A BIM-based construction schedule optimization system based on big data analysis

Abstract

This invention discloses a BIM construction schedule optimization system based on big data analysis, comprising: a task mapping module for generating a set of planned tasks, a set of components, and a point cloud of component models; a semantic indexing module for generating a BIM semantic constraint index; a corridor construction module for generating a temporally continuous corridor; a four-point basis construction module for generating a candidate four-point basis set; a transformation solving module for generating a candidate transformation set; an evidence ranking module for outputting the target registration transformation; and an optimization output module for generating the construction schedule optimization result. This invention improves registration accuracy, schedule recognition stability, and optimization result reliability.

Description

Technical Field

[0001] This invention relates to the field of building information modeling technology, and in particular to a BIM construction progress optimization system based on big data analysis. Background Technology

[0002] With the increasing demand for Building Information Modeling (BIM) technology, 3D perception technology for construction sites, and collaborative management of engineering data, BIM-based construction progress analysis and optimization technology has received widespread attention. Existing construction progress management systems mainly rely on manual inspection records, static BIM model comparison, or simple point cloud registration for construction status identification. However, the following problems are commonly found in practical applications: Construction plan data, BIM model data, and monitoring phase point cloud data are scattered and have inconsistent correlation granularity. Existing mapping methods often struggle to establish stable task-component correspondences, leading to inaccurate mapping results between the planned task set and the component set. The simultaneous presence of repetitive components, occluded areas, and unfinished boundaries on the construction site, coupled with the lack of semantic and temporal continuity constraints in traditional four-point base construction and ordinary point cloud registration methods, easily results in pseudo-matches, misregistrations, and cross-temporal jumps, affecting the stability of target registration transformations. Existing progress analysis methods primarily rely on geometric overlap, making it difficult to jointly utilize visible coverage, voxel occupancy rate, occlusion uncertainty, and process matching degree, resulting in insufficient accuracy in component completion status identification and low reliability in determining construction progress deviations.

[0003] Therefore, how to provide a BIM construction progress optimization system based on big data analysis is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] One objective of this invention is to propose a BIM construction progress optimization system based on big data analysis. This invention introduces an improved Super4PCS semantic constraint four-point base construction, temporally continuous corridor constraint, and completion evidence joint sorting mechanism. It describes in detail the component registration, state identification, and progress optimization implementation method driven by the collaborative driving of construction plan data, BIM model, and monitoring phase point cloud. It has the advantages of high registration accuracy, high progress identification stability, and high reliability of optimization results.

[0005] A BIM construction schedule optimization system based on big data analysis according to an embodiment of the present invention includes: The task mapping module is used to perform task mapping on construction plan data and BIM models to form a set of planned tasks, a set of components, and a point cloud of component models. The semantic indexing module is used to perform spatial association organization and process time binding on the component set to form a BIM semantic constraint index; The corridor construction module is used to perform temporal arrangement on the monitoring time phase point cloud sequence, extract the current time phase point cloud, and combine it with the previous monitoring time phase registration transformation, the set of completed components and the boundary of the unfinished area to form a temporally continuous corridor. The four-point basis construction module is used to perform improved Super4PCS semantic constraint four-point basis construction on the current phase point cloud and component model point cloud under BIM semantic constraint index and temporal continuous corridor constraints, forming a candidate four-point basis set; The transformation solution module is used to perform conformal matching search, rigid body transformation solution, candidate transformation clipping, and rollback verification based on the candidate four-point basis set to form a candidate transformation set. The evidence sorting module is used to apply the candidate transformation set to the component set to form the completed evidence vector set, and to perform joint sorting on the geometric consistency of the completed evidence vector set and the candidate transformations, and output the target registration transformation. The optimized output module is used to identify the completion status of components and construction progress deviations based on target registration transformation, and to perform task rearrangement on the planned task set to generate construction progress optimization results.

