A BIM-based method for managing construction progress information in building projects

Abstract

This invention discloses a BIM-based method for managing construction progress information in building engineering, belonging to the field of computer information management technology. This invention utilizes a pre-trained language model to semantically parse the natural language reports of construction personnel, abstracting information such as floors, areas, component types, and construction actions into component description elements. These elements are then combined with a component graph containing type attributes, spatial adjacency relationships, and construction sequence dependencies. This constrains the matching range at both the semantic space and topological structure levels, significantly improving the accuracy of component location under fuzzy spatial representations. Through multi-channel semantic similarity calculation and a confidence-driven weight adaptive mechanism, the system can automatically adjust the contribution of channels such as type, action, graph structure, contextual text, and discrete labels based on different corpus quality.

Description

Technical Field

[0001] This invention relates to the field of computer information management technology, and in particular to a BIM-based method for managing construction progress information of building projects. Background Technology

[0002] In construction progress information management, some complex indoor construction scenarios involve multiple floors, areas, and various component types. Project managers need to update digital models and schedules based on the progress status reported regularly by construction workers. Currently, Building Information Modeling (BIM) is widely used as a carrier for construction information management, requiring that progress data accurately correspond to components in the model, such as through unique identifiers or spatial coordinates. However, in large-scale indoor projects, frontline construction workers often use natural language to describe the construction situation, such as "the east wall on the second floor of Area A has been plastered." Such descriptions often contain non-standard spatial terms and relative position information, which are difficult to directly map into the structured data fields of the BIM model. This results in progress information entry relying on manual identification and secondary processing, which can easily lead to entry delays or matching errors, thus affecting the timeliness and reliability of overall progress management.

[0003] To reduce the workload of manual data processing, several auxiliary solutions based on BIM and natural language processing have been proposed in existing technologies. For example, some systems use speech recognition or text analysis to obtain oral or text reports from construction workers, and use pre-trained language models or rule engines to parse elements such as construction areas, floors, and component types, and then retrieve candidate components from the BIM database to generate model update instructions. Some solutions also improve the mapping efficiency from natural language to component objects by associating common vague descriptions such as "east side" and "near the window" with preset coordinate ranges. However, these solutions are mostly based on general corpora for training and lack targeted modeling for the professional context and complex spatial relationships in the field of architectural engineering. When multiple components share similar descriptions or construction areas overlap, the system still has difficulty accurately distinguishing the target components, often requiring manual verification and correction. In addition, natural language processing models are sensitive to context and expression habits. When new terms, abbreviations, or dialects frequently appear in a dynamic construction environment, the parsing results are prone to instability, which in turn leads to inconsistencies in the automatically generated progress update instructions.

[0004] In summary, existing BIM-based construction progress information management still has the following shortcomings when dealing with natural language reports in complex indoor construction scenarios: the mapping efficiency between natural language reports from construction personnel and BIM structured data is low, relying on manual judgment, which can easily cause delays in progress information entry; the ability to understand spatial semantics and fuzzy descriptions in the construction field is limited, resulting in low component matching accuracy and affecting the authenticity and traceability of progress information; the progress information update process lacks stable and reliable automated support, making it difficult to reflect the real progress of the construction site in a timely and accurate manner, leaving management loopholes.

[0005] Therefore, it is necessary to provide a BIM-based method for managing construction progress information to better handle the natural language reports from construction personnel, achieve efficient and accurate association between progress information and BIM model components, and improve the real-time performance and reliability of construction progress management. Summary of the Invention

[0006] In view of the aforementioned existing problems, the present invention is proposed.

[0007] This invention provides a BIM-based method for managing construction progress information in building projects. This addresses the problems in existing technologies, such as construction workers in complex indoor construction sites often using natural language to report progress, vague descriptions that are difficult to automatically map to BIM components, progress entry relying on manual judgment, and delayed and error-prone updates.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] This invention provides a BIM-based method for managing construction progress information in building projects, executed by computer equipment, and includes the following steps:

[0010] Step S1: Receive natural language input of construction progress report submitted by construction personnel. The natural language input includes at least fuzzy description information of floor, area and component type and corresponding progress status.

