A BIM-based intelligent management system for building engineering projects

Abstract

The application discloses a kind of based on BIM's construction engineering project intelligent management system, belong to engineering management technical field;The system of the present application runs in round mode, each round locks BIM object and space topology snapshot, specification structured snapshot, WBS plan and resource snapshot, field event and evidence snapshot and are parsed into base table, cross-domain alignment is realized by object number, space number, task number, article number, risk number;Joint inference module sorts specification hard constraint, plan executable boundary and risk situation in rectification domain, and the executable rectification action package is combined and output from action template;Write-back audit closed loop module writes back action package to plan version record, field control account and work order / evidence account, and based on closed loop result trigger recalculation into next round, to realize " computable-executable-writeable-auditable " construction phase intelligent rectification closed loop.

Description

Technical Field

[0001] This invention relates to the field of engineering management technology, and more specifically, to a BIM-based intelligent management system for building engineering projects. Background Technology

[0002] As construction projects become larger and construction periods become shorter, the industry is gradually using BIM for engineering information integration, design verification and construction management, and is trying to introduce methods such as knowledge graphs and risk propagation models to improve automation and intelligence.

[0003] Existing publicly available literature 1 (Research on the Review Method of 3D Design Specifications for Power Transmission and Transformation Engineering Based on BIM and Knowledge Graph, Luo Quan, Master's Thesis, China University of Mining and Technology, 2024) achieves 3D design specification review and result visualization through "specification knowledge graph - knowledge base storage - BIMBase secondary development review", such as Figure 1 As shown, its core contribution lies in structuring the standard provisions and integrating them with the BIM review process, thereby improving the efficiency and consistency of model compliance checks. However, this solution is mainly geared towards "design compliance review," and its output is primarily "inspection results and visualization." On the one hand, it lacks linkage with real-time events, resource constraints, and plan disturbances during the construction phase; on the other hand, it lacks cross-phase risk transmission modeling and executable output for plan correction, resulting in "being able to check does not necessarily mean being able to modify, and being able to modify does not necessarily mean being able to write back in a closed loop."

[0004] Existing publicly available literature 2 (Research on Cross-Stage Risk Constraints in BIM Application of Smart Construction Sites, Shao Lijia, Master's Thesis, Wuhan University, 2024) analyzes the cross-stage transmission of risks in BIM application of smart construction sites based on a cascading failure model, such as... Figure 2 As shown, a multi-layered network is constructed and key risk transmission chains are extracted. At the same time, a network immunity constraint strategy is proposed. Its core contribution lies in structuring and quantifying the risk propagation process and identifying key risk chains. However, this solution is mainly aimed at the analysis of the "risk factor network level": its risk nodes lack a unified semantic anchor point with BIM objects, specification clauses, and planned tasks; its immunity strategy is difficult to directly translate into executable process adjustments, resource reallocation, and inspection work orders at the WBS level; in addition, there is a lack of a mechanism to audit and solidify the risk reasoning conclusions with BIM attributes and on-site evidence chains, resulting in "analysis is possible but it is difficult to form an auditable corrective closed loop".

[0005] Therefore, there is an urgent need for an intelligent management system that can achieve risk-standard-plan linkage reasoning and closed-loop correction under real engineering conditions. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, this invention provides a BIM-based intelligent management system for building construction projects. This system addresses the problems mentioned in the background art through a systematic mechanism that includes snapshot locking and basic table solidification, unified alignment of five types of business keys, four-map anchor point-residual joint reasoning, action package output and multi-carrier write-back auditing, and recalculation rolling updates.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A BIM-based intelligent management system for building construction projects includes a BIM access and parsing module, a specification structuring module, a risk assessment module, a planning and resource module, a joint reasoning module, and a write-back audit closed-loop module. The system operates in rounds, locking BIM object and spatial topology snapshots, specification clause structured snapshots, WBS planning and resource snapshots, and site event and evidence snapshots in each round and parsing them into a base table, aligning them based on object number, spatial number, task number, clause number, and risk number. The joint reasoning module constructs an object relationship diagram, a specification constraint diagram, a risk propagation diagram, and a planning and resource diagram, generating a system composed of five types of numbers. The anchor points are constructed and the correction domains are merged according to the anchor point sharing relationship; within the correction domain, the anchor point-residual simultaneous solution algorithm is executed to iteratively calculate the specification residual and the plan residual to obtain actionable solutions that satisfy the specification hard constraints and the plan executable boundary, and calculate the risk residual to form a risk situation. The actionable solutions are sorted and optimized, and the correction action package is output by combining action templates generated from task window adjustment, resource reallocation, spatial flow limiting and peak shifting, and specification inspection work order; the back-write audit closed loop module writes the correction action package back to the plan version record, the on-site control ledger, the specification inspection work order and the evidence audit ledger, and triggers recalculation to enter the next round based on the closed loop feedback.