[0006] Optionally, the task mapping module specifically comprises: Read the task identifier, task start time, task end time, task area identifier, and task sequence constraints from the construction plan data; read the component identifier, component spatial location, component boundary range, and component geometric information from the BIM model to form task records and component records. The task records and component records are compared by region affiliation to screen out candidate mapping records where the task region identifier covers the spatial location of the component. Then, the start time, end time, and task sequence constraint are compared for the candidate mapping records to form task mapping records. Task mapping records are grouped by task identifier to form a planned task set, and task mapping records are grouped by component identifier to form a component set; Geometric sampling and point cloud transformation are performed on the geometric information of the components along the boundary range of the corresponding components in the component set to form the component model point cloud, and the component model point cloud is written into the component set according to the component identifier.

[0007] Optionally, the semantic indexing module specifically comprises: Expand the component identifier, component spatial location, component boundary range and task mapping relationship in the component set according to the component identifier to form a component index record. Expand the task identifier, task start time, task end time and task area identifier in the planned task set according to the task identifier to form a task index record. Based on the continuity between the spatial locations of components and the continuity between the boundary ranges of components, adjacency determination and pairing and merging are performed on the component index records to form component adjacency records. Based on the task mapping relationship and task area identifier, the area of ​​the component adjacency records is adjusted to form spatial association records. Bind the task identifier, task start time and task end time in the task index record to the component identifier in the spatial association record to form a process time binding record; Spatial association records and process time binding records are collected by component identifier to form a BIM semantic constraint index that includes component identifier, spatial association relationship and process time binding relationship.

[0008] Optionally, the corridor construction module specifically comprises: The monitoring time phase point cloud sequence is expanded according to the monitoring time phase order to form a time phase point cloud record, and time phase arrangement is performed according to the monitoring time phase order to form a time phase arrangement sequence; Extract the time phase point cloud record located in the current monitoring time phase from the time phase arrangement sequence to form the current time phase point cloud, and extract the time phase point cloud record located in the previous monitoring time phase from the time phase arrangement sequence; The registration transformation of the previous monitoring time phase is applied to the set of completed components and the boundary of the unfinished area to form the mapping results of the completed components and the mapping results of the unfinished boundary. Based on the spatial connection relationship between the current temporal point cloud, the mapping results of completed components, and the mapping results of unfinished boundaries, corridor connectivity is performed to form a temporally continuous corridor.

[0009] Optionally, the four-point base construction module specifically comprises: The current phase point cloud is expanded according to spatial location to form point cloud point records. The component model point cloud is expanded according to component identifier and spatial location to form model point records. Based on the component identifier, spatial relationship and process time binding relationship in the BIM semantic constraint index, semantic correspondence filtering is performed on the point cloud point records and model point records to form semantic constraint point records. The semantic constraint point records are mapped to the temporal continuous corridor. Based on the spatial coverage and spatial extension direction in the temporal continuous corridor, the semantic constraint point records are subjected to position constraint filtering to form corridor constraint point records. Improved Super4PCS semantic constraint four-point basis construction is performed on the corridor constraint point records. Point cloud four-point combinations and model four-point combinations are established according to the spatial extension direction in the temporally continuous corridor. The opposite side length relationship, diagonal relationship and intersection point segmentation ratio relationship in the point cloud four-point combination are calculated. The opposite side length relationship, diagonal relationship and intersection point segmentation ratio relationship in the model four-point combination are also calculated. The point cloud four-point combinations that satisfy the component identification consistency relationship, spatial association consistency relationship, process time binding consistency relationship, opposite side length consistency relationship, diagonal consistency relationship and intersection point segmentation ratio consistency relationship are paired with the model four-point combinations to form four-point basis candidate records. Perform component identification consistency judgment, spatial association consistency judgment, and process time binding consistency judgment on the four-point basis candidate records to form a candidate four-point basis set.