[0011] Step S2: Perform semantic parsing on the natural language input to extract at least one component description element and the progress status;

[0012] Step S3: Based on the component description element, retrieve candidate component information from the BIM database associated with the BIM model. The BIM database stores component identifier, component type, spatial location, and construction progress data for each component.

[0013] Step S4: Based on the pre-established semantic relationship model of components, calculate the matching degree of the candidate component information to generate a candidate component list;

[0014] Step S5: Adjust the weights of the candidate components in the candidate component list based on construction history data, and determine the target component from the candidate component list;

[0015] Step S6: Based on the progress status, update the construction progress information of the target component in the BIM model, and display the update result in the construction progress management interface.

[0016] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the component semantic relationship model is realized by constructing a component atlas, which includes:

[0017] The type attributes of components, the spatial adjacency relationships between components, and the construction sequence dependencies between component construction procedures.

[0018] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the semantic parsing step includes:

[0019] A pre-trained language model is used to identify spatial terms and construction action terms in the natural language input, and the identified spatial terms and construction action terms are mapped to nodes and edges in the component graph to obtain the component description elements.

[0020] As a preferred embodiment of the BIM-based construction progress information management method of the present invention, the matching degree calculation includes:

[0021] Based on the semantic similarity between the component description element and the nodes in the component graph, the candidate components are semantically matched, and the candidate components are filtered by combining the spatial coordinate range of the candidate components. The semantic similarity is obtained by a preset semantic similarity calculation rule, and the spatial coordinate range is filtered by a preset spatial distance calculation rule.

[0022] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the weight adjustment step includes:

[0023] The priority of each candidate component is dynamically adjusted based on the update frequency of the candidate component in the construction history data and the most recent operation time, wherein the priority is dynamically calculated based on the time decay factor and frequency weighting rule.

[0024] In a preferred embodiment of the BIM-based construction progress information management method of the present invention, in the retrieval step, if the component description element corresponds to multiple candidate components, then based on the construction progress sequence data associated with the candidate components, the candidate components in the active construction stage are selected as priority candidate objects to determine the target component.

[0025] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the update step includes:

[0026] The progress status is compared with the planned progress stored in the BIM model to obtain progress deviation information, and the target components with progress deviations are automatically marked in the construction progress management interface.

[0027] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the method further includes:

[0028] After receiving the natural language input, multimodal data associated with the target component is acquired, the multimodal data including at least one type of on-site image data and sensor data;

[0029] The reliability of the progress status is verified by cross-checking the multimodal data with the component status in the BIM model.

[0030] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the method further includes:

[0031] Based on the updated construction progress information, the differences between the planned and actual progress of each floor, construction area, and component type are statistically analyzed, and a construction progress management report including a timeline view and a Gantt chart view is generated and presented in the construction progress management interface.

[0032] As a preferred embodiment of the BIM-based construction progress information management method for building projects described in this invention, the method further includes:

[0033] Based on the construction progress information and the progress deviation information, adjustment suggestions for subsequent construction tasks are automatically generated. The adjustment suggestions include at least suggestions for adjusting the start and end times of subsequent construction tasks, suggestions for construction team configuration, and suggestions for adjusting material arrival times. These suggestions are then confirmed and implemented by project management personnel in the construction progress management interface.