[0008] As a further embodiment of the present invention, the basic tables include at least an object data table, a spatial topology table, a clause structured constraint table, a task and resource table, and an event evidence table, and the version number and number of records of each basic table are fixed in the round record.

[0009] As a further embodiment of the present invention, each record in the object data table includes at least an object number, object category, professional category, location identifier, parameter set, geometric reference, belonging space number and model version number, and each record in the spatial topology table includes at least a space number, space type, net area, boundary polygon, adjacent space set and entrance / exit set.

[0010] As a further aspect of the present invention, each record in the structured constraint table includes at least the clause number, applicable object category, applicable space type, judgment parameter list, threshold or allowable deviation, evidence list and rectification time limit, and the standard hard constraint uses the judgment parameters and thresholds of the structured constraint table as the judgment criteria.

[0011] As a further aspect of the present invention, the planning resource module generates a space occupancy prediction table and a shared resource conflict table, and generates a boundary quantum table containing the upper limit number of work groups, the predicted peak number of work groups, the space congestion index, the number of hours of shared resource conflict and the number of rework triggers. The executable boundary of the plan is determined by the boundary quantum table.

[0012] As a further embodiment of the present invention, the anchor point is composed of object number, space number, task number, clause number, and risk number. The anchor point is generated by chain completion, and the anchor point is merged by using the closure rule of shared business key to form a correction domain.

[0013] As a further aspect of the present invention, the risk situation module normalizes and weights the number of events, conflict hours, rework triggers and space congestion index based on a statistical window to obtain the observation intensity, which is used to update the propagation edge weights in the risk propagation graph and form the basis for risk situation ranking.

[0014] As a further embodiment of the present invention, the action template includes at least a task window adjustment template, a resource reconfiguration template, a space flow limiting and peak shaving template, and a standard inspection work order template. The correction action package includes at least an action package number, a round number, a set of affected objects, a set of affected spaces, a set of affected tasks, and a set of referenced clause numbers.

[0015] As a further aspect of the present invention, the evidence audit ledger records at least a set of evidence file indexes and an evidence summary verification code, the verification code being used to verify the consistency of the evidence files during the write-back and recalculation update process.

[0016] As a further aspect of the present invention, the write-back audit closed-loop module triggers a recalculation update when the plan version is switched, the work order is closed, or the space congestion alarm decreases, and updates the risk status and candidate action ranking with the recalculation results, and enters the next round of rolling solution.

[0017] The technical effects and advantages of the BIM-based intelligent management system for building construction projects of this invention are as follows: This invention locks in BIM object / space topology, structured specification clauses, WBS plan resources and on-site event evidence through round-by-round snapshots, and solidifies them into basic tables and version records, avoiding drift in terms of scope and basis, ensuring consistency in judgment and review within the same round, and realizing a construction management data base that is "calculable, recalculateable and traceable". This invention achieves cross-domain semantic alignment using object numbering, spatial numbering, task numbering, clause numbering, and risk numbering, constructs an object relationship diagram, a specification constraint diagram, a risk propagation diagram, and a plan resource diagram, and forms a correction domain with anchor point closures. Under the premise of satisfying the specification hard constraints and the plan executable boundary, it outputs a correction action package to solve the problem of "easy to check but difficult to modify" breakpoints. This invention writes the corrective action package back to the plan version record, on-site control ledger, standard inspection work order, and evidence audit ledger, and uses the evidence summary check code to solidify the consistency of the link; based on the work order closure and indicator changes, it triggers recalculation and rolling updates, so that the risks of congestion, conflict, rework and other risks can be quantified and reduced, forming an "executable-writeable-auditable" closed loop. Attached Figure Description

