BIM-based civil construction whole-process progress management and control system and method

Abstract

The invention provides a BIM-based civil engineering construction whole process progress management and control system and method, and relates to the technical field of progress data management, and the method comprises the steps: obtaining a civil engineering information model BIM, building a construction three-dimensional information model, importing a construction progress plan, and generating 4D progress simulation, the construction three-dimensional information model is converted into a construction unit set defined by a floor-partition-axis net, each construction unit has a unique construction unit identifier and establishes a corresponding spatial index relationship, construction process site progress data is collected and converted into site evidence items, a coverage construction unit is determined based on the site evidence items, and the coverage construction unit is used for covering the construction unit. According to the method, actual state slices are formed in coverage construction units, corresponding planned state slices are inquired according to identifiers corresponding to the coverage construction units, the actual state slices and the planned state slices are aligned in a unified data structure, the risk that model presentation in the whole construction process deviates from site reality is reduced, and the construction efficiency is improved. And the accuracy of progress statistics and key node judgment is improved.

Description

Technical Field

[0001] This invention relates to the field of progress data management technology, and in particular to a BIM-based system and method for controlling the progress of civil construction throughout the entire process. Background Technology

[0002] Schedule management during the civil construction phase is typically based on the construction organization design, construction schedule, and network plan. It involves establishing plans around floors, zones, grid lines, and various sub-items and work processes, and continuously tracking actual progress during construction to support management activities such as schedule statistics, deviation verification, and resource allocation. With the increasing application of Building Information Modeling (BIM) technology in the construction phase, engineering management practices are gradually establishing a link between 3D information models and schedule plans. This creates a correspondence between construction sequence, stage nodes, and model components, enabling visual representation of the construction process within the model environment and aiding in schedule coordination and process control.

[0003] Under this management model, the acquisition of actual on-site progress information still relies primarily on personnel collection. Image data is typically collected by construction workers, quality management personnel, or supervisors during daily inspections, process acceptance, and node checks. Specifically, relevant personnel enter the site according to predetermined inspection routes and frequencies, taking photos or short videos of the work area at designated or agreed-upon locations. These photos or videos are then uploaded to the project management platform after filling in information such as floor, zone, grid range, process name, and inspection conclusions. This serves as a basis for process documentation, quality acceptance, and progress assessment. For construction areas with a large coverage area, there are also methods that use drone aerial photography or fixed camera equipment to periodically acquire on-site image data, which is then combined with model and planning information to conduct progress verification and process analysis.

[0004] For example, Chinese invention patent CN120297918B discloses an intelligent management method and system for civil engineering projects, which includes the following steps: acquiring personnel distribution, equipment status, and material inventory data; extracting key task nodes and their execution status; identifying tasks lagging behind schedule; screening critical path nodes and analyzing time differences; marking progress trends; identifying equipment-affected sections; judging resource allocation imbalances; and generating progress monitoring indicators. Through multi-source data collection of personnel distribution, equipment status, and material inventory, it enhances the identification of differences between task status and planned nodes, improves the efficiency of progress delay early warning, achieves dynamic classification and trend identification of progress fluctuations, assists in optimizing construction rhythm, identifies construction efficiency bottlenecks through overlapping analysis of equipment operation and task fluctuation areas, improves scheduling accuracy, and further achieves refined progress monitoring through comparison of input and cycle differences, comprehensively enhancing progress control capabilities and the synergy of on-site resource allocation.

[0005] For example, Chinese invention patent CN119721997B discloses a BIM-based method and system for managing construction progress. The method includes: acquiring the actual 3D model and BIM simulation model corresponding to the construction task; matching the actual 3D model with the BIM simulation model to determine the virtual construction node corresponding to the current construction node in the BIM simulation model; determining the actual construction time and the estimated total construction time of the construction task corresponding to the current construction node, and determining the difficulty of each construction stage in the construction task; determining the virtual construction time corresponding to the virtual construction node; determining the construction progress status of the current construction stage based on the actual construction time, the estimated total construction time, and the virtual construction time; and managing the construction progress of the construction task according to the construction progress status of the current construction stage.

[0006] The above-mentioned technology has the following technical problems:

[0007] In the overall progress control of civil construction, the acquisition of actual on-site progress information is usually carried out on a work surface-by-work basis. Construction workers, quality management personnel, or supervisors enter the site during daily inspections, process acceptance, and node checks to photograph the construction status of formwork erection, rebar tying, concrete pouring, and masonry and plastering, creating photos or short videos. When uploading these videos, information such as floor, zone, grid range, construction section, process name, and inspection conclusions are also recorded as process documentation and progress verification. Since the photographed content is mostly of local construction status, and the information is mainly labeled by floor, zone, and construction section, this type of video data can usually reflect the overall progress within a certain work area, but it is difficult to naturally correspond to individual component objects.

[0008] In contrast, building information modeling (BIM)-based schedule management often uses model components as the basic objects for schedule association. It establishes correspondences between construction tasks, stage nodes, and components such as beams, slabs, columns, walls, and openings to display and update the schedule status within the model. Within the same floor and zone, it's common for structural components to be arranged in a regular, repetitive pattern along the grid. Multiple beams with the same cross-section, structural columns of the same specification, or openings of the same size exhibit high consistency in geometry, structural attributes, and construction procedures. In this case, the on-site imagery and its accompanying floor, zone, and procedure annotations have a clear regional and aggregate nature, limiting the actual construction status to a specific operational area and making it difficult to provide a unique and stable correspondence for any particular component.

[0009] Under the aforementioned conditions, if the existing system directly uses image data and work reporting information from the work surface level for progress write-back and node matching at the component level, inconsistencies may easily arise between the component status records and the actual construction objects. For example, progress information within the same work surface may be scattered across multiple similar components or updated to incomplete component objects, causing the component completion status presented in the model to deviate from the actual site conditions. This deviation further leads to the reliance on unstable component states for plan comparison, deviation identification, and critical node judgment, resulting in problems such as distorted progress statistics, inaccurate deviation location, and unreliable early warning triggering. This makes it difficult to meet the accuracy, stability, and traceability requirements of progress control throughout the entire civil construction process. Summary of the Invention

[0010] Therefore, embodiments of the present invention provide a BIM-based system and method for controlling the progress of civil construction throughout the entire process, which can meet the requirements of accuracy, stability and traceability for controlling the progress of civil construction throughout the entire process.

[0011] The technical solution of this invention is implemented as follows:

[0012] This invention provides a BIM-based system for managing the entire construction progress of civil engineering projects. The system includes: an information conversion module, used to acquire a BIM model of the civil engineering information model and establish a 3D construction information model, import the construction schedule plan to generate a 4D progress simulation, and convert the 3D construction information model into a set of construction units defined by floors, zones, and grids. Each construction unit has a unique identifier and a corresponding spatial index relationship is established. A slice alignment module is used to collect on-site progress data and convert it into on-site evidence entries. Based on these on-site evidence entries, it determines the covering construction units, forms actual state slices within the covering construction units, and queries the corresponding planned state slices using the identifier of the covering construction unit. The actual state slices and planned state slices are aligned using a unified data structure. A progress management module is used to output management results for progress comparison, early warning, and scheduling based on the aligned actual state slices and planned state slices. When it is necessary to present the actual state slices at the component level, it performs write-back control on the component status of the actual state slices, ensuring that components that do not meet the write-back conditions remain confirmed at the construction unit level without undergoing component-level write-back.

[0013] This application also provides a BIM-based method for full-process progress control of civil engineering construction. The method includes: Step 1: Acquiring a civil engineering information model (BIM) and establishing a three-dimensional construction information model, importing the construction schedule plan to generate a 4D progress simulation, and converting the three-dimensional construction information model into a set of construction units defined by floors, zones, and grids. Each construction unit has a unique construction unit identifier and a corresponding spatial index relationship is established. Step 2: Collecting on-site progress data during the construction process and converting it into on-site evidence items. Based on the on-site evidence items, determining the covering construction units, forming actual state slices within the covering construction units, and querying the corresponding planned state slices using the identifiers corresponding to the covering construction units. Aligning the actual state slices and planned state slices with a unified data structure. Step 3: Based on the aligned actual state slices and planned state slices, outputting control results for progress comparison, early warning, and scheduling. When it is necessary to present the actual state slices at the component level, write-back control is performed on the component status of the actual state slices, so that components that do not meet the write-back conditions remain in the construction unit level for confirmation and are not written back at the component level.