[0010] Optionally, the transformation solution module specifically comprises: Extract point cloud four-point combinations and model four-point combinations from the candidate four-point base set, establish point pairing relationships according to the order of the four points, and perform conformal matching search along the point pairing relationships to form pairing transformation records; Perform rigid body transformation on the point pairing relationship in the pairing transformation record, calculate the rotation and translation parameters of the four-point combination of the point cloud to the four-point combination of the model, and organize the rotation parameters, translation parameters and point pairing relationship into candidate transformation records; Based on the component identifiers, spatial relationships, and process time binding relationships in the BIM semantic constraint index, and combined with the spatial coverage and spatial extension direction in the temporally continuous corridor, the candidate transformation records are subjected to component identifier consistency judgment, spatial relationship consistency judgment, process time binding consistency judgment, spatial coverage consistency judgment, and spatial extension consistency judgment. Candidate transformation records that meet the judgment conditions are then subjected to candidate transformation clipping to form clipped transformation records. The rotation and translation parameters in the clipping transformation record are applied to the four-point combination of the point cloud to form the transformation point. The positional deviation between the transformation point and the spatial position in the four-point combination of the model is calculated, and the transformation point is returned to the original spatial position in the four-point combination of the point cloud according to the reverse mapping relationship. The clipping transformation record that meets the positional deviation requirements and the consistency relationship of the rollback position is retained to form a candidate transformation set.

[0011] Optionally, the evidence sorting module specifically comprises: Apply each candidate transformation in the candidate transformation set to the component spatial location, component boundary range, and component model point cloud in the component set to form a transformation component record corresponding to each candidate transformation; Based on the spatial positional correspondence between the transformation component record and the current temporal point cloud, coverage and occupancy determination are performed on the transformation component record, the visible coverage and voxel occupancy rate corresponding to each candidate transformation are calculated, and occlusion determination is performed on the transformation component record based on the point cloud distribution in the current temporal point cloud to form an occlusion uncertainty record. Based on the task mapping relationship and process time binding relationship between the transformation component record and the planned task set, process correspondence analysis is performed on the transformation component record to calculate the process matching degree corresponding to each candidate transformation. The visible coverage, voxel occupancy rate, occlusion uncertainty and process matching degree are merged according to the candidate transformation to form a completion evidence vector set. Perform a joint sorting of the completed evidence vector set and the geometric consistency of the candidate transformations, and output the target registration transformation corresponding to the sorting result.

[0012] Optionally, the optimized output module specifically comprises: The target registration transformation is applied to the spatial location, boundary range, and point cloud of the component model in the component set to form the target transformation component record; Based on the spatial position correspondence, boundary coverage, and point cloud occupancy relationship between the target transformation component record and the current temporal point cloud, the component identifiers in the target transformation component record are judged one by one to form a component completion status record. According to the component identifier in the component completion status record and the task mapping relationship in the planned task set, the task correspondence is performed. Combined with the task start time, task end time and task sequence constraints in the planned task set, the schedule deviation is calculated on the task correspondence result to form a construction schedule deviation record. Based on the task identifiers, schedule deviation values, and task sequence constraints in the construction schedule deviation record, the task start time, task end time, and task sequence are adjusted for the planned task set to generate a task rearrangement result. The task rearrangement results, component completion status records, and construction progress deviation records are merged according to task identifiers to form the construction progress optimization results.

[0013] The beneficial effects of this invention are: Compared with existing construction progress management methods that rely on manual inspection records, static BIM model comparison, or ordinary point cloud registration, this invention establishes a stable correlation between construction plan data, BIM models, and monitoring phase point cloud sequences through a task mapping module, semantic indexing module, and corridor construction module. This unifies task identifiers, component identifiers, spatial correlations, and process time binding relationships into the same technical link, effectively solving the problems of difficulty in accurately matching planned tasks with component objects and discontinuous spatial relationships under changes in monitoring phases. On this basis, this invention utilizes an improved Super4PCS semantic constraint four-point base construction, candidate transformation pruning, and rollback verification, so that the four-point base construction no longer depends solely on geometric relationships, but is simultaneously constrained by BIM semantic constraint index and temporally continuous corridor constraints. This reduces the probability of pseudo-matches and misregistrations caused by the combined effects of duplicate components, occluded areas, and unfinished boundaries from the source, significantly improving the registration accuracy between the current phase point cloud and the component model point cloud and the stability of cross-temporal registration.