[0034] The beneficial effects of this invention are as follows: By constructing an end-to-end information processing link from natural language reporting to BIM component progress updates, this invention specifically improves this problem at multiple stages: This invention utilizes a pre-trained language model to semantically parse the natural language reports of construction personnel, abstracting information such as floors, areas, component types, and construction actions into component description elements. These elements are then combined with a component graph containing type attributes, spatial adjacency relationships, and construction sequence dependencies. This constrains the matching range at both the semantic space and topological structure levels, significantly improving the accuracy of component location under fuzzy spatial representations. Through multi-channel semantic similarity calculation and a confidence-driven weight adaptive mechanism, the system can automatically adjust the contribution of channels such as type, action, graph structure, contextual text, and discrete labels according to different corpus quality, thus maintaining relatively stable matching performance even in cases of dialects, colloquialisms, and incomplete expressions. Furthermore, this invention introduces construction history logs, performing time attenuation and frequency weighted fusion on candidate components based on their most recent operation time and update frequency. This prioritizes recent and high-frequency construction areas, making the selection of target components closer to the actual on-site construction status and reducing misjudgments and manual verification workload in scenarios with multiple components having the same name or overlapping areas. At the progress information application level, this invention automatically compares the updated actual progress with the planned progress in the BIM and marks deviations. Simultaneously, it generates timeline views and Gantt chart views by floor, region, and component type, centrally displaying the overall and local progress deviation distribution. This facilitates managers in quickly identifying critical paths and risk points, and adjusting construction sequences and resource allocation in a timely manner. In addition, by introducing on-site images and sensor data for cross-verification of natural language reports, this invention improves the reliability and traceability of progress data even in complex on-site environments and with diverse expression habits. Overall, it reduces manual data entry and communication costs, enhances the automation, real-time performance, and accuracy of construction progress information management, and effectively compensates for the shortcomings of existing technologies in managing progress in complex and ambiguous reporting scenarios. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0036] Figure 1 This is a flowchart illustrating the BIM-based construction progress information management method in the embodiments. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0038] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0039] For example, the terms “first” and “second” used in this application are only used to distinguish and describe similar objects, to differentiate the first object from another object, and are not used to describe a specific order or sequence, nor should they be interpreted as indicating or implying relative importance.

[0040] This application proposes a BIM-based method for managing construction progress information in building projects, executed by computer equipment, and combined with... Figure 1 As shown, it includes the following steps:

[0041] Step S1: Receive natural language input of construction progress report submitted by construction personnel. The natural language input includes at least fuzzy description information of floor, area and component type and corresponding progress status.

[0042] Step S2: Perform semantic parsing on the natural language input to extract at least one component description element and progress status;

[0043] Step S3: Based on the component description elements, retrieve candidate component information from the BIM database associated with the BIM model. The BIM database stores component identifier, component type, spatial location, and construction progress data for each component.

[0044] Step S4: Based on the pre-established semantic relationship model of components, calculate the matching degree of the candidate component information and generate a list of candidate components;

[0045] Step S5: Adjust the weights of the candidate components in the candidate component list based on historical construction data, and determine the target component from the candidate component list;

[0046] Step S6: Update the construction progress information of the target component in the BIM model according to the progress status, and display the update results in the construction progress management interface.

[0047] In this embodiment, natural language input is preferably string data in text form. Construction personnel can directly type in the report content on the progress reporting interface of a mobile terminal or desktop terminal, or transcribe the oral report into text using existing speech recognition services and submit it. The system only receives natural language input that has been structured into text and does not process the speech signal itself. The construction progress status can be configured as a limited discrete level or percentage range in the project implementation. For example, the statuses of not started construction, completed rework, etc., can be mapped to preset labels or percentage values. The progress status is stored in the database in a one-to-one correspondence with the corresponding component identifier, so that it can be directly used for comparison with the planned progress in subsequent steps. The BIM database can directly use the data tables exported from the existing BIM model of the project. The component identifier, component type, spatial location, and planned progress fields are imported into the management system through the data interface. The spatial location field includes coordinate information for spatial distance calculation, and the planned progress field includes information such as the planned start time and planned completion time. Specifically, during project deployment, the system allows setting the recommended submission frequency for natural language reports based on project management requirements. The reporting cycle can be set between 10 minutes and 1 hour, and the interface prompts construction personnel to submit reports according to their shift schedule. The refresh cycle of the construction progress management interface can be set to minutes or shifts to balance real-time performance and system load. Optionally, if a natural language report fails to identify a unique target component after subsequent processing steps, or if its semantic similarity and priority are both below a preset threshold, the system marks the report as requiring manual confirmation. This is displayed in the construction progress management interface as a prompt or a dedicated list, and the progress information in the BIM model is not modified temporarily to ensure the consistency and traceability of the model data. Similarly, when a component identifier matching the natural language description cannot be found in the BIM database due to modeling omissions or synchronization failures, the system records this anomaly and allows the re-triggering of steps S1 to S6 after subsequent data synchronization or model completion, preventing information loss due to update failures.