[0014] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0015] (1) Obtain the civil engineering information model (BIM) to establish a three-dimensional construction information model, and import the construction schedule plan to generate a four-dimensional progress simulation. The three-dimensional construction information model covering the construction unit is converted into a set of construction units defined by floors, zones and grids. Each construction unit is given a unique construction unit identifier and a spatial index relationship is established. This abstracts the progress expression on the planning side from the component granularity to the work space granularity, avoiding the unstable mapping caused by the inherent regionality of on-site information and the difficulty in uniquely corresponding to a single component. Collect on-site progress data of the construction process and convert it into on-site evidence items containing time, procedures, location and image supporting materials. Based on the on-site evidence items, determine the covering construction unit, form the actual state slice within the covering construction unit, and query the corresponding planned state slice with the covering construction unit identifier. This aligns the actual state slice and the planned state slice with a unified data structure, enabling the plan and the actual to be comparable under the same work space and the same procedure dimension, thereby supporting progress comparison, deviation location and early warning and scheduling output. To address the error-prone issue of component rewriting caused by the coexistence of duplicate components and regional evidence, when component-level presentation is required, rewriting control is implemented on the component status derived from the actual state slice. Components that do not meet the rewriting conditions are kept in the construction unit level for confirmation without being rewritten at the component level, and are jointly recorded with the evidence chain and control results. This reduces the risk of deviation between the model presentation and the actual site throughout the construction process, and improves the accuracy, stability and traceability of progress statistics and key node judgment.

[0016] (2) After mapping the on-site evidence items to actual state slices within the construction unit, this scheme uses the construction unit identifier as an index to synchronously query the planned state slices. It also uses the construction unit and process as a unified data structure to achieve alignment between the plan and the actual situation, so that the on-site images and work reports can be compared at the granularity of the work space first, avoiding the mismatch risk introduced by directly writing regional evidence back to the component granularity. Furthermore, the coverage area of ​​the actual state slice is adaptively corrected by adjusting the mapping domain: when the actual coverage area is inconsistent with the corresponding range of the plan, it is shrunk to the minimum connected coverage area, which can reduce the coverage spillover caused by the locality of the shooting field or the coarse granularity of the annotation, so that the slice only acts on the work range that the evidence can support; when continuous construction events drive the synchronous advancement of adjacent areas, it is expanded along the advancement direction and marked as pending confirmation, which can reserve a buffer for potential advancement areas without interrupting progress tracking, reducing the omission caused by excessive shrinkage; when the coverage area span is too large, it is split into sub-coverage areas according to the axis grid direction and forms slices respectively, which can decompose the collective information of densely repeated component areas into finer-grained alignment units, thereby improving the spatial resolution of deviation positioning. After the mapping domain adjustment is completed, process consistency adjustment is performed. Conflicting states that violate the process state machine are set to pending review and limited to the construction unit level. Simultaneously, a targeted review scope or supplementary sampling requirement is generated, which can block the transmission effect of unreasonable process sequences on subsequent plan comparisons and early warning triggers. Thus, under conditions where duplicate components are common and evidence is difficult to uniquely point to a single component, the probability of component states deviating from the actual site conditions can be significantly reduced, improving the accuracy, stability, and traceability of progress statistics, deviation identification, and early warning scheduling.

[0017] (3) This invention addresses the problem of distorted progress write-back caused by the difficulty in stably corresponding individual components when collecting on-site images and work reports on a work surface basis. By forming actual state slices at the construction unit level and introducing a component disambiguation and write-back control mechanism when component-level presentation is required, the component state update is based on constrained and convergent derivation results. Specifically, the key construction units are first determined by combining the actual state slices and the planned state slices to avoid blind write-back across all components. Then, candidate components are extracted within the key construction units, and components with the same geometric shape, structural attributes, and process characteristics are clustered to form duplicate component groups. The one-to-many ambiguity caused by repeated arrangement is converged to the solution boundary within the group. At the same time, constraints such as continuous spatial advancement, consistency of duplicate groups, consistency of adjacency, process state machine, and event boundary are integrated to suppress the risk of evidence in the work surface area being scattered and written into multiple similar components or mistakenly written into incomplete components. For components that cannot meet the constraints, write-back is not forced, but the evidence chain index is retained at the construction unit level to reduce the deviation in plan comparison, misjudgment of key nodes, and early warning drift caused by the inconsistency between the completed state of the model components and the actual situation. Furthermore, when the number of components that cannot be uniquely determined reaches a threshold, structural adjustments are triggered. By subdividing construction units along the direction of advancement, merging areas with overly fragmented evidence, reconstructing according to the form of construction sections, and issuing minimum replenishment instructions, the symmetry is actively broken and disambiguation convergence is promoted. This improves the stability and traceability of component-level progress presentation and ensures the reliability of progress statistics and scheduling decisions. Attached Figure Description

[0018] Figure 1 This is a structural diagram of the BIM-based civil construction progress control system provided in this embodiment of the invention;

[0019] Figure 2 This is a schematic diagram of the process of BIM-based civil construction progress control provided in the embodiments of the present invention;

[0020] Figure 3 This is a flowchart of the construction of a BIM-based 3D model and the generation of progress slices provided in an embodiment of the present invention.

[0021] Figure 4 This is a flowchart of the progress control result processing and the formation of a closed loop throughout the entire process provided by an embodiment of the present invention;

[0022] Figure 5 This is a flowchart of the component disambiguation optimization and structural adjustment loop provided in the embodiments of the present invention;

[0023] Figure 6 This is the overall view of the construction scene for four-dimensional progress simulation provided in the embodiments of the present invention. Figure 1 ;

[0024] Figure 7This is the overall view of the construction scene for four-dimensional progress simulation provided in the embodiments of the present invention. Figure 2 ;

[0025] Figure 8 This is the overall view of the construction scene for four-dimensional progress simulation provided in the embodiments of the present invention. Figure 3 . Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0027] In actual civil construction, the collection of construction progress information is usually carried out on a work surface basis, such as floor, zone, or construction section. Construction workers, quality inspectors, or supervisors often record the status of formwork, reinforcement, pouring, masonry, and plastering processes by taking photos or short videos during inspections, acceptance, and node verification. When uploading, they also fill in the process name, floor zone, and grid range, etc., for record keeping and progress verification. However, on-site images often present a local perspective, and the annotation information has obvious regional and aggregate characteristics, making it difficult to naturally and uniquely correspond to individual beams, slabs, columns, walls, or openings. Within the same work area, the repetitive arrangement of components along the grid makes their geometry and properties highly similar. If the evidence at the work surface level is directly used for component-level progress writing, it is easy to cause incorrect updates of component status, distorted progress statistics, inaccurate deviation positioning, and unreliable early warnings.

[0028] Example 1: Based on the above, this embodiment of the invention provides a BIM-based system for managing the entire construction progress of civil engineering projects. For example... Figure 1 The structure shown is that of a BIM-based civil construction progress control system. This system may include the following modules: information conversion module, slice alignment module, progress control module, and database.

[0029] The information conversion module is connected to the slice alignment module, which in turn is connected to the progress control module. All three modules—information conversion, slice alignment, and progress control—are connected to the database.

[0030] The database is used to store the parameters involved in the BIM-based civil construction progress control system. Based on the actual needs of the project, the data format, accuracy threshold and association rules of each parameter are defined to build a multi-dimensional data association system indexed by the unique identifier of BIM components. Finally, in combination with the dynamic requirements of progress control, a database architecture that supports real-time data access, historical data traceability and multi-source data fusion is designed to ensure that the data can efficiently support the entire process of construction progress planning, process tracking, deviation warning and debriefing optimization.

[0031] BIM, or Building Information Modeling, is a three-dimensional information model that digitally integrates information about the entire lifecycle of building components, including their geometry, material properties, construction, and operation and maintenance.

[0032] Figure 3 This invention provides a flowchart for the construction of a BIM-based 3D construction model and the generation of progress slices. The process involves acquiring the BIM data and establishing a 3D construction model, then directly proceeding to the core judgment step: whether the on-site evidence items carry directly locating identifiers. If the judgment result is yes, direct location is performed to the corresponding construction unit; if the judgment result is no, reverse location is performed to cover the construction units. The results of these two paths ultimately converge, both pointing to the formation of actual status slices and the query of planned status slices. After completing this step, the process terminates with the output of progress control results, clearly presenting the entire chain of pre-processing data from model acquisition to the generation of control results.