[0014] Furthermore, this invention, through an evidence sorting module and an optimization output module, jointly sorts the visible coverage, voxel occupancy rate, occlusion uncertainty, process matching degree, and candidate transformation geometric consistency degree. This ensures that the target registration transformation not only meets spatial registration requirements but also construction status identification requirements. Consequently, it can more accurately identify the completion status of components, determine construction progress deviations, and adjust the task start time, task end time, and task sequence according to task identifiers and task sequence constraints. Ultimately, it forms a more engineering-applicable construction progress optimization result. Therefore, compared with the prior art, this invention has the advantages of more reliable registration results, more accurate component completion determination, more stable progress deviation identification, and task rearrangement that better reflects the actual construction status. It can effectively improve the automation level and on-site application value of BIM construction progress optimization. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of a BIM construction progress optimization system based on big data analysis proposed in this invention; Figure 2 This is a schematic diagram of the improved Super4PCS semantic constraint four-point basis structure of a BIM construction progress optimization system based on big data analysis proposed in this invention. Detailed Implementation

[0016] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0017] refer to Figures 1-2 A BIM-based construction schedule optimization system based on big data analysis includes: The task mapping module is used to perform task mapping on construction plan data and BIM models to form a set of planned tasks, a set of components, and a point cloud of component models. The semantic indexing module is used to perform spatial association organization and process time binding on the component set to form a BIM semantic constraint index; The corridor construction module is used to perform temporal arrangement on the monitoring time phase point cloud sequence, extract the current time phase point cloud, and combine it with the previous monitoring time phase registration transformation, the set of completed components and the boundary of the unfinished area to form a temporally continuous corridor. The four-point basis construction module is used to perform improved Super4PCS semantic constraint four-point basis construction on the current phase point cloud and component model point cloud under BIM semantic constraint index and temporal continuous corridor constraints, forming a candidate four-point basis set; The transformation solution module is used to perform conformal matching search, rigid body transformation solution, candidate transformation clipping, and rollback verification based on the candidate four-point basis set to form a candidate transformation set. The evidence sorting module is used to apply the candidate transformation set to the component set to form the completed evidence vector set, and to perform joint sorting on the geometric consistency of the completed evidence vector set and the candidate transformations, and output the target registration transformation. The optimized output module is used to identify the completion status of components and construction progress deviations based on target registration transformation, and to perform task rearrangement on the planned task set to generate construction progress optimization results.

[0018] In this embodiment, the task mapping module specifically comprises: When creating task records and component records, the system reads task identifiers, task start times, task end times, task area identifiers, and task sequence constraints from the construction plan data, and component identifiers, component spatial locations, component boundary ranges, and component geometric information from the BIM model. It extracts task identifiers, task start times, task end times, task area identifiers, and task sequence constraints according to the field order in the construction plan data, and extracts component identifiers, component spatial locations, component boundary ranges, and component geometric information according to the field order in the BIM model. Fields corresponding to the same task identifier are grouped into one task record, and fields corresponding to the same component identifier are grouped into one component record, thus creating task records and component records. The task records and component records are compared for regional affiliation to filter out candidate mapping records where the task region identifier covers the component spatial location. Then, the candidate mapping records are compared for start time, end time, and task order constraint. When forming a task mapping record, the region range corresponding to the task region identifier in the task record is read, and the component spatial location in the component record is read to determine whether the component spatial location falls within the region range. The task records and component records that meet the region coverage relationship are paired to form candidate mapping records. The task start time, task end time, and task order constraint are read for each candidate mapping record. The candidate mapping records are filtered according to the time arrangement relationship and task arrangement relationship. The candidate mapping records that meet the region affiliation relationship, start time relationship, end time relationship, and task order constraint relationship are retained to form the task mapping record. Task mapping records are grouped by task identifier to form a planned task set. When grouping task mapping records by component identifier to form a component set, the same identifier merging is performed along the task identifier in the task mapping record. The component identifiers and mapping relationships merged under the same task identifier are organized into the same task unit to form a planned task set. The same identifier merging is performed along the component identifier in the task mapping record. The task identifiers and mapping relationships merged under the same component identifier are organized into the same component unit to form a component set. Geometric sampling and point cloud transformation are performed on the component geometric information along the boundary range of the corresponding component in the component set to form a component model point cloud. When writing the component model point cloud into the component set according to the component identifier, the boundary range of each component identifier is read from the component set, geometric sampling points are extracted along the boundary range of the component geometric information, and the geometric sampling points are converted into a point set arranged according to spatial coordinates to form a component model point cloud. The component model point cloud is then written into the corresponding component unit in the component set according to the component identifier.

