failures.

[0025] Environmental sensors are installed at various points according to the functions of the construction area. Monitoring points are set up in areas with high-intensity work, material storage areas, and high-altitude work areas. Temperature, humidity, dust concentration, and noise data are collected every 5 minutes. The equipment is automatically calibrated before data collection to ensure accuracy. Temperature and humidity measurement errors are controlled within ±0.5℃ and ±5%RH, respectively, and dust concentration detection accuracy reaches 0.1 mg / m³. 3 The noise measurement range covers 30-130dB.

[0026] The UAV inspection unit plans inspection routes based on key nodes and hazardous areas in the construction schedule. It uses GPS + inertial navigation combined positioning technology, sets the flight altitude to 10-20 meters above the highest point of the construction area, and adjusts the flight speed to 3-8 m / s according to the complexity of the operation. It performs all-round shooting and laser scanning of high-altitude operation areas and large-span structure areas to generate high-density three-dimensional point cloud data with a point cloud density of not less than 50 points / square meter, which is used for geometric accuracy comparison with the BIM model.

[0027] The collaborative platform interface adopts standardized connection protocols such as API and WebService to connect with the digital management platforms of the client, designer and construction party. It automatically captures design change documents, progress adjustment notices, test reports and instruction information at a frequency of once per hour. It supports the parsing of multiple file formats such as PDF, CAD and Excel, and performs format standardization processing after extracting key information. The collected real-time data is compared with the preset data bound in the BIM model to identify data entry delays, unsynchronized changes, parameter mismatches, and abnormal associations of safety hazards; the data comparison and analysis process adopts an edge-central fusion architecture.

[0028] Edge computing nodes are deployed on local servers at the construction site to preprocess multi-source data. Abnormal data filtering uses the Laida criterion to remove data exceeding three times the standard deviation, while mean filtering smooths noisy data. Data format conversion unifies heterogeneous data output from different devices into JSON format to ensure data structure consistency. Timestamp alignment uses UTC time as the benchmark, calibrating the timestamps of data collected by each device, with errors controlled within one second. Cross-device data cross-validation compares data records of the same event on different devices; for example, material arrival data is verified by simultaneously verifying RFID reader / writer data and visual sensor images of the arrival process, ensuring data authenticity.

[0029] The preprocessed multi-source data is integrated using a multi-source data fusion algorithm that combines the weighted average method with DS evidence theory. This algorithm integrates visual sensor data, RFID data, environmental sensor data, and collaborative platform data to construct a three-dimensional comparison space.

[0030] When comparing with the preset data in the BIM model, a dynamic weight allocation strategy is adopted. The weight values ​​are determined based on factors such as process criticality, data reliability, and scope of influence. The weights of each indicator are calculated using the analytic hierarchy process, thereby obtaining the data deviation value, synchronization delay time, and safety risk correlation.

[0031] For process progress data, compare the time difference between actual construction nodes and planned nodes to calculate the progress deviation rate. At the same time, compare the absolute and relative deviations between actual construction parameters and design parameters, such as the coordinate deviation of component installation positions and the deviation of rebar tying spacing. For material data, compare the consistency of the model, specifications, and quantity of incoming materials with the planned material information, verify the validity of the certificate of conformity number through online verification, and compare the compliance of the temperature and humidity of the material storage environment, protective measures, and preset storage standards. Based on the weight of abnormal impact, a warning message of the corresponding level is generated, with handling suggestions attached, and pushed to the relevant responsible persons. The weight of abnormal impact is determined by the analytic hierarchy process to determine the evaluation index system, which includes four primary indicators: process criticality, scope of impact, severity, and urgency. Each primary indicator has 3-4 secondary indicators, such as the scope of impact including the number of components affected, the number of processes associated, and the amount of cost impact. The subjective weight is corrected by the entropy weight method, and the comprehensive abnormal impact weight is finally calculated.

[0032] The warning level is dynamically determined by combining the weight of the abnormal impact with the real-time project status. The real-time project status includes the current construction progress, resource allocation, number of existing warnings, and processing progress.