[0033] Taking a newly built comprehensive office building project in a certain city as an example: This project is a 12-story above-ground and 2-story underground reinforced concrete frame-shear wall structure building. The construction organization divides the standard floors into two major zones, east and west, and uses a grid (e.g., axes 1 to 8, A to H) as the positioning benchmark. The main civil engineering processes include formwork erection, rebar tying, concrete pouring, and masonry and plastering. To achieve full-process progress control, this embodiment is equipped with an information conversion module. Its function is to convert the BIM model on the design side and the progress plan on the planning side into a unified data structure that is verifiable and traceable on site and can be used for 4D progress simulation and deviation identification.

[0034] The information conversion module first acquires the Building Information Model (BIM) and establishes a 3D construction information model. The Building Information Model (BIM) refers to a digital building model containing the geometric shape, spatial location, attribute parameters, and structural relationships of components. The 3D construction information model refers to a set of 3D model data after the BIM has been processed for construction applications. It includes at least two types of objects: firstly, architectural spatial hierarchy elements, which are spatial organization elements used to define the boundaries of the work space and positioning benchmarks, such as buildings, floors, zones, grids, construction sections, and room / area boundaries, expressing the spatial scope of construction activities; secondly, civil engineering component objects, which are constructible objects that constitute the main structure and enclosure system of the building, such as beams, slabs, columns, shear walls, stairs, openings, and lintels for doors and windows. Each civil engineering component object has an object identifier that can be recognized by the system within the model, such as an enterprise component code, and retains its spatial affiliation with floors, zones, and grids for subsequent planning association and status display.

[0035] The information conversion module imports the construction schedule plan to generate a 4D progress simulation, thereby obtaining the planned status of each construction task at different times. The construction schedule plan refers to a set of tasks compiled according to a network plan, which at least includes information such as task name, process type, planned start and end times, logical relationships, and milestones. The 4D progress simulation refers to overlaying a time dimension onto a three-dimensional (3D) model, establishing a connection between each construction task and its corresponding spatial range or component set, thus displaying the planned progress of the task in the model at any given time. The planned status refers to the expression of the planned progress of a task at a certain moment, such as not started, in progress, completed, or with a planned completion interval. This planned status serves as a benchmark for subsequent comparison with the actual on-site status.

[0036] In one example, Figure 6 This is the overall view of the construction scene for four-dimensional progress simulation provided in the embodiments of the present invention. Figure 1 , Figure 7 This is the overall view of the construction scene for four-dimensional progress simulation provided in the embodiments of the present invention. Figure 2 , Figure 8 This is the overall view of the construction scene for four-dimensional progress simulation provided in the embodiments of the present invention. Figure 3 This example constructs a four-dimensional progress simulation model based on the acquisition of the building information model and the establishment of a three-dimensional construction information model. The construction schedule is imported and associated with the spatial scope, thereby overlaying the time dimension on the three-dimensional model to form a progress simulation that can advance over time. The system is driven by the time axis, automatically retrieving the planned status of the corresponding task within any selected time window and highlighting it in the model space as the planned coverage area. At the same time, the task list and Gantt time axis are displayed synchronously at the bottom of the interface to keep the spatial status consistent with the planned status. Through this four-dimensional progress simulation model, the system can provide an intuitive and verifiable reference for subsequent querying of planned status slices, and supports the comparison and viewing of key processes and their coverage areas from different perspectives, providing a unified planning benchmark for progress comparison, deviation location, and early warning scheduling.

[0037] After completing the 4D progress simulation, the information conversion module performs a benchmark comparability layer conversion on the 3D construction information model, generating a set of construction units. The benchmark comparability layer conversion transforms the original model representation, primarily based on component objects, into a unified comparison benchmark based on spatial operation units more commonly used in on-site operations. This ensures that regional evidence such as on-site images and work reports can be reliably matched within the model's scope for verification. A construction unit is the smallest comparable operation unit defined by a floor-zone-grid area. For example, 3F-East Zone-grid 1~4 / A~D can be considered a single construction unit, or further subdivided into multiple sub-units when the grid is large. Each construction unit is assigned a unique construction unit identifier, a code that uniquely identifies the unit within the system. For example, BLD1-F03-ZE-1_4-A_D represents building 1, 3 floors, East Zone, grids 1 to 4, and A to D, respectively, used to anchor data from both the planning and on-site sides to the same object.

[0038] Two types of index relationships are established for each construction unit to enable direct execution of subsequent positioning, comparison, and deviation aggregation: The first is the spatial label index, which uses floor number, zone name, grid range, construction section name, etc. as spatial labels to establish a rapid retrieval relationship between spatial labels and construction unit identifiers. This is used to directly convert common annotation information in site data, such as axes 1 to 4 and A to D in the 3rd floor east zone, into the corresponding set of construction units. The second is the adjacency relationship index, which records the adjacency relationships between construction units in the plane or vertical direction, such as adjacent grid units on the same floor and coaxial grid units on different floors. This allows discrete unit states to be naturally aggregated into connected regions during progress analysis, which is used to identify spatial forms with construction significance, such as continuous lag and the forefront of advancement.

[0039] During routine inspections, process acceptance, and node checks, project construction personnel take photos or short videos of procedures such as formwork erection, rebar tying, concrete pouring, and masonry plastering. When uploading these videos to mobile devices, they typically only include regional information such as floor, zone, grid range, construction section, process name, and inspection conclusions. To avoid directly writing the above-mentioned work surface image data back to the component layer, which could lead to problems such as mismatched duplicate components, component status deviating from the actual site conditions, distorted progress statistics, and unreliable early warnings, this embodiment sets up a slice alignment module. This module is used to uniformly convert on-site data and planned data into a slice data structure anchored by construction units, thereby achieving stable plan comparison and deviation location without relying on the unique identification of individual components.

[0040] During construction, construction workers and supervisors take photos of local construction progress, such as formwork erection and rebar tying, during daily inspections and upload them to their mobile devices. The slice alignment module receives on-site progress data including photos or short videos, as well as process and location information entered manually or automatically during upload. For example, at 10:12 AM on March 18, 2025, a 15-second video was taken during a rebar tying inspection on the 3rd floor east zone. The selected process for upload was rebar tying, and the location was entered as 3F-East Zone-Grid 2~4 / B~D. The inspection conclusion was checked as "self-inspection completed, awaiting concealed acceptance." At 2:35 PM on the same day, the construction worker took three photos of formwork erection and directly scanned the zone QR code at the east zone entrance. The system automatically retrieved the range of 3F-East Zone-Grid 1~3 / A~C and entered "formwork reinforcement completed, ready for rebar entry." Each upload will be organized by the system into a field evidence entry. The entry will retain at least the collection time, corresponding procedures, coverage area, core elements of the video evidence, and conclusions. The core elements of the video evidence refer to the media file and a brief summary, so that the original video can be traced and used for subsequent positioning and alignment.

[0041] Verify whether the on-site evidence items contain directly locating markers that can be used to directly anchor construction units. Directly locating markers refer to markers whose content has already fixed boundary elements such as floors, zones, and grid ranges, and can form a one-to-one correspondence with construction unit markers.

[0042] Taking scanning the zone QR code as an example, the zone QR code is bound to a specific floor, zone and grid range when it is pre-deployed. Therefore, when the on-site evidence item carries the QR code information, the system can directly retrieve the corresponding covered construction unit without component identification, such as directly locating the construction unit 3F-East Zone-Grid 1~3 / A~C.

[0043] Under this direct positioning path, the system uses construction units as the carriers of on-site evidence. It only needs to confirm the work area covered by the on-site evidence item to complete the subsequent generation of actual state slices, without needing to further determine which beam, column, or opening component corresponds to which one in the image. Since the construction unit itself is defined by the floor-zone-grid range, it is naturally consistent with the regional labeling standard commonly used when uploading on-site. Therefore, it can maintain the stability of the positioning basis in scenarios where structural components are repeatedly arranged along the grid and the geometric shapes of the components are highly similar. This avoids the erroneous distribution and updating of progress information at the same work surface level to multiple similar components, or to components that have not yet been completed. This provides a reliable data foundation for subsequent planned state slice alignment, deviation identification, and early warning triggering.