[0019] In this embodiment, the semantic indexing module specifically comprises: The component identifier, component spatial location, component boundary range, and task mapping relationship in the component set are expanded according to the component identifier to form a component index record. The task identifier, task start time, task end time, and task area identifier in the planned task set are expanded according to the task identifier to form a task index record. When expanding the component records in the component set according to the component identifier, the component spatial location, component boundary range, and task mapping relationship corresponding to the same component identifier are organized into the same component index record. The task records in the planned task set are expanded according to the task identifier to form a component index record. The task start time, task end time, and task area identifier corresponding to the same task identifier are organized into the same task index record. Based on the continuity between component spatial locations and the continuity between component boundary ranges, adjacency determination and pairing merging are performed on component index records to form component adjacency records. Then, based on task mapping relationships and task area identifiers, region consolidation is performed on component adjacency records to form spatial association records. When forming spatial association records, the continuity between component spatial locations is determined based on the distance and arrangement relationships between component spatial locations, and the continuity between component boundary ranges is determined based on the adjacency and connection relationships between component boundary ranges. Component index records that satisfy the continuity and continuity relationships are paired and merged into component adjacency records. Finally, based on the task identifier corresponding to the task mapping relationship and the region range corresponding to the task area identifier, region division and region merging are performed on component adjacency records to form spatial association records. When binding the task identifier, task start time, and task end time in the task index record to the component identifier in the spatial association record to form a process time binding record, the correspondence between the task identifier and the component identifier is determined according to the task mapping relationship. The task start time and task end time are written into the spatial association record where the corresponding component identifier is located. The correspondence between the task identifier and the component identifier, as well as the correspondence between the task start time and the task end time, are organized into the same record to form a process time binding record. When spatial association records and process time binding records are collected by component identifier to form a BIM semantic constraint index containing component identifier, spatial association relationship and process time binding relationship, spatial association records are merged according to component identifier, and process time binding records are merged according to component identifier. Spatial association relationships and process time binding relationships corresponding to the same component identifier are organized into the same index unit to form a BIM semantic constraint index.

[0020] In this embodiment, the corridor construction module specifically includes: The monitoring time phase point cloud sequence is unfolded according to the monitoring time phase order to form a time phase point cloud record. The time phase arrangement is then performed according to the monitoring time phase order to form a time phase arrangement sequence. The point cloud data corresponding to each monitoring time phase is extracted along the arrangement position in the monitoring time phase point cloud sequence, and the point cloud data corresponding to each monitoring time phase is organized into a time phase point cloud record. Then, the time phase point cloud records are sequentially arranged according to the monitoring time phase order so that the time phase point cloud records are continuously unfolded along the monitoring time phase order to form a time phase arrangement sequence. Extract the time phase point cloud record located at the current monitoring time phase from the time phase arrangement sequence to form the current time phase point cloud. When extracting the time phase point cloud record located at the previous monitoring time phase from the time phase arrangement sequence, locate the corresponding time phase point cloud record of the current monitoring time according to the arrangement position in the time phase arrangement sequence, and determine the point cloud data in the corresponding time phase point cloud record of the current monitoring time as the current time phase point cloud. Locate the corresponding time phase point cloud record of the previous monitoring time according to the arrangement position of the current monitoring time, and determine the point cloud data in the corresponding time phase point cloud record of the previous monitoring time as the previous monitoring time phase point cloud. When the registration transformation of the previous monitoring phase is applied to the set of completed components and the boundary of the unfinished area to form the mapping results of the completed components and the mapping results of the unfinished boundary, the spatial transformation relationship in the registration transformation of the previous monitoring phase is applied to the spatial position of the components and the boundary range of the components in the set of completed components, so that the set of completed components obtains the mapping result of the completed components under the registration transformation of the previous monitoring phase. The spatial transformation relationship in the registration transformation of the previous monitoring phase is applied to the boundary position and the boundary extension range of the unfinished area, so that the boundary of the unfinished area obtains the mapping result of the unfinished boundary under the registration transformation of the previous monitoring phase. Based on the spatial connection relationship between the current temporal point cloud, the completed component mapping result, and the unfinished boundary mapping result, corridor connectivity is performed to form a temporally continuous corridor. When forming a temporally continuous corridor, position correspondence determination is performed according to the point cloud distribution position in the current temporal point cloud and the component mapping position in the completed component mapping result. Boundary correspondence determination is performed according to the point cloud boundary in the current temporal point cloud and the boundary mapping position in the unfinished boundary mapping result. The spatial regions that satisfy the position correspondence relationship and the boundary correspondence relationship are connected and organized along the monitoring temporal phase sequence to form a temporally continuous corridor.