[0018] Figure 1 A system development framework diagram for existing technologies based on standardized knowledge graphs and BIM secondary development; Figure 2 A flowchart for extracting the key risk transmission chain based on the cascading failure model in existing technologies; Figure 3 This is a module diagram of a BIM-based intelligent management system for building construction projects according to the present invention; Figure 4 This is a flowchart of the anchor point-residual simultaneous solution algorithm of the present invention; Figure 5 This is a flowchart of the rolling correction-writeback-audit closed loop process of this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This example uses a real-world engineering project: a new construction project of a municipal-level general hospital with a total building area of ​​approximately 186,000 m², including outpatient and emergency medical technology buildings, inpatient buildings, infectious disease buildings, an energy center, and underground parking garages and connecting corridors. The implementation window was selected during the peak period from the 9th to the 13th month after construction began, with electromechanical installation, secondary structure work, equipment arrival, and special acceptance proceeding in parallel. Multiple disciplines were working simultaneously in spaces such as the third operating room, fire control room, and vertical shaft corridors. Model changes, plan switching, and the number of on-site work orders increased simultaneously. On-site management was prone to breakpoints where "compliance could be checked, risks could be calculated, and plans could be scheduled, but it was difficult to coordinate and output executable corrective actions and closed-loop feedback."

[0021] This invention's system employs a collaborative approach between a field server and archive nodes. The field server handles model analysis, constraint determination, risk status updates, simultaneous problem solving, and action package generation, while the archive nodes provide long-term storage for evidence documents and audit records. The system uses a rolling solution process: a full solution is executed daily at 08:10, with a BIM incremental synchronization cycle of 30 minutes. When the number of key object changes is ≥20 or the total number of object changes is ≥200, an additional solution is executed within 15 minutes, outputting the difference actions. The total number of object changes is the sum of the number of objects in the object data table whose fields were updated between two adjacent BIM incremental synchronizations. Field updates include object additions, deletions, and updates to object attributes, geometry, or relational fields. The total number of object changes characterizes the scale of input changes in this round. Each round generates a round number and fixes the trigger source and the start and end times of the statistical window. Action packages, plan versions, work orders, and audit ledgers are all traced using the round number to avoid data drift within rounds.

[0022] The number of critical object changes is determined by field-based criteria in the object data table. An object change refers to an update to a field used for determination in the object data table. An object is marked as a critical object and included in the critical object change count if it meets any of the following conditions: the object category belongs to critical equipment or components such as valves, fans, pumps, distribution boxes, main cable trays, or fire-resistant partition components; the object number appears in the deliverable object set of the task and resource table, and the corresponding task is in the critical path task set; the object is mapped to at least one mandatory clause in the current round of structured constraint table; the object's space number belongs to the hit space set in the current round of risk situation table, and the hit risk includes cross-conflict risk or acceptance rework risk. The system embeds the critical object determination source field in the round record, with source enumerations including "critical category," "critical path delivery," "mandatory clause hit," and "risk hit space," used for audit tracing of triggering reasons.

[0023] The system modules of this invention are as follows: Figure 3 As shown, the system includes a BIM access and parsing module, a standardized structure module, a risk assessment module, a planning and resource module, a joint reasoning module, and a write-back audit closed-loop module. The responsibilities, boundaries, inputs, and outputs of each module are clearly defined as follows: The BIM Access and Parsing module is used to access BIM model files or model interface data, parse and number engineering objects such as components, floors, rooms, work surfaces, and passageways, and generate object data tables, spatial topology tables, and object relationship diagrams. The Standardization and Structuring module is used to access applicable code clauses, enterprise standards, or project management rules, decomposing the clauses into machine-readable constraint tuples and forming a structured constraint table, which in turn generates a code constraint diagram. The Risk Situation module is used to access event evidence data such as sensor data, inspection data, quality and safety records, and work orders, forming an event evidence table, and combining the object relationship diagram and code constraint diagram to construct a risk propagation diagram, used to depict the propagation path and impact range of risks across objects and spaces. The Planning and Resource module is used to access construction schedule plans, task lists, and resource allocation data, forming a task and resource table and constructing a planning and resource diagram, used to describe the priority relationships of tasks, resource occupancy, and space occupancy status. The Joint Reasoning module is used to perform joint judgment and constraint solving based on the object relationship diagram, code constraint diagram, risk propagation diagram, and planning and resource diagram, generating corrective action packages and outputting candidate action combinations and their priorities. The write-back audit closed-loop module is used to write back the corrective action package to the BIM model attributes, planning system, work order system or audit ledger, record the action execution results and version changes, and initiate a recalculation when the trigger conditions are met to achieve continuous closed-loop optimization.