[0044] When on-site evidence items do not carry directly locating markers, reverse location is performed within the execution area to obtain the covered construction unit. The specific execution process is as follows:

[0045] Semantic information of on-site evidence items is extracted from the structured fields and unstructured descriptions of the evidence items. Semantic information refers to a set of text fields that can express the spatial range and process attributes of the evidence, such as floor number, zone name, grid range, construction section name, process name, inspection conclusion, and remarks, such as near the elevator shaft, east side beam and slab area.

[0046] Based on the semantic information of the on-site evidence items, a matching degree analysis is performed sequentially with the spatial label index of each construction unit in the construction unit set. The similarity between the semantic information of the on-site evidence items and the spatial label index of each construction unit can be analyzed by cosine similarity analysis, and the final similarity is marked as the matching degree.

[0047] The matching scores are sorted in descending order, and the top N construction units corresponding to the highest matching scores are extracted and marked as candidate construction units. The highest matching score is defined as a threshold number of candidates that the system has set in the database beforehand. After sorting all construction units by matching score, only the top N construction units are selected as candidate construction units, where N is the highest matching score. Candidate construction units refer to the set of construction units that may cover the construction area temporarily retained by the system when the coverage area cannot be uniquely determined based on semantic information alone, in order to avoid premature locking and spatial mislocation.

[0048] To further eliminate the positioning instability caused by overly coarse regional markings and the prevalence of repetitive components, the system obtains the located on-site evidence entries of the same work group or the same process within adjacent time windows, determines the priority connectivity range, and performs trimming or center-of-gravity shifting on each candidate construction unit of the covered construction unit to obtain the covered construction unit.

[0049] Among these, "same work team" refers to on-site evidence items uploaded or associated by the same construction work team (e.g., the same formwork team, rebar team, or masonry team); "same work process" refers to on-site evidence items corresponding to the same construction stage type, such as formwork erection, rebar tying, concrete pouring, or masonry plastering; "adjacent time window" refers to a continuous time range extending forward and backward from the current on-site evidence item's collection time, used to express the construction pattern of the same work team typically advancing continuously on similar work surfaces within a short period; "located on-site evidence item" refers to evidence items whose coverage of construction units can be determined during the preceding processing, such as carrying a zoning QR code, clearly defining the axis grid range, fixed camera position binding information, or containing stable scene anchor points, thus successfully converging into the construction unit set. Through the above limitations, the system obtains a set of reference evidence items that are consistent with the current evidence in terms of personnel organization or work process attributes and are consecutively adjacent in time.

[0050] After obtaining the set of reference evidence entries, the system determines the priority connectivity range. The priority connectivity range refers to a spatially continuous range of construction units, starting from the construction unit covered by the reference evidence entries and combining it with the adjacency index between construction units (i.e., an index structure recording the adjacency relationships of construction units on the same floor or vertically). This range reflects the fact that the same work group or the same process typically does not cross multiple non-adjacent work surfaces within a short period, but rather proceeds continuously along the grid direction or the direction of work advancement. This priority connectivity range is used to spatially constrain the current candidate construction units, thereby reducing candidate range drift caused by coarse area labeling.

[0051] The system performs trimming or center-of-gravity shifting on each candidate construction unit to obtain the coverage construction unit. Among them, "pruning" refers to removing candidate construction units from the candidate set when there are construction units that are not obviously adjacent to the priority connectivity range or are spatially far away from it, so that the candidate set converges to a spatial range consistent with the continuous operation pattern. "Center of gravity shifting" refers to moving the spatial center of the candidate set (i.e., the concentrated area of ​​candidate construction units in the grid coordinates or plane position) towards the priority connectivity range when the candidate construction unit set is still scattered but cannot be directly removed, so that the final retained candidate construction units are closer to the actual progress path reflected by the reference evidence item. The pruning and center of gravity shifting are repeated until the covering construction unit is obtained. After pruning or center of gravity shifting, the covering construction unit output by the system is the final construction unit set that can carry the construction status reflected by the current site evidence item. This covering construction unit can be directly used to generate actual status slices and align and compare them with planned status slices, so as to achieve more stable progress positioning and subsequent progress writing without forcibly mapping to a single component object, and reduce the problems of progress statistics distortion and unreliable warning caused by erroneous updates of component status in the case of repeated components.

[0052] On-site evidence is mostly limited to localized images of the work area, and the annotations are often regionalized. If these images are used directly for plan comparison without adjusting their coverage, it can easily lead to erroneous diffusion or shrinking of progress status, resulting in unreliable deviation identification and early warning triggering. This embodiment aligns the data structure of construction units and processes using a unified approach, and introduces mapping domain adjustment and process consistency adjustment before and after alignment. This ensures that the spatial carrying capacity used for progress verification is stable, traceable, and consistent with the plan's baseline.

[0053] Within the corresponding construction unit of the on-site evidence item, an actual state slice is generated. An actual state slice refers to a record of the actual on-site state within the covered construction unit, expressed according to a unified data structure of construction unit-process. For example, the process status (e.g., not started, in progress, completed, pending confirmation) is given for processes such as formwork erection, rebar tying, and concrete pouring. Photos, short video clips, work reports, or acceptance conclusions associated with this process status are linked in an index format. An evidence chain index refers to an index identifier used to point to the corresponding evidence item and its media file or record item, ensuring that any subsequent status change can be traced back to the original evidence source.

[0054] The system uses the construction unit identifier covering the construction unit as an index to query the corresponding planned status slice from the construction 3D information model. The planned status slice refers to the planned baseline state obtained from the construction schedule plan and 4D schedule simulation. It is also expressed using a unified data structure of construction unit-process, and is used to describe the planned status of the construction unit in the corresponding process within the current time window.

[0055] By using the above methods, the actual state slice and the planned state slice are consistent in data structure, indexing method and process scope, so that the plan can be directly compared and deviations can be identified, avoiding alignment difficulties caused by the inability to correspond to components due to the fact that the on-site evidence is mainly area-based.

[0056] Mapping domain adjustment is performed on the actual state slices of the covered construction unit. Mapping domain adjustment is to structurally adjust the spatial range covered by the actual state slices so that it is as consistent as possible with the visible range of the planned state slices in terms of diameter, and to suppress the range deviation caused by local shooting or continuous construction.

[0057] Mapping domain adjustment specifically refers to: first analyzing whether the visible ranges of the actual state slice and the planned state slice are consistent. Specifically, the system calculates the set similarity of the construction unit sets within the visible ranges of the two, for example, based on the set similarity ratio of intersection and union. When the set similarity is less than the minimum allowed value of similarity preset in the database, it is determined that the visible ranges are inconsistent.

[0058] If inconsistencies in the visible range occur, the actual state slice is shrunk: using the construction units clearly covered by the on-site evidence as the initial seed set, connectivity filtering is performed on the adjacency index of the construction units, eliminating discrete units that are not connected to the seed set or only touch through a small boundary, thus obtaining the minimum connected coverage region. The minimum connected coverage region refers to the smallest set of construction units that can cover the range supported by the on-site evidence while maintaining connectivity with it. This is used to prevent the progress status caused by local images from being mistakenly diffused to other construction units corresponding to a large number of similar components within the same partition.

[0059] When the spatial range corresponding to the actual state slice is not less than the preset defined spatial range, the system splits the covered construction unit according to the axis grid direction, dividing the original coverage area into multiple smaller sub-coverage areas along the axis grid boundary. The specific number of divisions is determined by the ratio between the spatial range and the defined spatial range. The ratio between the spatial range and the defined spatial range is marked as the range ratio. The database stores a mapping table between the range ratio and the number of sub-coverage area divisions. Therefore, the corresponding number of divisions can be queried from the mapping table based on the range ratio. For example, the range of axes one to four can be split into two segments: axes one to two and axes three to four, or divided into zones according to axes A to B and C to D, and independent actual state slices can be generated for each sub-coverage area. Among them, the defined spatial range refers to the preset spatial range threshold for triggering refinement in the database (which can be characterized by the number of construction units or the axis grid span), which is used to avoid the state being too coarse due to the excessive coverage of a single piece of evidence; the sub-coverage area refers to the smaller coverage area obtained after splitting. Its slices can be used for more refined plan comparison and deviation location, reducing the probability of erroneous updates in scenarios with dense repetitive components.