[0048] In one embodiment, the component semantic relationship model is implemented by constructing a component graph, which includes:

[0049] The type attributes of components, the spatial adjacency relationships between components, and the construction sequence dependencies between component construction procedures;

[0050] In one embodiment, the semantic parsing step includes:

[0051] A pre-trained language model is used to identify spatial terms and construction action terms in natural language input, and the identified spatial terms and construction action terms are mapped to nodes and edges in the component graph to obtain component description elements.

[0052] In one embodiment, the matching degree calculation includes:

[0053] Based on the semantic similarity between the component description elements and the nodes in the component graph, the candidate components are semantically matched, and the candidate components are filtered by combining the spatial coordinate range of the candidate components. The semantic similarity is obtained by a preset semantic similarity calculation rule, and the spatial coordinate range is filtered by a preset spatial distance calculation rule.

[0054] The rules for calculating semantic similarity are defined as follows:

[0055] Step S41: Map the component description elements obtained from semantic parsing into multi-channel vectors, and calculate the channel similarity for each candidate component. The channel results are then uniformly normalized. :

[0056]

[0057]

[0058]

[0059] in, Indicates the candidate component index. Represents the type vector of the query. Indicates candidate type vector, This represents the construction action vector being queried. Indicates candidate Construction action vector, This indicates a query for the structural vector on the component map. Indicates candidate The structure vector, The text context vector representing the query. Indicates candidate The text context vector, Represents the vector dot product. Represents the L2 norm, Represents a linear mapping function. The kernel width hyperparameter represents the text channel, and its value is... , This indicates a query for a set of discrete labels. Indicates candidate A set of discrete labels These represent the similarity of the five channels, with values ​​ranging from [value 1] to [value 2]. ;

[0060] Step S42: Perform weight adaptation based on channel confidence to obtain the total semantic similarity and maintain it. scope:

[0061]

[0062] in, Indicates candidate Total semantic similarity Represents a set of channels. Indicates candidate In the passage similarity, Indicates channel Normalized weights, taking values and , Indicates channel The confidence level, with values ​​of , Indicates channel The scaling factor, with values... , For sets The summation index;

[0063] Step S43: Define the confidence level using the channel average of the feature recognition probability, which is used to drive the weights.

[0064]

[0065] in, Indicates channel The number of valid elements is a positive integer. Indicates channel No. The probability of recognizing each element, with values ​​ranging from 1 to 2. , The value of ;

[0066] Step S44: Threshold-based filtering and sorting of the total semantic similarity results to form a set for spatial coordinate filtering.

[0067]

[0068] in, This represents the set of candidate indices selected through semantic filtering. This represents the semantic similarity threshold, with values ​​ranging from [value 1] to [value 2]. ;

[0069] Input features are derived from a standard component type table, a construction action vocabulary, node / adjacency embeddings of component diagrams, and sentence vectors, and are processed before online computation. Normalization and tag inverted index retrieval, , , Criteria were determined using the project's historical data set.

[0070] Specifically, the above implementation decomposes semantic matching into multi-channel metrics and uses similarity output within a unified interval as a fusion premise to avoid offset caused by cross-dimensional superposition. Three channels—type, action, and graph structure—provide stable characterizations of component categories, process descriptions, and topological relationships. The text channel absorbs contextual clues to address differences in natural language expressions, and the tag channel supplements the overlap of discrete elements. The fusion stage employs soft-normalized weights, mapping channel confidence to contribution. When the input is more complete in a certain type of clue, the influence of that channel is automatically increased, thereby reducing interference from missing descriptions or noisy words. Threshold filtering is connected to subsequent spatial range filtering and progress data verification, providing an adjustable recall / precision balance. The implementation path is parameter-calibrated around the project corpus, accelerating retrieval through near-nearest neighbor analysis and maintaining response speed through online lightweight fusion.