[0024] To ensure consistency in computational methods, the system of this invention completes input snapshot locking and base table solidification before entering joint inference. The specific process is as follows: When entering a single-round rolling solution, the system locks four types of input snapshots: BIM object and spatial topology snapshot, specification clause structured snapshot, WBS plan and resource snapshot, and site event and evidence snapshot, and parses them into object data table, spatial topology table, clause structured constraint table, task and resource table, and event evidence table. The system writes the base table version number, record number, and verification summary into the round record. The verification summary is used to identify whether the base table has been replaced or rearranged within the round.

[0025] The event evidence table uses events as the smallest unit and includes event number, event type, timestamp, associated space set, associated object set, associated task set, evidence file index set, evidence digest check code, and reporting source. The event types are fixed as five categories: space congestion alarm, cross-operation conflict, passageway occupancy and material stacking, finished product protection damage, and rework trigger.

[0026] In this embodiment, the irreversible digest is generated using the SHA-256 algorithm, with the input encoding in UTF-8. For the base table verification digest, the system generates a key-value string of "field name = field value" for each record according to the primary key field order. The primary key fields of the same record are connected by "|", and then the key-value strings of each record are connected by a newline character. This string is then sorted in ascending order by primary key and used as the digest input. For the evidence digest verification code, the system directly calculates the SHA-256 of the binary content of the evidence file and records the hexadecimal digest string. The field order, separator, and sorting rules are locked in the configuration, and the configuration version number is written to the round record to ensure consistency in verification across different nodes.

[0027] Subsequently, the system uniformly adopted five types of business keys—object ID, space ID, task ID, clause ID, and risk ID—to complete cross-table alignment. Mapping table solidification and boundary quantity calculation were both completed based on the solidification results of this round before proceeding. Figure 4 The four figures shown are used to construct and solve the system of equations.

[0028] The structured constraint table uses clauses as the smallest unit and includes at least the clause number, applicable object category, applicable space type, list of judgment parameters, unit, threshold or allowable deviation, list of evidence, rectification deadline, and severity weight. Each parameter in the list of judgment parameters includes the parameter name, unit, source type of value, and calculation method. The list of evidence must always include at least two items: measurement records and timestamped photographs. Some clauses may also include a supervisor's signed document or special acceptance record.

[0029] The Task and Resource Table uses tasks as the smallest unit and includes task number, planned start date, planned completion date, duration, set of preceding tasks, set of succeeding tasks, set of workspaces, set of deliverables, set of resource allocations, and minimum resource requirements. The deliverable set is established by the construction unit during the WBS preparation phase according to the "Task-Component / Equipment List" format. The list object numbers are consistent with the BIM object numbers, and are generated by merging the BIM component and equipment lists and validating them according to coding rules. When a task does not have a deliverable set, only weak workspace associations are allowed, and threshold fields must not be triggered to avoid accidentally triggering mandatory thresholds that could render the plan unexecutable.

[0030] This invention system performs dictionary normalization and solidifies the mapping table for object categories, space types, task types, and event types. The mapping between objects and tasks employs both strong and weak associations: an object's ID appearing in the task delivery object set is considered a strong association (0.95 confidence); an object's space ID appearing in the task's work space set is considered a weak association (0.70 confidence). If the same object has weak associations with multiple tasks, the primary mapping is selected based on: professional consistency (1.00), similar professional background (0.60), and unrelated professional background (0.20). Weak association scores <0.60 are only used for risk ranking and not for threshold fields to avoid accidental triggering that could render the plan unexecutable.

[0031] After the mapping table is solidified, the executability boundary quantity is calculated and written into the round boundary quantum table: the upper limit of the number of work groups is determined by the net area of ​​the space and the space type. One work group is allowed per 120m² in the clean confined area and one work group is allowed per 150m² in the ordinary corridor, rounded down.

[0032] The peak number of work groups is predicted using minute-level task window overlap prediction: an interval set is formed according to the start and completion of tasks, the number of tasks executed simultaneously per minute is calculated, and then the number of work groups working simultaneously is obtained by mapping according to the task work group configuration; when the number of work group members is not explicitly configured for a task, the default work group table for the job type is filled in and the source of the filling is recorded.