[0060] After the construction worker uploads the rebar tying inspection video in the third-floor east section, the system has completed the mapping domain adjustment and generated the corresponding actual state slice for the construction unit. At this time, the system further performs process consistency adjustment on the actual state slice. Process consistency adjustment refers to standardizing the sequential relationship between different process states within the same construction unit, avoiding errors caused by skipping processes due to local evidence on the work surface. The process state machine refers to predefined process sequence constraints, such as rebar tying only proceeding after formwork erection is completed, and concrete pouring only proceeding after rebar tying meets certain conditions, used to determine whether the combination of process states within the same construction unit is reasonable.

[0061] In this embodiment, when a state combination in the actual state slice of the same construction unit violates the process state machine, for example, if the concrete pouring of a construction unit is marked as completed while the formwork erection or rebar tying is still in a state of not having started or being obviously insufficient, the system does not directly use this conflicting state for plan comparison and subsequent write-back. Instead, it adjusts the conflicting process state in the actual state slice to a state pending verification. Here, a conflicting process state refers to the one or more process states that cause the process sequence contradiction; a state pending verification means that the process state is not yet a definite conclusion, but is only retained as a state mark that needs further verification.

[0062] To avoid error propagation, the system also restricts the pending verification status of the construction unit to be effective only at the construction unit level. That is, the status is only used for prompting and verification at the slice level of the construction unit-process, and does not directly drill down to the component object level to update the component completion status, thereby reducing progress distortion caused by erroneous write-back in scenarios with dense repetitive components.

[0063] Simultaneously, the system generates targeted review scopes or evidence supplementation requirements for conflicting construction units. The targeted review scope refers to the review space defined down to the floor, zone, grid line, and boundary of the construction unit, guiding construction workers or supervisors to prioritize verifying conflicting units during subsequent inspections. Evidence supplementation requirements specify the constraints on the supplementary data, such as requiring additional images of stable reference points including grid line intersections, beam ends, or opening edges, to reconfirm the process status within the construction unit and eliminate conflicts.

[0064] Figure 4 This is a flowchart of the progress control result processing and closed-loop formation provided by this embodiment of the invention. Based on the progress control result, the process begins with a judgment step to determine whether the actual state slice needs to be presented at the component level. If the judgment result is negative, the process directly executes the confirmation of maintaining the construction unit level; if the judgment result is positive, the process executes write-back control on the component state, and then proceeds to the next judgment step to determine whether the component meets the write-back conditions. If the write-back conditions are met, the component level write-back is executed; if the write-back conditions are not met, the process executes the confirmation of not writing back at the construction unit level. Finally, the three paths of maintaining the construction unit level confirmation, component level write-back, and construction unit level confirmation of not writing back all converge to form a closed-loop progress control system, completing the processing of control results and the construction of the closed loop.

[0065] When a project requires component-level progress display, component-level node acceptance, or fine-grained critical path location—for example, when the completion status of components such as beams, slabs, and columns needs to be displayed in the model with coloring, or when critical node acceptance needs to be assigned to a specific set of components—the system, based on the aforementioned process consistency adjustment, performs component disambiguation optimization on the actual state slices. Component disambiguation optimization refers to filtering and converging candidate component sets based on the construction unit range, adjacency relationships, and process state machine constraints when there are many duplicate components in the same floor and zone, and work surface evidence is difficult to uniquely correspond to a single component. This outputs a stable set of components that can be rewritten, preventing the same work surface information from being scattered across multiple similar components or updated to incomplete components. This ensures that component-level progress display and critical node judgment are based on a stable and traceable state.

[0066] First, key construction units are identified based on the actual state slice and the planned state slice. Key construction units refer to those that exhibit significant deviations in the plan comparison, fall within the scope of tasks associated with the critical path, or have triggered a pending review state. These key units are used to limit the scope of disambiguation optimization and avoid indiscriminate drilling across the entire floor.

[0067] Driven by the actual state slice corresponding to the key construction unit, the candidate component set of the key construction unit is extracted to form the component domain to be judged: where the candidate component set refers to the set of component objects that fall within the spatial range of the construction unit and are related to the process; the component domain to be judged refers to the component judgment range formed by the candidate component set, which is the object domain that needs to be further disambiguated and screened before the component layer is written back.

[0068] The system clusters components with the same geometric shape, structural attributes, and process characteristics in the candidate component set to form duplicate component groups. Among them, duplicate component groups refer to a set of components with the same cross-section or size, the same structural type, and the same process path in the same construction area, such as multiple beams with the same cross-section or structural columns of the same specification. The system uses repeating component groups as disambiguation boundaries. Within these boundaries, the goal is to ensure that the component process status of the component domain to be determined is consistent with both the actual and planned state slices. The following constraints are applied: First, a spatial continuity constraint, which reflects the pattern that the progress of the work surface usually unfolds continuously in a continuous manner or along the axis, preventing the same group of components from being marked as completed in a discrete, jump-like manner; Second, a repeating group consistency constraint, which restricts the write-back results within the same repeating component group to be more consistent with the progress of the entire area rather than random dispersion; Third, an adjacency consistency constraint, which requires that the component states derived from adjacent construction units be compatible with each other, avoiding discontinuous conflicts at the boundaries; Fourth, a process state machine constraint, which ensures that the sequence of processes such as formwork, reinforcement, and pouring is not violated at the component layer; Fifth, an event boundary constraint, which limits events with a clear scope, such as pump truck pouring and material requisition, to the scope of their covered construction units, avoiding out-of-boundary updates.

[0069] Figure 5 This is a flowchart of the component disambiguation optimization and structural adjustment loop provided in this embodiment of the invention. In this process, the core judgment step is to determine whether the candidate component meets the defined quantity constraint. If it does, it is placed into the set of components that can be stably written back; if not, it is placed into the set of components that cannot be uniquely determined, and the number of abnormal components is counted. Then, the next judgment step is to determine whether the number of abnormal components is not lower than a defined value, i.e., to define the number of abnormal components. If the judgment result is negative, the process of maintaining the construction unit layer confirmation without writing back is executed; if the judgment result is positive, structural adjustment is executed, leading to the judgment step of selecting a structural adjustment method, corresponding to four adjustment methods: Method 1: Subdivide construction units to reconstruct slices; Method 2: Merge adjacent construction units; Method 3: Reconstruct construction units into construction segment morphology reconstruction slices; Method 4: Generate minimum supplementary sampling instructions to obtain supplementary evidence. After all four adjustment methods are executed, they all point to re-executing component disambiguation optimization and flowing back to the step of whether the candidate component meets the defined quantity constraint, forming a loop optimization logic. Placing components into the set of components that can be stably written back and maintaining the construction unit layer confirmation without writing back both ultimately converge to completing the component-level progress presentation and recording, terminating the process.

[0070] After the constraints are applied, the system determines whether the number of constraints that each candidate component satisfies reaches the preset threshold. The threshold is a constraint satisfaction threshold stored in the database to determine whether the constraints are sufficiently consistent and can be stably written back (e.g., multiple constraints need to be satisfied simultaneously rather than just a single constraint).

[0071] If a candidate component satisfies the constraint condition of the defined quantity, then the component is placed in the set of components that can be stably written back; otherwise, if a candidate component does not satisfy the constraint condition of the defined quantity, then the component is placed in the set of components that cannot be uniquely determined.

[0072] For example, at 15:30 on a certain day, the system receives a site evidence entry uploaded by the construction worker. The entry specifies the area of ​​the third-floor east zone, the first and second axes of the grid, and the area of ​​A to B axes, and includes a rebar tying inspection, along with a short video clip. The information conversion module has already located this area into several construction units, one of which is identified as Third Floor—East Zone—Grid First and Second Axes / A to B Axes—Construction Unit C1. The system generates an actual status slice of the rebar work within construction unit C1 (e.g., displayed as "In Progress" and linked to the evidence chain index of the video), and simultaneously retrieves the planned status slice of the same construction unit C1 within the same time window from the four-dimensional progress simulation (e.g., the plan should be "Completed and ready to proceed with concrete pouring"). Because construction unit C1 showed a significant deviation in the plan comparison, indicating that the plan had been completed but the actual work was still in progress, and because the beam and slab reinforcement corresponding to C1 was a key pre-construction process for the capping of the structure of this floor (subsequent pouring and formwork removal depended on its completion), the system identified C1 as a key construction unit and proceeded to the component layer for disambiguation, instead of drilling down to process other construction units in the east section of the third floor that had no deviation.