[0071] Specifically, the component description elements obtained from semantic parsing are organized into vector representations across multiple channels in engineering implementation. The type channel can be constructed based on the encoding or pre-trained embedding vectors of each component type in the standard component type table. The construction action channel can be obtained based on the construction action vocabulary and its corresponding embedding representation. The graph structure channel can generate structural vectors based on nodes in the component graph and their adjacency relationships. The text channel can use context vectors obtained by encoding the entire natural language input sentence. The label channel is directly generated based on a set of discrete labels related to the component. The vector dimension is determined by the pre-trained model or project configuration and can be set in the range of tens to hundreds of dimensions to achieve a balance between expressive power and computational overhead. Input features are normalized when written to vector storage to ensure that the similarity outputs of different channels all fall within the range of zero to one. In actual deployment, the semantic similarity threshold can be initially selected according to the suggested range, for example, a midpoint close to 0.7, and then adjusted based on the manual annotation results of historical project samples to ensure that the candidate set selected through semantic screening achieves a trade-off between recall and false match rate that meets management requirements. Optionally, when a channel lacks effective elements in the current query, such as the absence of construction action terms or available tags in the natural language input, the system treats this channel as having zero or near-zero confidence. During weight normalization, its weight is automatically reduced, making the contribution of this channel's similarity to the total semantic similarity approach zero, thus avoiding interference from invalid features in the matching results. Furthermore, when multiple channels generally have low confidence due to short or incomplete text, the system can expand the candidate set by lowering the semantic similarity threshold and further filter based on subsequent spatial distance calculation rules and progress data verification, thereby improving the recall rate of hard samples without significantly increasing the risk of mismatches. In the implementation of approximate nearest neighbor retrieval, an existing vector index library can be optionally used to segment or jointly index multi-channel vectors. By pre-constructing the index structure, the number of online similarity calculations is reduced, thus maintaining a low response time even in multi-project concurrent scenarios.

[0072] In one embodiment, the weight adjustment step includes:

[0073] The priority of each candidate component is dynamically adjusted based on the update frequency of the candidate component in the construction history data and the most recent operation time. The priority is dynamically calculated based on the time decay factor and frequency weighting rule.

[0074] Priority is dynamically calculated based on time decay factor and frequency weighting rule, and the steps include:

[0075] Step S51, when candidate components The time of the most recent operation was Moment, at the present moment Calculate the time interval and give the decay in the form of half-life:

[0076]

[0077] in, Indicates candidate The time difference from the most recent operation to the present, in hours. This indicates the current time, in hours. Indicates candidate The time of the most recent operation, in hours. Represents the time decay factor, with a range of values. , This represents the half-life parameter, in hours, with a recommended range of values. ;

[0078] Step S52, in a length of The number of updates to candidate components is counted within a sliding time window, and a saturation weight is used to suppress the over-amplification of frequent components.

[0079]

[0080] in, This represents the frequency weighting factor, with a range of values. , Indicates candidate In the time window The number of updates within, in seconds. This indicates the length of the sliding time window, in hours. A recommended value range is... , This represents the frequency saturation coefficient, and its value range is... Dimensionless, recommended value is ;

[0081] Step S53: To ensure that both recentity and frequency work together and that the overall priority is suppressed when a single item is too low, a weighted geometric average is used to obtain the original priority:

[0082]

[0083] in, Indicates candidate The original priority and its range of values , This represents the proximity weighting coefficient, with a range of values. Dimensionless;

[0084] Step S54: Incorporate historical priorities into exponential smoothing to obtain the current smoothing priority:

[0085]

[0086] in, Indicates candidate Smoothing priority, value range , Indicates candidate The previous time step smoothing priority, value range , This represents the smoothing coefficient, and its value range is... Dimensionless, recommended to take ;

[0087] Step S55: To facilitate comparability between different batches, perform cross-candidate normalization on the smoothing priority and provide a selection set:

[0088]

[0089] in, Indicates candidate Normalization priority, range of values , Indicates a candidate index, dimensionless. To represent a small constant to prevent the denominator from being zero, it is recommended to take... , This indicates the set of candidates to proceed to the next step of the decision or to be selected directly. This indicates the priority threshold, and its range. ;