[0033] The space congestion index is the ratio of the predicted peak number of work groups to the maximum number of work groups in the space, with a hard boundary of 1.00. Shared resource conflict hours are calculated using the difference between the request window and the available calendar, with each consecutive 30 minutes counted as 0.5 hours, and conflicting task pairs and conflict time periods are recorded. Rework triggers are grouped into one trigger if the same object in the same space is within a 2-hour window.

[0034] The boundary quantum table includes at least the space number, the upper limit of the space number of work groups, the predicted peak number of work groups, the space congestion index, the shared resource number, the number of conflict hours, the number of rework triggers, and the start and end times of the statistical window, which serve as the benchmark for hard boundary pruning and recalculation.

[0035] Once the base table and boundary quantities are fixed within a round, the system enters an anchor-residual simultaneous solution algorithm based on object relationship diagrams, specification constraint diagrams, risk propagation diagrams, and plan resource diagrams, such as... Figure 4 As shown, the system of this invention constructs an object relationship diagram, a specification constraint diagram, a risk propagation diagram, and a plan resource diagram, using anchor point quintuples as unified alignment keys. Anchor points consist of object number, space number, task number, clause number, and risk number. Anchor point generation employs chain-like completion: when a specification hard constraint hits an object, an anchor point containing the object number, space number, and clause number is first formed. If a strong correlation mapping exists, the task number is completed; if a risk hits the space or task, the risk number is completed. Anchor point merging uses a closure rule; if any business key is shared, they are merged into the same correction domain. The solution is initiated when a mandatory clause within the correction domain is not closed or the risk situation reaches a threshold.

[0036] The edges in the object relationship graph are generated from the base table fields and weighted accordingly. Connecting edges: Two objects share the same system number and the distance between their endpoint control points is ≤0.30m. Crossing edges: The object's bounding box intersects the spatial boundary polygon, and the object's category is duct, pipe, or cable tray. Adjacent edges: Two spaces are adjacent in the adjacent space set, and each space contains at least one object to be corrected. Containing edges: The object's space number matches the space number. Belonging space edges: The object node points to the belonging edge of the space node. Edge weights range from 0 to 1, initially assigned according to edge type: connecting edges 0.90, crossing edges 0.80, adjacent edges 0.60, containing edges 0.70, belonging space edges 0.70. If an object has "critical equipment = yes" or is hit by a mandatory clause, the weights of its connecting and crossing edges are increased by 0.05 and truncated to 1.00. The above rules rely only on the object data table, spatial topology table, and geometric reference fields, and are bound to the round record to ensure reconstruction within the same round.

[0037] Within the correction domain, the anchor-residual simultaneous solution algorithm combines the determination of the specification residual, the plan executable residual, and the risk suppression residual, and uses hard boundaries to ensure the executableness of the output action. For each mandatory hard constraint term, the specification residual is calculated by taking the actual value x of the determination parameter and the threshold requirement x of the clause. n The definition is as follows: ; Where x comes from the object parameter set, geometric calculation, or measurement record, x n The data comes from structured fields in the clauses. The condition for zeroing the specification residual is that the object's compliance status is updated to closed loop and the evidence list meets the necessary requirements, or the parameters meet the threshold after rectification and the review is completed. The executable residual of the plan adopts a hard boundary caliber: the application of candidate actions must not disrupt the preceding and following relationships; resource reconfiguration must not cause any task's resource allocation to fall below the minimum resource limit; space occupancy prediction must not cause any space congestion index to remain greater than 1.00; the number of shared resource conflict hours must not decrease but increase, and the increase must not exceed 0.5 hours. The critical path prediction delay is calculated based on the current locked plan version as the baseline, and the difference between the current version and the critical path recalculation after the action is applied. The global threshold does not exceed 7 days, used to constrain the cascading effect and the diffusion of "non-critical to critical" events.

[0038] Risk suppression residuals are used for ranking. Observations are normalized and mapped to 0–1 and truncated: event counts are mapped to 0–5 times, conflict hours to 0–4 hours, rework triggers to 0–3 times, and congestion indices to 0.60–1.00. Exceeding the upper limit is counted as 1, and missing data is counted as 0 (missing data refers to no hit records within the statistical window or records that failed the minimum integrity check and were removed). The system aggregates observations according to the hit space set and hit task set configured by the propagation edge and obtains the observation intensity s according to the weights. An incremental update operator with smoothing is used to update the edge weights: the edge weights w from the previous round are updated... t 1. The edge weight w in this round t Press w t =clip(0.70×w t Update 1+0.30×s, 0, 1); for two consecutive rounds on the same side, take the moving average of s from two rounds and substitute it; if there is no hit in this round, record s=0 and allow it to decay naturally. Update the operator coefficients and the moving average window length and write them into the round record to ensure consistency in verification. The strength of the critical risk chain is calculated based on the continuous product of the weights of the edges on the chain. After the candidate action is applied, it is recalculated according to the same method and the decrease is compared. It only participates in the ranking after the hard boundaries of the specification and plan are met. Therefore, the risk suppression residual and the strength of the critical risk chain are used as the basis for the risk side ranking of candidate action combinations; the hard constraints of the mandatory provisions of the specification and the executable boundaries of the plan are satisfied as hard boundaries in the candidate generation and pre-screening stages to avoid bringing unexecutable combinations into the ranking calculation.