[0073] When drilling down into the critical construction unit C1, the system uses the actual state slice corresponding to C1 as the driving force to extract component objects related to the rebar process within the spatial range of C1, forming a candidate component set, which constitutes the component domain to be determined. For example, within the range of C1, there are 4 frame beams with the same cross-section (denoted as beams L1 to L4), 2 floor slab partition strips (slabs S1 and S2), and 2 structural columns of the same specification (columns Z1 and Z2). These components all fall within the spatial boundary of C1 and all involve the rebar binding process, so they are included in the candidate component set. The component domain to be determined is composed of the above candidate component set. It is the object domain that the system needs to determine which components have reached the rebar state that matches the actual slice and which components have not yet reached the state. This is used to avoid directly mapping the work surface evidence to a certain beam or column, which would cause erroneous write-back.

[0074] Subsequently, the system clusters components with the same geometric shape, structural attributes, and process characteristics in the candidate component set to form duplicate component groups. In this example, beams L1 to L4 are all frame beams with the same cross-section and the same reinforcement process path, so they are clustered into the same duplicate component group; columns Z1 and Z2 are structural columns of the same specification and have the same process path, so they also form another duplicate component group; slabs S1 and S2 are slab strips of the same thickness and use the same binding and acceptance rules, so they can form a slab-type duplicate component group. The system uses repeating component groups as disambiguation boundaries. Within each group, constraints are applied with the goal of ensuring that the component's process status is consistent with the actual and planned state slices of construction unit C1. For example, the spatial continuous advancement constraint requires that the completion status of beams in the same group should better conform to the common pattern of continuous advancement along the axis, avoiding randomly marking "completed" on distant beams. The repeating group consistency constraint requires that the write-back results of beams in the same group should prioritize the progress of large areas rather than discrete and scattered. The adjacency consistency constraint requires that if adjacent construction units of C1 (e.g., construction unit C2 of axes two to three / A to B in the same floor grid) have been confirmed as completed, the beams in C1 closer to the boundary of C2 are more likely to be completed first, thus ensuring that the states on both sides of the boundary are compatible and that there is no abrupt break. The process state machine constraint ensures that the reinforcement status cannot cross the formwork and that the pouring status cannot be written as completed when the reinforcement conditions are not met. The event boundary constraint is used to limit the impact of events with a clear coverage area. For example, if the pump truck pouring event only covers C2 and not C1, it cannot be inferred that the beams and slabs in C1 have met the reinforcement completion conditions before pouring because of the pouring of adjacent areas.

[0075] Under the aforementioned constraints, the system can divide candidate components into two output categories: For example, after considering both adjacency consistency and continuous advancement, the system determines that beams L1 and L2 (close to the boundary of C2 and with visible steel mesh and binding completion features in the evidence chain) meet the constraint conditions of the defined quantity, and includes them in the component set that can be stably rewritten, and rewrites their steel reinforcement process status as "completed"; while beams L3 and L4 and columns Z1 and Z2 are included in the component set that cannot be uniquely determined because the evidence only covers a local area and cannot be uniquely distinguished within the repeated component group, and at the same time cannot meet a sufficient number of constraint conditions.

[0076] Suppose that there are 6 components in the set of components that cannot be uniquely determined, and these are marked as the number of anomalous components. This number is then compared with the defined number of anomalous components stored in the database (e.g., 4). The defined number of anomalous components represents the minimum allowed number of anomalous components. When the number of anomalous components is not less than the defined number, the system determines that it is difficult to stably distinguish duplicate components based solely on existing evidence and constraints, and therefore initiates structural adjustment. Structural adjustment refers to adjusting the spatial organization of construction units or supplementing the minimum evidence that can break the symmetry, without changing the construction unit-process comparison criteria, so that subsequent component disambiguation optimization can more easily achieve unique and stable results.

[0077] Structural adjustments in this embodiment include the following types of actions, which can be selected or combined according to the triggering cause: First, if it is determined that the conflict mainly stems from the continuous advancement of the work surface along the axial direction but the unit is too thick, resulting in multiple beams with the same cross-section being covered by the same domain, then the construction unit C1 to which the abnormal component belongs is subdivided into several smaller construction units along the advancement direction. The specific number of subdivisions is determined by the number of abnormal components. The database stores a mapping table between the number of abnormal components and the number of subdivisions of construction units. Based on the number of abnormal components, the corresponding number of subdivisions of construction units can be queried from the database. For example, C1 is split into C1-1 and C1-2 along the first and second axis directions of the grid, and the planned state slices and actual state slices corresponding to these new construction units are reconstructed simultaneously. Then, component disambiguation optimization is re-executed on the subdivided units. Among them, the advancement direction subdivision refers to cutting the coverage area along the main advancement axis direction of the work surface on site, so that similar components are no longer driven by the same slice as a whole. Secondly, when fragmented evidence leads to local conflicts—for example, multiple photos taken from different corners, creating seemingly contradictory states near unit boundaries—the system merges several adjacent construction units into larger units to stabilize a consistent state. Then, it gradually refines the units on this stable foundation to reduce boundary conflicts caused by fragmented evidence. Thirdly, when the site is organized by construction sections / flow sections, but the construction units remain regular grids, causing evidence to fail to converge over a long period, the system reconstructs the construction units from a grid shape to a construction section shape. This involves regenerating units according to the actual boundaries of the construction sections on site and simultaneously reconstructing slices, ensuring that the evidence labeling caliber matches the comparable caliber of the model. This reduces the uncertainty caused by misalignment between area labeling and unit division.

[0078] When the aforementioned spatial organization adjustments are insufficient to break the symmetry between duplicate components, the system generates a minimum supplementary sampling instruction and performs component disambiguation optimization again. The minimum supplementary sampling instruction refers to minimizing the constraints on the scope and content of the supplementary sampling without increasing the on-site workload: the supplementary sampling scope is limited to conflicting construction units and their adjacent areas; the supplementary sampling content must include structural anchor points that can distinguish adjacent duplicate components, such as grid intersection markers, beam end nodes, column base positions, opening edges, or core tube boundaries; structural anchor points refer to reference features that are stable in position during construction, correspond to them in the model, and can break the symmetry of the uniform appearance of multiple components. After the supplementary sampling is completed, the system updates the actual state slice with new evidence and reruns the component disambiguation optimization, thereby gradually promoting the convergence of the disambiguation results. Finally, it outputs a set of components that can be stably written back, while keeping the remaining non-uniquely determinable components only confirmed at the construction unit level without being written back at the component level.

[0079] After aligning the actual state slices with the planned state slices, the progress control module outputs control results for progress comparison, early warning, and scheduling, and performs write-back control when component-level presentation is required. The progress control module refers to the functional module that performs difference analysis, cluster aggregation, early warning triggering, and scheduling suggestion generation between the planned and actual status under the unified standard of construction unit-process.

[0080] The system first determines the deviation type and calculates the deviation amount for each construction unit under each process based on the difference in process status between the actual and planned state slices. Process status difference refers to the difference between the actual and planned states within the same construction unit and process. Deviation types include: ahead of schedule, as planned, behind schedule, and pending confirmation: ahead of schedule means the actual state arrives earlier than planned; as planned means the actual state matches the plan; behind schedule means the actual state has not reached the planned state; pending confirmation means the actual state is marked as requiring verification due to insufficient or conflicting evidence. Deviation amount is a numerical value used to quantify the degree of deviation, such as expressed as time difference (the difference between the actual arrival time and the planned arrival time) or process stage difference (planned completion but actual progress). For example, if a construction unit in the third-floor east section has a planned state of "completed" for the rebar tying process, but an actual state of "in progress," then this unit is marked as behind schedule, and the corresponding deviation amount is recorded.

[0081] Subsequently, the system aggregates spatially adjacent construction units with consistent deviation types within the same process, forming deviation clusters and outputting their range and evidence association. Spatially adjacent refers to two construction units sharing a boundary in the planar division or being directly connected through an adjacency relationship index (e.g., adjacent vertically, horizontally, or vertically), thus forming a continuous construction unit area in space. A deviation cluster refers to a set of spatially adjacent and connected construction units exhibiting the same deviation type in the same process, used to reflect deviation patterns significant in construction organization, such as delayed or advanced contiguous areas. The system outputs the spatial range of the deviation cluster (floor, zone, grid range, and the set of construction units it contains), the involved process (e.g., rebar tying or concrete pouring), and the associated evidence chain index. The evidence chain index refers to the index identifier traceable to on-site evidence entries, used to explain the source of the images, work reports, or acceptance records upon which the deviation determination of the cluster is based.