[0090] The actual calculation process includes: inputting construction history logs → calculation. and →Statistics and obtain → Synthesis →Exponential smoothing → Normalization And according to Sort / filter → Output target component candidates for determination in step S5;

[0091] Specifically, the above implementation uses a half-life function to characterize the impact of the most recent operation on priority, making the time interval decay exponentially, which is convenient for intuitive calibration through half-life; on the frequency side, a saturated fractional weight is used to map the number of updates to a monotonically increasing interval with a limited upper limit, reducing the crowding out of the ranking by extremely high-frequency samples; the two are geometrically weighted and fused, so that if any one of them is too low, the overall value will be reduced, which is more in line with the management intention of needing both recent activity and historical frequency; online smoothing is used to suppress short-term fluctuations and improve the stability of priority curves in reports and interfaces; cross-candidate normalization and threshold provide the system with a portable scoring scale, which is convenient for unified scheduling strategies under different projects and different time windows; in terms of parameter selection, half-life, time window and saturation coefficient can all be calibrated by the project's historical data grid, and when the construction rhythm changes, only a small amount of scalar needs to be adjusted to maintain the availability of the ranking;

[0092] Furthermore, historical construction data in engineering practice originates from update logs automatically recorded by the progress management system. Each log entry contains at least a component identifier, operation time, and the updated progress status. During the offline phase, the system aggregates logs by component identifier to generate a time series for calculating the most recent operation time and the number of updates within a time window. The half-life parameter can be determined by analyzing the typical cycles of construction pace under different project types. For example, for fast-paced interior decoration projects, a half-life of approximately one to two days can be selected; for slower-paced main structure projects, a half-life of several days to a week can be selected, ensuring that the time decay factor changes on a time scale consistent with on-site intuition. The time window length can be determined based on the project's planned cycle and shift rotation cycle, and can be set within a range of one to two weeks, ensuring that frequency statistics reflect recent activity levels without excessively amplifying short-term fluctuations. The frequency saturation coefficient can be selected by analyzing the distribution of high-frequency components in historical logs, ensuring that the frequency weighting factors of most components are within a clearly distinguishable range without premature saturation. Optionally, in the early stages of a project or when the amount of imported historical data is insufficient, the system can temporarily adopt a weighting strategy based solely on the most recent operation time, uniformly setting the frequency weighting factor to a neutral value. Simultaneously, historical logs are gradually accumulated, and once the data volume reaches a preset scale, the system switches to a combination of time decay and frequency weighting. Similarly, when a component has no update records within a time window, the system can treat its update count as zero, thus obtaining a near-zero frequency weighting factor, making the component's original priority primarily determined by the time decay factor. In the normalization step, if the smoothed priorities of all candidate components are the same or have extremely small differences, the system uses a preset small constant to avoid a zero denominator and treats the normalized priorities as similar values. Subsequent use of semantic similarity or construction progress sequence data can further break up ties, thereby ensuring the sorting process can be stably executed under various boundary conditions.

[0093] In one embodiment, during the retrieval step, if a component description element corresponds to multiple candidate components, then based on the construction progress sequence data associated with the candidate components, the candidate components in the active construction phase are selected as priority candidate objects to determine the target component.

[0094] In one embodiment, the update step includes:

[0095] The progress status is compared with the planned progress stored in the BIM model to obtain progress deviation information, and the target components with progress deviations are automatically marked in the construction progress management interface.

[0096] In one embodiment, the method further includes:

[0097] After receiving natural language input, multimodal data associated with the target component is acquired. The multimodal data includes at least one type of on-site image data and sensor data.

[0098] Cross-checking the status of components in the BIM model with multimodal data is used to verify the reliability of the progress status.

[0099] In one embodiment, the method further includes:

[0100] Based on the updated construction progress information, the differences between the planned and actual progress of each floor, construction area, and component type are statistically analyzed. Construction progress management reports, including timeline views and Gantt chart views, are generated and presented in the construction progress management interface to support project managers in scheduling construction time and adjusting resources.