[0039] The anchor-residual simultaneous solution algorithm uses action template projection to generate candidate action combinations. The task window adjustment template has a step size of 1 day, with a maximum backward step of 3 days for critical path tasks and a maximum backward step of 7 days and a maximum forward step of 3 days for non-critical tasks. The resource reallocation template has a labor adjustment step size of 2 people / time, with a maximum of 6 people per time and not lower than the minimum lower limit; shared resources have a step size of 0.5-hour shifts. The spatial flow limiting and peak-shifting template has a maximum number of simultaneous work groups within a 2-hour time window that is an integer, and the written value does not exceed the upper limit of the number of work groups in the space. The standard inspection work order template requires the evidence list to include at least measurement records and timestamped photos, with deadlines of 12 hours, 24 hours, and 48 hours. In this embodiment, the mandatory clause defaults to the 24-hour timeframe. If the measured value and the model value are inconsistent, a difference explanation must be filled out and a review record uploaded; otherwise, the loop cannot be closed. The "Space Flow Restriction and Peak Shifting Template - Overlay Restriction Rules" stipulate that hot work and hoisting operations should not be overlapped within the same 2-hour time window in the same space. If they must be overlapped due to the indivisible nature of the work process, it will be considered a violation and trigger a level 2 alarm. Simultaneously, the maximum number of parallel work shifts allowed within that time window will be reduced by the minimum (original maximum). 1,3) Downward adjustment is used to drive the generation of peak shifting, window adjustment, or resource reallocation actions; the hit flag and downward adjustment result of this rule are written into the action package and ledger write-back fields. It should be noted that the step size, maximum adjustment range, upper limit of candidate combination, and scoring weight in this embodiment are exemplary values, and the project can be configured and adjusted according to project management requirements; to ensure verifiable consistency, the actual values ​​are fixed with each round of recording.

[0040] The system performs hierarchical enumeration for each correction domain, limiting the upper limit of candidate combinations to 500 groups. It then fixes the work order and threshold fields, the flow control and peak-shaving fields, and enumerates window adjustments of ±1 day, ±2 days, and resource reallocation of ±2 people and ±4 people. After each enumeration, a boundary quantum table is used for rapid pre-screening. When the number of groups exceeds 500, the system truncates and retains groups based on the combined change in congestion index and conflict hours. The pruning and filtering order is fixed: groups lacking mandatory work orders and threshold fields are removed; groups disrupting pre- or post-critical path prediction delays exceeding 7 days are removed; groups with resources below the minimum lower limit are removed; groups with congestion indices greater than 1.00 in any space are removed; groups with shared resource conflict hours increasing by more than 0.5 hours are removed. Filtered combinations are then sorted, and action packages record fixed scores: risk suppression weight 0.55, plan disturbance weight 0.25, and compliance closed-loop advancement weight 0.20. The planned disturbance indicator is the sum of the critical path delay increment and the absolute value of the total task window adjustment days, and then normalized to one. The compliance closed-loop progress indicator is scored at the field level based on whether the work order corresponding to the mandatory clause is generated, whether the necessary evidence items are complete, and whether the threshold field is written and can be judged. If all three are satisfied, a score of 1 is given; otherwise, the value is taken according to the satisfaction ratio. The highest score is output as the correction action package, and an action package number and plan version number are generated.