[0082] The system generates an early warning message when the deviation meets preset trigger conditions, or when a deviation cluster is associated with a critical path segment. The preset trigger conditions refer to threshold rules used to trigger the early warning, such as a lag duration exceeding the preset maximum allowed lag duration in the database, a lag cluster size exceeding a threshold, or the duration of the pending confirmation state exceeding the preset maximum allowed duration in the database. The critical path segment refers to the spatial and operational range covered by the task chain in the network plan that plays a decisive role in the overall project duration; if this segment is affected, subsequent critical nodes may be directly delayed. The early warning message includes at least: the location of the deviation cluster, the deviation type, the involved operational processes, the potentially affected nodes, and the corresponding evidence chain index.

[0083] Based on the early warning information, the system further generates scheduling suggestions and clarifies their scope of application. These scheduling suggestions refer to outputs aimed at adjusting the construction organization, and include at least one or more of the following: First, resource adjustment suggestions, such as suggesting additional rebar teams or key equipment for bottleneck processes in lagging clusters, specifying the scope of the construction units and processes involved; second, work surface adjustment suggestions, such as splitting lagging clusters into parallel work surfaces according to the grid direction, or adjusting the handover sequence of flow sections to reduce waiting time; third, plan window adjustment suggestions, such as moving local process windows forward, backward, or compressed without changing the overall project duration control target, and simultaneously updating the four-dimensional progress simulation to maintain consistency in the plan. The system ultimately outputs early warning information and scheduling suggestions as the control results, and archives these results along with the corresponding actual status slices, planned status slices, and evidence chain indexes. This allows for tracing back how deviations were determined, why early warnings were triggered, and what evidence was used to generate the suggestions during subsequent reviews, forming a stable and traceable closed loop for the entire process of progress control.

[0084] The second aspect of this invention provides a BIM-based method for full-process progress control in civil engineering construction, comprising: Step 1: acquiring a civil engineering information model (BIM) and establishing a three-dimensional construction information model, importing the construction schedule plan to generate a 4D progress simulation, and converting the three-dimensional construction information model into a set of construction units defined by floors, zones, and grids, with each construction unit having a unique construction unit identifier and establishing a corresponding spatial index relationship; Step 2: collecting on-site progress data during the construction process and converting it into on-site evidence items, determining covering construction units based on the on-site evidence items, forming actual state slices within the covering construction units, and querying the corresponding planned state slices using the identifiers corresponding to the covering construction units, aligning the actual state slices and planned state slices with a unified data structure; Step 3: based on the aligned actual state slices and planned state slices, outputting control results for progress comparison, early warning, and scheduling, and when it is necessary to present the actual state slices at the component level, performing write-back control on the component status of the actual state slices, so that components that do not meet the write-back conditions remain confirmed at the construction unit level without being written back at the component level.

[0085] In Example 2, since the semantic information entered on-site usually consists of floors, zones, grid ranges, construction sections, process names, and a small amount of directional descriptions, there are problems such as inconsistent expressions, missing words, frequent synonym substitutions, and vague descriptions such as "east side" and "near the elevator shaft" that are difficult to directly correspond to fixed text templates. If only pure text similarity is used for measurement, it is easy to judge the real location as low similarity due to differences in wording, or to misjudge irrelevant areas as high similarity due to shared common vocabulary, thereby expanding the candidate range and introducing location jumps. At the same time, the components in the same floor and zone are repeatedly arranged along the grid, and the evidence of the work surface is naturally regional and collective. The key to positioning is not whether the text is similar, but whether the floor and zone are consistent, whether the grid range is included, whether the directional anchor point is hit, and whether the adjacent relationship is connected, etc., which can be objectively verified spatial conditions. Under the condition that Example 1 remains unchanged, the semantic information is decomposed into verifiable spatial component evidence, and the matching degree is calculated and verified item by item according to the component.

[0086] Specifically, the system first performs structured parsing of the semantic information of the on-site evidence items: extracting floor number, zone name, grid range, construction section name, process name, and location description (e.g., near elevator shaft, near facade, and east side) as independent fields, and normalizing synonymous expressions. Spatial component evidence refers to elements in the above fields that can be verified item by item against the spatial label index of the construction unit. For example, floor and zone are spatial hierarchy elements, grid range is a positioning reference element, and elevator shaft, stairwell, and facade are spatial reference elements corresponding to stable scene anchor points. Through this parsing, what was originally a loose text label is converted into multiple individually identifiable consistency components, thus providing input for subsequent item-by-item verification.

[0087] Subsequently, the system performs component-by-component verification for each construction unit, converting the verification results of each component into numerical contributions to form the matching degree value of the construction unit. Specifically, the system performs strong consistency verification on the floor zoning component: if the floor and zoning labels of the construction unit are consistent with the evidence entry, the component is recorded as a full contribution; if they are inconsistent, the construction unit is directly judged as unmatchable and eliminated. The system performs inclusion relationship verification on the grid range component: if the evidence entry provides clear grid endpoints (such as axis 1 to 2, axis A to B), it determines whether the grid range of the construction unit is equivalent to or contains that range; equivalence is recorded as a higher contribution, inclusion as a second-highest contribution, and non-inclusion as zero contribution. The system performs anchor point hit verification on the orientation / adjacent component: if the evidence entry contains orientation descriptions such as near elevator shafts, it queries the spatial label of the construction unit or its pre-labeled adjacent element labels; a hit is recorded as a contribution, and a miss is recorded as zero contribution. The system performs consistency checks on construction segment / process components: if a task mapping strongly related to the construction segment or process exists in the construction unit label, it is recorded as a contribution; otherwise, it is recorded as a low or zero contribution. The contributions of each component are weighted and summed according to preset weights to obtain a matching degree value, so that the matching degree is ultimately expressed as a sortable quantitative result, rather than just a judgment of similarity / dissimilarity between texts.

[0088] Taking the third-floor east zone of this project as an example, the semantic information of a certain on-site evidence item is: third-floor east zone, grid lines 1 to 2 and A to B, rebar tying, near the elevator shaft. The system decomposes it into spatial component evidence: floor = third floor, zone = east zone, grid range = 1 to 2 / A to B, process = rebar tying, directional anchor point = adjacent to elevator shaft. For candidate construction unit C1 (labeled as third floor - east zone - 1 to 2 / A to B - adjacent to elevator shaft), the floor zone consistency contribution is 1.0; the grid range equivalent contribution is 0.6; the anchor point hit contribution is 0.2; and the process consistency contribution is 0.2. Then, the matching degree is obtained by weighted summation as M(C1) = 1.0 × 0.4 + 0.6 × 0.3 + 0.2 × 0.2 + 0.2 × 0.1 = 0.64, where 0.4, 0.3, 0.2, and 0.1 are the weights corresponding to each contribution in the example, and the matching degree is a value between 0 and 1. For another construction unit C2 (also a three-story east zone, but with a grid between axes two and three / A and B and not adjacent to the elevator shaft), the floor zone consistency contribution is still satisfied; the contribution of the grid range partially containing or not containing the area approaches 0; the contribution of missing anchor points is 0; and the contribution of process consistency is 0.2, resulting in a lower matching degree, for example, M(C2) = 0.40 + 0.00 + 0.00 + 0.02 = 0.42. Based on this, the system sorts the matching degree values ​​of all construction units, extracts the top-ranked construction units as candidate construction units, and then performs pruning or centroid shifting based on the priority connectivity range formed by the location evidence in adjacent time windows, thereby outputting the final covered construction units.