[0021] In this embodiment, the four-point base construction module is specifically as follows: The current phase point cloud is expanded according to spatial location to form point cloud point records. The component model point cloud is expanded according to component identifier and spatial location to form model point records. Based on the component identifier, spatial association relationship and process time binding relationship in the BIM semantic constraint index, semantic correspondence filtering is performed on the point cloud point records and model point records. When forming semantic constraint point records, each point in the current phase point cloud is arranged according to spatial location order and the spatial location of each point is organized into the point cloud point records. Each point in the component model point cloud is arranged according to component identifier order and spatial location order and the component identifier and the spatial location of each point are organized into the model point records. Then, using the component identifier, spatial association relationship and process time binding relationship in the BIM semantic constraint index as filtering conditions, the correspondence relationship comparison is performed on each point cloud point record and model point record. Point records that satisfy the component identifier correspondence relationship, spatial association correspondence relationship and process time binding relationship are retained to form semantic constraint point records. Semantic constraint point records are mapped to a temporally continuous corridor. Based on the spatial coverage and spatial extension direction in the temporally continuous corridor, position constraint filtering is performed on the semantic constraint point records. When forming corridor constraint point records, the spatial position in the semantic constraint point records is written into the corresponding position in the temporally continuous corridor. The inclusion relationship between each record in the semantic constraint point records and the spatial coverage in the temporally continuous corridor is determined. The direction consistency relationship between the arrangement direction in the semantic constraint point records and the spatial extension direction in the temporally continuous corridor is determined. Semantic constraint point records that satisfy the spatial coverage constraint and the spatial extension direction constraint are retained to form corridor constraint point records. An improved Super4PCS semantic constraint four-point basis construction is performed on the corridor constraint point records. Point cloud four-point combinations and model four-point combinations are established according to the spatial extension direction in the temporally continuous corridor. The relationships of opposite side lengths, diagonals, and intersection point segmentation ratios in the point cloud four-point combinations and the model four-point combinations are calculated. Point cloud four-point combinations that satisfy the consistency relationships of component identification, spatial association, process time binding, opposite side length, diagonal, and intersection point segmentation ratios are paired with model four-point combinations to form four-point basis candidate records. In the spatial extension direction of the time-series continuous corridor, four-point selection and combination arrangement are performed on the corridor constraint point record. A four-point combination of point cloud is formed on the point cloud point record side, and a four-point combination of model is formed on the model point record side. Based on the relative positional relationship in the four-point combination, the correspondence between the lengths of the two sets of opposite sides, the correspondence between the two diagonals, and the division ratio of the diagonal intersection point on the diagonal are calculated. Then, the four-point combination of point cloud and the four-point combination of model are paired according to the consistency of component identification, spatial association, process time binding, opposite side length, diagonal, and intersection point division ratio to form four-point base candidate records. For the candidate four-point basis records, perform component identifier consistency judgment, spatial association consistency judgment, and process time binding consistency judgment to form a candidate four-point basis set. When forming a candidate four-point basis set, extract the component identifier correspondence, spatial association correspondence, and process time binding correspondence for each of the point cloud four-point combinations and model four-point combinations in the candidate four-point basis records. Perform consistency judgment on the component identifier correspondence, spatial association correspondence, and process time binding correspondence. Retain the four-point basis candidate records that satisfy the component identifier consistency, spatial association consistency, and process time binding consistency to form a candidate four-point basis set.