[0101] In one embodiment, the method further includes:

[0102] Based on construction progress information and progress deviation information, adjustment suggestions for subsequent construction tasks are automatically generated. The adjustment suggestions include at least the start and end times of subsequent construction tasks, construction team configuration suggestions, and material arrival time adjustment suggestions. These suggestions are then made available for project management personnel to confirm and implement in the construction progress management interface.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0104] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A BIM-based method for managing construction progress information in building projects, executed by computer equipment, characterized in that: Includes the following steps: Step S1: Receive natural language input of construction progress report submitted by construction personnel. The natural language input includes at least fuzzy description information of floor, area and component type and corresponding progress status. Step S2: Perform semantic parsing on the natural language input to extract at least one component description element and the progress status; Step S3: Based on the component description element, retrieve candidate component information from the BIM database associated with the BIM model. The BIM database stores component identifier, component type, spatial location, and construction progress data for each component. Step S4: Based on the pre-established semantic relationship model of components, calculate the matching degree of the candidate component information to generate a candidate component list; Step S5: Adjust the weights of the candidate components in the candidate component list based on construction history data, and determine the target component from the candidate component list; Step S6: Based on the progress status, update the construction progress information of the target component in the BIM model, and display the update result in the construction progress management interface.

2. The BIM-based construction progress information management method for building projects as described in claim 1, characterized in that, The component semantic relationship model is implemented by constructing a component graph, which includes: The type attributes of components, the spatial adjacency relationships between components, and the construction sequence dependencies between component construction procedures.

3. The BIM-based construction progress information management method for building projects as described in claim 2, characterized in that, The semantic parsing steps include: A pre-trained language model is used to identify spatial terms and construction action terms in the natural language input, and the identified spatial terms and construction action terms are mapped to nodes and edges in the component graph to obtain the component description elements.

4. The BIM-based construction progress information management method for building projects as described in claim 3, characterized in that, The matching degree calculation includes: Based on the semantic similarity between nodes in the component graph and the component description elements, the candidate components are semantically matched, and the candidate components are filtered by combining the spatial coordinate range of the candidate components. The semantic similarity is obtained by a preset semantic similarity calculation rule, and the spatial coordinate range is filtered by a preset spatial distance calculation rule.

5. The method for managing construction progress information of building projects based on BIM as described in claim 1, characterized in that, The weight adjustment steps include: The priority of each candidate component is dynamically adjusted based on the update frequency of the candidate component in the construction history data and the most recent operation time, wherein the priority is dynamically calculated based on the time decay factor and frequency weighting rule.

6. The method for managing construction progress information of building projects based on BIM as described in claim 1, characterized in that, In the retrieval step, if the component description element corresponds to multiple candidate components, then based on the construction progress sequence data associated with the candidate components, the candidate components in the active construction stage are selected as priority candidate objects to determine the target component.

7. The BIM-based construction progress information management method for building projects as described in claim 1, characterized in that, The update steps include: The progress status is compared with the planned progress stored in the BIM model to obtain progress deviation information, and the target components with progress deviations are automatically marked in the construction progress management interface.

8. A method for managing construction progress information of building projects based on BIM as described in claim 1, characterized in that, The method further includes: After receiving the natural language input, multimodal data associated with the target component is acquired, the multimodal data including at least one type of on-site image data and sensor data; The reliability of the progress status is verified by cross-checking the multimodal data with the component status in the BIM model.

9. A method for managing construction progress information of building projects based on BIM as described in claim 1, characterized in that, The method further includes: Based on the updated construction progress information, the differences between the planned and actual progress of each floor, construction area, and component type are statistically analyzed, and a construction progress management report including a timeline view and a Gantt chart view is generated and presented in the construction progress management interface.

10. A method for managing construction progress information of building projects based on BIM as described in claim 7, characterized in that, The method further includes: Based on the construction progress information and the progress deviation information, adjustment suggestions for subsequent construction tasks are automatically generated. The adjustment suggestions include at least suggestions for adjusting the start and end times of subsequent construction tasks, suggestions for construction team configuration, and suggestions for adjusting material arrival times. These suggestions are then confirmed and implemented by project management personnel in the construction progress management interface.

Images