[0089] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0090] The above description is only an optional embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A BIM-based whole-process progress control system for civil construction, characterized in that, The system comprises: An information conversion module for obtaining a building information model BIM and establishing a construction three-dimensional information model, importing a construction progress plan to generate a 4D progress simulation, and converting the construction three-dimensional information model into a construction unit set defined by floor-partition-axis network boundaries, the construction unit having a unique construction unit identifier and establishing a corresponding spatial index relationship; A slice alignment module for collecting construction process site progress data and converting it into site evidence items, determining a coverage construction unit based on the site evidence items, forming an actual state slice within the coverage construction unit, querying the corresponding planned state slice with the identifier corresponding to the coverage construction unit, and aligning the actual state slice and the planned state slice in a unified data structure; A progress control module for outputting control results for progress comparison, early warning and scheduling based on the aligned actual state slice and planned state slice, and performing write-back control on the component state of the actual state slice when component-level presentation of the actual state slice is required, so that components that do not meet the write-back conditions remain at the construction unit level and do not perform component-level write-back. 2.The BIM-based whole-process progress management and control system for civil construction according to claim 1, characterized in that, The specific process of obtaining a building information model BIM and establishing a construction three-dimensional information model is as follows: Obtain a building information model BIM and establish a construction three-dimensional information model, the construction three-dimensional information model including building space level elements defining the boundaries of the work space and positioning reference, and civil component objects constituting the main structure and enclosure of the building; Import a construction progress plan to generate a 4D progress simulation to obtain the planned state of each construction task at different times; Perform reference comparable layer conversion on the construction three-dimensional information model to generate a construction unit set; The construction unit is divided by floor-partition-axis network range and has a unique construction unit identifier; Establish a spatial tag index and a neighboring relationship index for each construction unit. 3.The BIM-based whole-process progress management and control system for civil construction of claim 1, wherein, The specific process of collecting construction process site progress data and converting it into site evidence items is as follows: Collect construction process site progress data and perform source data conversion to generate site evidence items; The site progress data includes image data and corresponding process-related information; The site evidence items cover the core elements of time, process, location and image supporting materials; When the site evidence items carry a direct positioning identifier, the site evidence items are directly positioned to the corresponding construction unit based on the direct positioning identifier to obtain a coverage construction unit; When the site evidence items do not carry a direct positioning identifier, perform area reverse positioning to obtain a coverage construction unit. 4.The BIM-based whole-process progress management and control system for civil construction according to claim 3, characterized in that, The specific execution process of performing area reverse positioning to obtain a coverage construction unit is as follows: Obtain the semantic information of the site evidence items; Based on the semantic information of the site evidence items, perform matching degree analysis with the spatial tag index of each construction unit in the construction unit set in turn; Sort the matching degrees in descending order; Extract a number of construction units corresponding to the top matching degrees and mark them as candidate construction units; Obtaining positioned field evidence entries of the same team or the same process in adjacent time windows, determining a priority connectivity range, and performing pruning or barycenter migration on each candidate construction unit to obtain a coverage construction unit. 5.The BIM-based whole-process progress management and control system for civil construction according to claim 1, wherein, The actual state slice and the planned state slice are aligned in a unified data structure, and the alignment process is specifically as follows: generating an actual state slice of the field evidence entry in the corresponding coverage construction unit; querying a planned state slice from the construction three-dimensional information model by taking the construction unit identifier of the coverage construction unit as an index; wherein the actual state slice and the planned state slice are expressed in a unified data structure of a construction unit-process, and contain a process state and an evidence chain index associated with the process state; performing mapping domain adjustment on the actual state slice; the mapping domain adjustment specifically refers to: analyzing the consistency of the visible range of the actual state slice and the planned state slice; when the visible ranges of the actual state slice and the planned state slice are inconsistent, performing contraction on the visible range of the actual state slice to form a minimum connected coverage domain; when the spatial range corresponding to the actual state slice is not lower than the defined spatial range, splitting the coverage construction unit in the axis net direction to generate a plurality of sub-coverage domains and respectively generating actual state slices; after the mapping domain adjustment is completed, performing process consistency adjustment on the actual state slice. 6.The BIM-based whole-process progress management and control system for civil construction according to claim 5, characterized in that, The process consistency adjustment on the actual state slice is specifically as follows: when the actual state slice in the same construction unit appears a state combination that violates the process state machine; adjusting the conflict process state in the actual state slice to a to-be-reviewed state, and limiting the to-be-reviewed state to be effective only at the construction unit level; and synchronously generating a directed review range or evidence supplement requirement for the conflict construction unit; when there is a component-level progress display, a component-level node acceptance or a critical path fine positioning requirement, performing component disambiguation optimization on the actual state slice.

7. The BIM-based whole-process progress management and control system for civil construction of claim 6, wherein, The component disambiguation optimization on the actual state slice is specifically performed as follows: determining a key construction unit based on the actual state slice and the planned state slice; taking the actual state slice corresponding to the key construction unit as a driving force, extracting a candidate component set of the key construction unit to form a to-be-judged component domain; wherein, the components in the candidate component set that have the same geometric shape, construction attribute and process feature are clustered to form a repeated component group; within the boundary of the repeated component group, a spatial continuous propulsion constraint, a repeated group consistency constraint, an adjacency consistency constraint, a process state machine constraint and an event boundary constraint are applied to the to-be-judged component domain so that the component process state of the to-be-judged component domain is consistent with the actual state slice and the planned state slice; if a certain candidate component meets a defined number of constraint conditions, the component is put into a stably writable component set; otherwise, if a certain candidate component does not meet a defined number of constraint conditions, the component is put into a non-unique component set; outputting the stably writable component set and the non-unique component set, and keeping the non-unique component set at the construction unit level without performing component layer writing. 8.The BIM-based whole-process progress management and control system for civil construction of claim 7, wherein, the component is put into the non-unique component set, which also includes: Counting the number of components in the component set that cannot be uniquely determined, and marking it as an abnormal component number; Comparing the abnormal component number with the defined abnormal component number stored in the database; If the abnormal component number is not less than the defined abnormal component number, performing structural adjustment to promote disambiguation convergence; The structural adjustment includes: Subdividing the construction unit to which the abnormal component belongs into several construction units along the advancing direction, and synchronously rebuilding the plan state slice and the actual state slice corresponding to the several construction units, and performing component disambiguation optimization after rebuilding; When the field evidence item presents a fragmented feature, merging several adjacent construction units; When the field is organized by construction section and the construction unit division is inconsistent, reconstructing the construction unit from a grid form to a construction section form and synchronously rebuilding the slice; Generating a minimum supplementary mining instruction, which is limited to the conflict construction unit and its adjacent range and specifies the structural anchor points that must be included, to repeat the component disambiguation optimization after obtaining supplementary evidence to break symmetry. 9.The BIM-based whole-process progress management and control system for civil construction of claim 1, wherein, The output is used for progress comparison and warning and scheduling control results, and the specific process is: Based on the process state difference between the actual state slice and the plan state slice, determine the deviation type of each construction unit at each process, the deviation type includes ahead of schedule, on schedule, behind schedule and pending confirmation, and calculate the corresponding deviation amount; Aggregate the construction units that are spatially adjacent and have consistent deviation types under the same process to form a deviation cluster, and output the spatial range, involved process, and associated evidence chain index of the deviation cluster; When the deviation amount meets the preset triggering condition, or the deviation cluster is associated with a critical path section, generate a warning information; Based on the warning information, generate a scheduling suggestion, which includes at least one of resource adjustment suggestion, work face adjustment suggestion or plan window adjustment suggestion, and give the construction unit range and process range of the suggested effect; Output the warning information and the scheduling suggestion as the control result, and archive the control result with the corresponding actual state slice, plan state slice and evidence chain index to support subsequent tracing and review.

10. A BIM-based whole-process progress management and control method for civil construction, characterized in that, It includes: Step one, obtain the building information model BIM and establish the construction three-dimensional information model, import the construction progress plan to generate the 4D progress simulation, and convert the construction three-dimensional information model into a construction unit set defined by floor-zone-axis network, the construction unit has a unique construction unit identifier and establishes a corresponding spatial index relationship; Step two, collect the field progress data during construction and convert it into field evidence items, determine the coverage construction unit based on the field evidence items, form the actual state slice in the coverage construction unit, and query the corresponding plan state slice with the identifier corresponding to the coverage construction unit, align the actual state slice and the plan state slice in a unified data structure; Step three, based on the aligned actual state slice and the planned state slice, output the management and control results for progress comparison, early warning and scheduling, and when it is necessary to present the actual state slice at the component level, write back control is performed on the component state of the actual state slice, so that components that do not meet the write back conditions remain at the construction unit level confirmation without performing component layer write back.

Images