[0033] A Level 1 warning corresponds to key process data lag exceeding 8 hours, unsynchronized P0-level design changes, mismatched parameters affecting structural safety, and abnormal associations of safety hazards. The warning information is pushed to the project manager, supervising engineer, construction manager, and safety administrator through the following methods: a BIM platform pop-up that is forced to stay for 10 seconds, an instant SMS message, an automatic voice call made 3 times, a targeted @mention of all relevant personnel in the work group, and a forced pop-up reminder in the APP. At the same time, the progress update permissions of subsequent related nodes of the process are frozen, new progress reports are prohibited, and the material entry approval process is suspended until the anomaly is handled.

[0034] Level 2 warnings correspond to data lag of 2-8 hours in ordinary processes, unsynchronized P1 level design changes, and mismatched non-critical parameters. Warning information is pushed via mobile APP with vibration alerts, pinned messages in work groups, and floating reminders on the right side of the BIM platform. Relevant responsible persons must respond within 4 hours.

[0035] A Level 3 warning is issued in the BIM platform message center for minor data delays that do not affect subsequent procedures and for P2 level design changes that are not synchronized. The person in charge can handle the issue during work breaks. The system receives processing results and supplementary data from the responsible party, verifies and verifies them against both real and virtual data. Once verification is successful, the data is synchronized to the digital twin model, and relevant parameters are updated. The responsible party receives warning information through human-computer collaborative terminals such as mobile apps, tablets, or PCs, and selects the appropriate supplementary data method based on the data type. Voice input supports dialect recognition and semantic error correction, automatically converting the data into text. Template selection provides standardized forms with required and optional fields, guiding the responsible party to complete the form quickly. When uploading images of paper records, the system automatically extracts text information using OCR recognition technology and matches it with the form fields to fill in the information.

[0036] After the supplementary data is submitted, the first step is to check the data integrity, verifying whether any required fields are missing and whether all attachments are complete. Then, a standard consistency verification is performed, comparing the supplementary data with the preset standards and acceptance specifications bound to the BIM model to confirm that the data is within the allowable range. Finally, a multi-source data correlation verification is performed, comparing the supplementary data with relevant data collected by visual sensors, RFID devices, environmental sensors, etc., to see if they are consistent, such as whether the quantity of materials used in the supplementary data matches the quantity read by RFID upon entry.

[0037] Virtual-real verification is conducted through a BIM model simulation platform. The supplemented data and processing results are substituted into the model to simulate the construction process and results, and to verify the effectiveness of the treatment measures. For example, after parameter adjustment, whether the simulated stress of the component meets the safety standards, and whether the progress correction is smoothly connected with the overall construction plan. The initial monitoring parameters are dynamically optimized based on the data from each exception handling process. The construction of the digital twin mapping relationship specifically includes binding the construction schedule, material standard parameters, process acceptance specifications, and the components, processes, and spatial coordinates of the BIM model; The collection and synchronization of design change information specifically involves: obtaining design change documents through the collaborative platform interface, extracting the change content, involved components or process nodes, new parameter standards, and schedule adjustment requirements using natural language processing technology, identifying the impact of the change on upstream and downstream processes, the direction of material demand adjustment, and the magnitude of cost budget changes through correlation analysis algorithms, and generating a change impact analysis report that includes the scope of impact, severity, and response suggestions.

[0038] The extracted change information is compared item by item with the original parameters of the corresponding process in the BIM model to determine the model elements and related data that need to be updated, and to clarify the update content and priority.

[0039] If the model update is not completed within the preset change synchronization threshold time, it is judged as a change synchronization anomaly. The anomaly priority is marked according to the impact chain analysis results. Changes that affect the critical path and involve significant cost changes are marked as high priority, while changes that only affect local details and have no cost changes are marked as low priority. The data entry lag is identified by the following logic: based on the construction schedule plan bound to the BIM model, combined with the historical construction duration distribution of similar processes, environmental influencing factors, and resource allocation data, an LSTM time series prediction model is constructed.

[0040] The model training uses historical supervision data such as process completion time, construction conditions, and resource input as the training set. After data normalization and sequence partitioning, an LSTM network containing input, hidden, and output layers is built. The network parameters are optimized by gradient descent and iteratively trained until the loss function converges.

[0041] The trained model is used to generate dynamic completion time intervals for each process. The upper and lower limits of the intervals are determined based on the predicted values ​​plus or minus a 30% confidence interval.