[0022] This invention addresses the shortcomings of traditional Super4PCS in BIM construction scenarios. It relies solely on geometric consistency to construct four-point bases, struggles to distinguish between repetitive and symmetrical components, fails to utilize semantic information from different construction phases, cannot adapt to continuous changes in monitoring timeframes, and cannot directly apply registration results to construction progress assessment. The invention provides targeted improvements to Super4PCS for BIM construction progress optimization scenarios. These improvements include introducing BIM semantic constraint indexes into the four-point base construction process, enabling component identifiers, spatial relationships, and process time binding relationships to participate in point record filtering and four-point combination pairing. This reduces invalid search space and suppresses false matches caused by repetitive structures. Furthermore, it incorporates temporally continuous corridors into the four-point base construction and candidate transformation generation processes, ensuring that the current timeframe point cloud is registered and transformed with the previous monitoring timeframe, including the set of completed components and unfinished areas. Spatial extension constraints are established between boundaries to reduce registration jumps caused by temporal phase switching. The transformation verification process is then strengthened through candidate transformation pruning and rollback verification to eliminate candidate solutions inconsistent with the evolution of the construction state. The completion evidence vector set and the geometric consistency of the candidate transformations are jointly sorted so that the target registration transformation no longer stays at the geometric optimal level, but simultaneously meets the requirements of progress identification and spatial matching. This makes the improved Super4PCS have semantic constraint capabilities, temporal constraint capabilities, and progress-driven capabilities. It can significantly improve the accuracy of four-point base construction, the credibility of candidate transformations, the accuracy of component completion status identification, and the stability of construction progress deviation judgment. At the same time, it reduces the mismatch rate, reduces the amount of manual verification, enhances the applicability in complex occlusion environments and low overlap environments, and improves the continuity, interpretability, and engineering practical value of BIM construction progress optimization results.

[0023] In this embodiment, the transformation solution module is specifically as follows: Extract point cloud four-point combinations and model four-point combinations from the candidate four-point base set. Establish point pairing relationships according to the order of the four points. Perform conformal matching search along the point pairing relationships to form pairing transformation records. According to the order of the four points in the candidate four-point base set, establish pairing relationships between each point in the point cloud four-point combination and the corresponding point in the model four-point combination. Perform correspondence checks on the relative positional relationship between the point cloud four-point combination and the model four-point combination based on the point pairing relationships. Organize the point pairing results that satisfy the correspondence relationship into pairing transformation records. Rigid body transformation is performed on the point pairing relationships in the pairing transformation record to solve the rotation and translation parameters of the four-point combination of the point cloud to the four-point combination of the model. When the rotation parameters, translation parameters and point pairing relationships are organized into candidate transformation records, the spatial positions of the four-point combination of the point cloud and the four-point combination of the model are extracted based on the point pairing relationships in the pairing transformation record. The rotation and translation parameters are calculated according to the corresponding change relationship between the spatial positions. The rotation parameters, translation parameters and point pairing relationships are written into the same record to form candidate transformation records. Based on the component identifiers, spatial relationships, and process time binding relationships in the BIM semantic constraint index, and combined with the spatial coverage and spatial extension direction in the temporally continuous corridor, the candidate transformation records are subjected to consistency checks for component identifiers, spatial relationships, process time binding, spatial coverage, and spatial extension. Candidate transformation records that meet the criteria are then trimmed. When forming trimmed transformation records, a consistency comparison is performed between the point pairing relationships in the candidate transformation records and the component identifier correspondences in the BIM semantic constraint index. The spatial position change relationships in the candidate transformation records are also compared with the BIM semantic constraint index. The spatial relationships in the IM semantic constraint index are compared for consistency. The time correspondence in the candidate transformation record is compared for consistency with the process time binding relationship in the BIM semantic constraint index. The distribution range of the transformation points in the candidate transformation record is compared for consistency with the spatial coverage range in the temporal continuous corridor. The transformation direction in the candidate transformation record is compared for consistency with the spatial extension direction in the temporal continuous corridor. Candidate transformation records that satisfy the consistency of component identification, spatial relationship, process time binding, spatial coverage, and spatial extension are retained to form the trimming transformation record. The rotation and translation parameters in the clipping transformation record are applied to the four-point combination of the point cloud to form transformed points. The positional deviation between the transformed points and the spatial positions in the four-point combination of the model is calculated, and the transformed points are returned to their original spatial positions in the four-point combination of the point cloud according to the reverse mapping relationship. Clipping transformation records that meet the positional deviation requirements and the consistency relationship of the rollback position are retained to form a candidate transformation set. When forming a candidate transformation set, the rotation and translation parameters in the clipping transformation record are applied to the spatial positions of each point in the four-point combination of the point cloud to obtain transformed points. The point-by-point deviation between the transformed points and the corresponding spatial positions in the four-point combination of the model is calculated, and the transformed points are returned to their original spatial positions in the four-point combination of the point cloud according to the reverse mapping relationship. The positional consistency between the returned results and the original spatial positions is compared, and clipping transformation records that meet the positional deviation requirements and the consistency relationship of the rollback position are retained to form a candidate transformation set.