[0041] This embodiment uses the 268th day after commencement of work as a typical cycle, cycle number SOL26801, model version number V2683, plan version number PV26703, and clause version number NORM202501. The statistical window is from 08:10 on the 267th day to 08:10 on the 268th day. The object change is concentrated in the smoke exhaust system of the third zone of the operating room. Clause matching generates 47 sets of clauses and object associations, including 12 sets of mandatory clauses. Two non-closed-loop mandatory hard constraints are located to two smoke exhaust fire damper objects. The judgment parameters are F, H, and L. The thresholds are F≥1.50 hours, H allowable deviation ±0.010m, and L≥0.300m. The evidence list includes measurement records, timestamped photos, and supervisor's signature sheet. The rectification deadline is 24 hours. The net space area is approximately 520 square meters. The upper limit for the number of work groups is rounded down to 4 based on 120 square meters per work group. The peak number of work groups predicted at the minute granularity is 6, and the congestion index is 1.50. The shared resource tower crane conflict hours are 3.6 hours, with 3 cross-conflict events and 2 rework triggers. According to the hit rules, the cross-conflict risk and the plan disturbance risk are hit. After the anchor point chain is completed, the anchor points related to the same space and the same task are merged into the same correction domain according to the closure rule and entered into the solution.

[0042] The simultaneous solution outputs the correction action package AP26802 and generates the plan version PV26801: (1) Task window adjustment: The start time of the smoke exhaust branch pipe installation task is changed from 08:00 on the 270th day to 08:00 on the 272nd day, and the construction period remains unchanged at 3 days; (2) Compliance closed-loop promotion: The threshold field is written for the concealed acceptance before sealing the plate. The threshold is that the work order IW268021 and IW268022 are allowed to start after they are both closed; At the same time, a standard inspection work order is generated, and the evidence list includes measurement records, timestamp photos and supervisor's signature sheet, and the rectification time limit is 2 days. 4 hours; (3) Resource reallocation: The number of cable tray installation workers will be adjusted from 3 to 4 (to meet the minimum lower limit), and shared resources will be adjusted in 0.5-hour shift increments; (4) Space flow restriction and peak staggering: The maximum number of simultaneous work groups during the day shift is 4, the time period is from 08:00 to 18:00, and the non-cutting night shift window is opened from 20:00 to 02:00; the superimposed restrictions are executed according to the "Space Flow Restriction and Peak Staggering Template - Superimposed Restriction Rules", and the action package records whether it hits, hits the 2-hour time window and the lowered upper limit value, which are used for on-site execution and audit traceability. The action package also carries the set of referenced clause numbers, the set of referenced risk numbers, and the set of affected objects, spaces, and tasks to ensure that the "actions and basis" are traceable.

[0043] After the correction action package is generated, press Figure 5 Entering the write-back and audit closed loop. The system maintains the write-back field mapping table and fixes its version number in the round record. The minimum field set for plan write-back includes task number, start time, completion time, resource configuration difference, threshold field, plan version number, round number, and action package number; the minimum field set for work order write-back includes work order number, clause number, object number, space number, inspection parameter list, threshold, evidence list, deadline, closed loop status, evidence index set, and evidence summary check code; the minimum field set for ledger write-back includes space number, maximum number of shifts, time window, superimposed restriction rules, alarm threshold, and effective round number. After write-back, plan version records, on-site control ledgers, standard inspection work orders, and evidence audit ledgers are generated, and the supervisor reviews and updates the work order closed loop status and object compliance status. The threshold field adopts a full satisfaction judgment. Only when all work orders listed in the threshold are closed and the required evidence items are valid, the task status changes from not startable to startable and the release time and basis list are written; otherwise, the threshold remains unchanged.

[0044] The system triggers a recalculation update when a work order closes in, a plan version is switched, or the space congestion alarm significantly decreases. Within the same statistical window, the space congestion index, shared resource conflict hours, and rework trigger count are recalculated according to a unified statistical standard. The propagation edge weights and critical risk chain strengths are updated according to the same aggregation standard. The "significant decrease" refers to a decrease in the space congestion index of ≥0.20 or a decrease in shared resource conflict hours of ≥0.5h compared to the previous round. The threshold is written into the round record for consistency verification. The round comparison record includes at least the boundary quantity difference, work order closure difference, critical path prediction delay difference, and risk chain strength difference and is written into the audit ledger. The recalculation comparison shows that: the predicted peak number of shifts in the third operating room decreased from 6 to 4, and the congestion index dropped from 1.50 to 1.00; the number of hours of shared resource conflict decreased; two mandatory clause work orders were closed within 24 hours, and the object compliance status was written back to closed and the threshold field was removed; the critical path prediction delay converged from 6.4 days to 2.1 days. All indicators can be traced back in the audit log through round number, action package number, plan version number and work order number, forming a closed-loop operation mechanism of "input locking - simultaneous solution - action writing back - evidence audit - recalculation update".