[0042] The system monitors in real time whether the multi-source data acquisition module has collected process completion data and quality acceptance data within the data entry time limit of the corresponding dynamic completion time interval. If the relevant data is not collected within the time limit, the system analyzes the installation status of components in the construction area and whether the process identification is complete through visual sensor image analysis. It also checks the personnel positioning data to confirm whether the workers have left the area and retrieves the equipment operation records to verify whether the construction equipment has stopped operating, thus performing multi-dimensional verification.

[0043] If multiple dimensions of data show that the process has been completed, it is determined that the data entry is delayed, and a standardized supplementary entry template is pushed. If the data shows that the process has not been completed, further analysis is conducted using algorithms to determine whether there are objective delay factors and schedule adjustment notices. The objective delay factors include: weather factors, equipment failure factors, and material arrival delay factors. Weather factors are obtained by connecting to the meteorological platform to obtain records of severe weather such as rainfall and strong winds. Equipment failure factors are checked by verifying equipment maintenance records and alarm logs. Material arrival delay factors are compared with RFID arrival data and planned arrival time. If there are no objective delay factors and no schedule adjustment notices, it is determined that the schedule is abnormally delayed. If there are objective delay factors or schedule adjustment notices, the dynamic completion time interval and data entry time limit of the corresponding process in the model are automatically updated. The method for binding the construction schedule, material standard parameters, process acceptance specifications, and components, processes, and spatial coordinates of the BIM model consists of the following steps: parsing and representing the elements in the BIM model; using a BIM model parsing engine to extract all components, construction process nodes, and their spatial coordinate information from the model; component information including type, size, material, connection method, etc.; process node information including process name, process requirements, dependencies, etc.; and spatial coordinates represented using a three-dimensional rectangular coordinate system.

[0044] Each element is assigned a unique identifier, which consists of 16 characters. The first 4 characters are the project code, the middle 6 characters are the component or process type code, and the last 6 characters are the sequence number.

[0045] The process involves mapping external data to model elements, including linking task nodes in the construction schedule with corresponding model components or processes using unique identifiers to establish a correspondence between tasks and entities; matching specifications, models, performance indicators, and other information from the material standard parameter library with the material properties of components in the model to ensure that material parameters are consistent with component requirements; and binding specific clauses, testing standards, and qualification thresholds of process acceptance specifications to corresponding process nodes as data rules to provide a basis for subsequent acceptance.

[0046] Establish a mapping database, use a relational database to store the associated mapping data, design data tables to include fields such as element identifier, external data type, data content, association time, and update log, set primary keys and foreign keys to ensure data integrity, and configure a data backup mechanism to prevent data loss; The method for constructing a three-dimensional comparison space involves the following steps: using the national geodetic coordinate system as a reference and combining it with the design coordinate system of the BIM model, a unified coordinate system is established through a seven-parameter transformation method (translation, rotation, and scaling parameters). All collected real-time data is then transformed into this unified spatial framework to ensure spatial coordinate consistency.

[0047] Visual sensor data is used to extract feature points such as corners and edges from images through feature point detection algorithms. These feature points are then matched with the feature points of corresponding components in the BIM model to establish a coordinate transformation relationship from two-dimensional image to three-dimensional space, thereby achieving spatial positioning of image data.

[0048] RFID data is associated with the unique identifier of the model tag through the spatial coordinate information built into the physical tag, so as to locate the material's model, quantity and other information to the coordinates of the specific component, forming a spatial correspondence between the material and the component.

[0049] Environmental sensor data is bound to preset monitoring points. Each monitoring point corresponds to specific spatial coordinates in the model. The real-time collected data is stored according to the coordinate position, forming a dynamic parameter field with spatial distribution, which intuitively presents the spatial changes of environmental indicators.

[0050] The UAV point cloud data is geometrically aligned with the BIM model using the ICP (Iterative Closest Point) algorithm. The distance deviation between the point cloud data and the model surface is calculated, and the point cloud coordinates are corrected to ensure geometric consistency with the BIM model.

[0051] The collaborative platform documents use natural language processing technology to parse the text content, identify the components, processes and spatial ranges involved, and associate them with the affected spatial areas in the model to complete the spatial mapping of multi-source data.

[0052] A unified timestamp with millisecond precision is added to all spatial mapping data. A four-dimensional spatiotemporal coordinate system is established based on three-dimensional spatial coordinates and timestamps. The system automatically associates relevant data from different sources according to the principle of spatial proximity (setting a 5-meter distance threshold) and integrates them to form a three-dimensional comparison space containing spatial information, temporal information, and attribute information, providing a unified analysis framework for subsequent data comparison.

[0053] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A BIM technology-based intelligent supervision method, characterized in that, The method comprises the following steps: Obtaining the BIM model corresponding to the construction project, constructing the digital twin mapping relationship, calculating the weight based on the analytic hierarchy process and entropy weight method based on the project type, construction complexity and historical supervision data, initializing the data entry time limit, change synchronization threshold and quality parameter benchmark; Collecting real-time data, including process progress data, material arrival data, design change information and environmental safety data, and associating them with the elements corresponding to the digital twin model; Comparing the collected real-time data with the preset data bound in the BIM model to identify data entry lag, change synchronization, parameter mismatch and safety hazard correlation anomalies; Based on the abnormal influence weight determined by the analytic hierarchy process and entropy weight method, generate corresponding level warning information, with processing suggestions and push to the relevant person in charge; Receive the processing results and supplementary data feedback by the person in charge, and perform verification and virtual verification, and synchronize to the digital twin model after verification and verification, and update the relevant parameters; Based on each abnormal processing data, dynamically optimize the initial set supervision parameters. 2.The intelligent supervision method based on BIM technology of claim 1, wherein, The construction of the digital twin mapping relationship specifically includes binding the construction progress plan, material standard parameter, process acceptance specification and the components, processes and spatial coordinates of the BIM model. 3.The intelligent supervision method based on BIM technology of claim 1, wherein, The real-time data is collected through a multi-source data collection module; The multi-source data collection module includes a visual sensor array, an RFID read-write device, an environmental sensor, an unmanned aerial vehicle inspection unit and a collaborative platform interface; The visual sensor array extracts construction state characteristics, personnel and equipment distribution information and process compliance identification through image recognition; the RFID read-write device is used to read the model, quantity, arrival time, qualification certificate number and storage location data of the RFID tag of the incoming material; the environmental sensor is used to collect construction site temperature and humidity, dust concentration and noise data; the unmanned aerial vehicle inspection unit plans the inspection path based on the construction progress plan, collects the high-altitude work area and large-span structure area, generates three-dimensional point cloud data and compares it with the BIM model; the collaborative platform interface is used to interface the digital management platforms of the owner, designer and constructor, and to grab design change files, progress adjustment notifications, test reports and instruction information.

4. The intelligent supervision method based on BIM technology according to claim 1, characterized in that, The data comparison uses an edge-central fusion architecture in the comparison and analysis process, specifically: The edge computing node pre-processes the multi-source data, including abnormal data filtering, data format conversion, timestamp alignment and cross-device data cross-validation; The abnormal data filtering uses the Raydah criterion, the data format conversion is converted to JSON format, the timestamp alignment is based on UTC time, and the cross-device data cross-validation is by comparing the data records of the same event in different devices; The pre-processed multi-source data is based on a multi-source data fusion algorithm, which combines weighted average method and D-S evidence theory to integrate visual sensor data, RFID data, environmental sensor data and collaborative platform data, and construct a three-dimensional comparison space; When compared with preset data in the BIM model, a dynamic weight distribution strategy is adopted to calculate data deviation values, synchronization delay time lengths and safety risk correlation degrees; for process progress data, the time difference between actual construction nodes and planned nodes and the deviation between actual construction parameters and design parameters are compared; for material data, the consistency of incoming material information and planned material information, the validity of qualification certificates and the compliance of material storage environments are compared.

5. The intelligent supervision method based on BIM technology according to claim 1, characterized in that, The collection and synchronization of the design change information specifically comprises: design change files are obtained through a collaborative platform interface, change content, involved components or process nodes, new parameter standards and schedule adjustment requirements are extracted, and the associated influence of the change on upstream and downstream processes, material requirements and cost budgets is identified to generate a change impact analysis report; the extracted change information is compared with original parameters of corresponding processes in the BIM model to determine model elements and associated data that need to be updated; if model updating is not completed within a preset change synchronization threshold time, it is determined that there is a change synchronization lag anomaly, and the anomaly priority is marked according to the influence chain analysis result.