[0045] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0046] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A BIM-based intelligent management system for building construction projects, comprising a BIM access and parsing module, a standardized structured module, a risk assessment module, a planning and resource module, a joint reasoning module, and a write-back audit closed-loop module; characterized in that, The system operates in rounds, locking BIM object and spatial topology snapshots, specification clause structured snapshots, WBS plan and resource snapshots, and on-site event and evidence snapshots in each round and parsing them into a basic table, which is then aligned based on object number, spatial number, task number, clause number, and risk number. The joint reasoning module constructs an object relationship graph, a normative constraint graph, a risk propagation graph, and a plan resource graph, generating anchor points composed of five types of numbers and merging correction domains according to anchor point sharing relationships. Within the correction domain, the anchor point-residual simultaneous solution algorithm is executed to iteratively calculate normative residuals and plan residuals to obtain actionable solutions that satisfy normative hard constraints and plan executable boundaries. Risk residuals are calculated to form a risk profile, and the actionable solutions are sorted and optimized. Correction action packages are output by combining action templates from task window adjustment, resource reallocation, spatial flow limiting and peak shifting, and normative inspection work order generation. The write-back audit closed-loop module writes the correction action packages back to the plan version record, on-site control ledger, normative inspection work order, and evidence audit ledger, and triggers recalculation to enter the next round based on closed-loop feedback.

2. The BIM-based intelligent management system for building construction projects according to claim 1, characterized in that, The basic tables include at least an object data table, a spatial topology table, a clause structured constraint table, a task and resource table, and an event evidence table, and the version number and number of records of each basic table are fixed in the round record.

3. The BIM-based intelligent management system for building construction projects according to claim 2, characterized in that, Each record in the object data table contains at least the object number, object category, professional category, location identifier, parameter set, geometric reference, belonging space number, and model version number. Each record in the spatial topology table contains at least the space number, space type, net area, boundary polygon, adjacent space set, and entrance / exit set.

4. The BIM-based intelligent management system for building construction projects according to claim 2, characterized in that, Each record in the structured constraint table of the provisions shall include at least the provision number, the category of applicable objects, the type of applicable space, the list of judgment parameters, the threshold or allowable deviation, the list of evidence and the rectification time limit. The hard constraints of the norm shall use the judgment parameters and thresholds of the structured constraint table of the provisions as the judgment criteria.

5. The intelligent management system for building construction projects based on BIM according to claim 1, characterized in that, The planning resource module generates a space occupancy prediction table and a shared resource conflict table, and generates a boundary quantum table containing the upper limit number of work groups, the predicted peak number of work groups, the space congestion index, the number of hours of shared resource conflicts, and the number of rework triggers. The executable boundary of the plan is determined by the boundary quantum table.

6. The BIM-based intelligent management system for building construction projects according to claim 1, characterized in that, The anchor point consists of object number, space number, task number, clause number, and risk number. Anchor point generation adopts chain completion, and anchor point merging adopts the closure rule of shared business key to form the correction domain.

7. The BIM-based intelligent management system for building construction projects according to claim 1, characterized in that, The risk situation module normalizes and weights the number of events, conflict hours, rework triggers and space congestion index based on a statistical window to obtain the observation intensity, which is used to update the propagation edge weights in the risk propagation graph and form the basis for risk situation ranking.

8. The BIM-based intelligent management system for building construction projects according to claim 1, characterized in that, The action templates include at least a task window adjustment template, a resource reconfiguration template, a space flow limiting and peak shifting template, and a standard inspection work order template. The corrective action package includes at least an action package number, a round number, a set of affected objects, a set of affected spaces, a set of affected tasks, and a set of referenced clause numbers.

9. The intelligent management system for building construction projects based on BIM according to claim 1, characterized in that, The evidence audit ledger records at least the evidence file index set and the evidence summary verification code. The verification code is used to verify the consistency of the evidence files during the write-back and recalculation update process.

10. The BIM-based intelligent management system for building construction projects according to claim 1, characterized in that, The write-back audit closed-loop module triggers a recalculation update when the plan version is switched, the work order is closed, or the space congestion alarm decreases, and updates the risk status and candidate action ranking with the recalculation results to enter the next round of rolling solution.

Images