6. The intelligent supervision method based on BIM technology according to claim 1, characterized in that, The identification logic of the data entry lag specifically comprises: based on the construction schedule plan bound to the BIM model, in combination with the construction market distribution of historical similar processes, a LSTM time series prediction model is used with historical process completion time, environmental factors and resource allocation data as a training set, network parameters are optimized through gradient descent method, and dynamic completion time intervals of each process are generated; whether the multi-source data acquisition module acquires process completion data and quality acceptance data within the data entry time limit of the corresponding dynamic completion time interval is monitored in real time; if relevant data is not acquired within the time limit, multi-dimensional verification is performed through visual sensor images, personnel positioning data and equipment operation records, specifically: if multi-dimensional data all show that the process has been constructed, it is determined that there is a data entry lag, and a supplementary recording template is pushed; if the data show that the process has not been constructed, further algorithm analysis is performed to determine whether there are objective delay factors and progress adjustment notifications, the objective delay factors including weather factors, equipment failure factors and material arrival lag factors; if there are no objective delay factors and no progress adjustment notifications, it is determined that there is a progress lag anomaly; if there are objective delay factors or progress adjustment notifications, the dynamic completion time interval and the data entry time limit of the corresponding process in the model are automatically updated.

7. The intelligent supervision method based on BIM technology according to claim 1, characterized in that, The grade classification of the early warning information is dynamically determined based on abnormal influence weights and real-time engineering state, specifically: a first-level early warning corresponds to a key process data lag of more than 8 hours, a P0-level design change that is not synchronized, a parameter mismatch that affects structural safety and a safety hazard correlation anomaly, early warning information is pushed to the project manager, the supervising engineer, the construction manager and the safety administrator through a multi-channel forced pushing mode, and the progress update authority and the material arrival approval authority of subsequent associated nodes of the process are frozen; a second-level early warning corresponds to a common process data lag of 2-8 hours, a P1-level design change that is not synchronized and a non-key parameter mismatch, early warning information is pushed through a mobile phone APP, a work group message notification and a BIM platform reminder. The third level of early warning corresponds to slight data delay entry and does not affect subsequent processes, and the P2 level of design change is not synchronized, and a message is sent on the BIM platform to remind. 8.The intelligent supervision method based on BIM technology of claim 1, wherein, The synchronization and updating process of the supplementary recording data is specifically: The person in charge receives the early warning information through the man-machine collaborative terminal, and selects the voice recording, template selection, and uploading of paper record images to supplement the data; After the supplementary recording data is submitted, the data integrity, consistency of preset standards, and correlation of multi-source collected data are checked, and if the data is incomplete or does not meet the requirements, the supplementary recording prompt is returned. 9.The intelligent supervision method based on BIM technology of claim 2, wherein, The method steps of binding the construction progress plan, material standard parameter, process acceptance specification, and component, process, and space coordinates of the BIM model are: The elements in the BIM model are analyzed and represented, all components, construction process nodes, and their space coordinate information in the model are extracted, and each element is assigned a unique identifier; The external data is associated and mapped with the model elements, including linking the task nodes in the construction progress plan with the corresponding model components or processes through unique identifiers, matching the specifications and model of the material standard parameter library with the material properties of the components in the model, and binding the specific terms of the process acceptance specification to the corresponding process nodes as data rules; A mapping relationship database is established, and the data with established association and mapping is stored in the library.

10. The intelligent supervision method based on BIM technology according to claim 4, characterized in that, The method steps of constructing a three-dimensional comparison space are: A unified coordinate system is established based on the BIM model coordinate system, and all collected real-time data is converted to this unified space framework; The visual sensor data is matched with the BIM components through feature point recognition, the coordinate conversion from two-dimensional image to three-dimensional space is established, the RFID data is positioned to the specific component coordinate through the corresponding relationship between the physical tag and the model tag, the environmental sensor data is bound to the preset monitoring point, forming a dynamically parameter field with spatial distribution, the unmanned aerial vehicle point cloud data is geometrically aligned with the BIM model through registration algorithm, and the content of the collaborative platform document is associated to the affected space area through content analysis, completing the spatial mapping of multi-source data; A unified timestamp is added to all spatial mapping data, a four-dimensional space-time coordinate system is established, based on this framework, the system automatically associates different source data according to the spatial proximity principle, and finally generates a three-dimensional comparison space.

Images