[0024] In this embodiment, the evidence sorting module specifically comprises: When applying each candidate transformation in the candidate transformation set to the component spatial position, component boundary range, and component model point cloud in the component set to form a transformation component record corresponding to each candidate transformation, the spatial transformation relationship in the candidate transformation is extracted one by one according to the arrangement order in the candidate transformation set, and the spatial transformation relationship is applied to the component spatial position, component boundary range, and component model point cloud in the component set. The position results, boundary results, and point cloud results obtained by each component under the action of the candidate transformation are sorted out accordingly, so that the component spatial position change results, component boundary range change results, and component model point cloud change results corresponding to the same candidate transformation are included in the same record to form a transformation component record; Based on the spatial position correspondence between the transformation component record and the current phase point cloud, coverage and occupancy determination are performed on the transformation component record. The visible coverage rate and voxel occupancy rate corresponding to each candidate transformation are calculated. Based on the point cloud distribution in the current phase point cloud, occlusion determination is performed on the transformation component record. When forming an occlusion uncertainty record, the spatial position and boundary range of the component in the transformation component record are compared with the spatial position of the point cloud in the current phase point cloud. The distribution range of the point cloud that falls within the component boundary range and maintains a correspondence with the component spatial position is counted. The visible coverage rate is calculated based on the coverage ratio between the corresponding point cloud distribution range and the component boundary range. The voxel occupancy rate is calculated based on the occupancy of the current phase point cloud in the voxel position corresponding to the component model point cloud. Occlusion determination is performed based on the distribution of missing regions, sparse regions and occluded regions in the point cloud distribution in the component boundary range. The occlusion determination results are organized into an occlusion uncertainty record. Based on the task mapping relationship and process time binding relationship between the transformed component records and the planned task set, process correspondence analysis is performed on the transformed component records to calculate the process matching degree corresponding to each candidate transformation. When the visible coverage, voxel occupancy rate, occlusion uncertainty and process matching degree are merged according to the candidate transformations to form the completion evidence vector set, a corresponding retrieval is performed according to the component identifier in the transformed component record and the task mapping relationship in the planned task set to determine the task identifier and task time position corresponding to each component. The time position corresponding to the transformed component record is compared with the task start time, task end time and task arrangement relationship in the planned task set according to the process time binding relationship. The process matching degree is calculated based on the matching results. The visible coverage, voxel occupancy rate, occlusion uncertainty and process matching degree corresponding to the same candidate transformation are organized into the same evidence unit to form the completion evidence vector set. When performing joint sorting on the geometric consistency of the completion evidence vector set and candidate transformations, and outputting the target registration transformation corresponding to the sorting result, the visible coverage, voxel occupancy, occlusion uncertainty, process matching degree, and geometric consistency of each candidate transformation are extracted according to the correspondence between the candidate transformation and the completion evidence vector. The sorting value is calculated based on the visible coverage, voxel occupancy, occlusion uncertainty, process matching degree, and geometric consistency of the candidate transformation, and the sorting is performed in order according to the size of the sorting value. The first candidate transformation in the sorting result is taken as the target registration transformation.

[0025] In this embodiment, the optimized output module specifically refers to: When applying the target registration transformation to the spatial position, boundary range, and point cloud of the component in the component set to form the target transformation component record, the spatial position, boundary range, and point cloud of the component are extracted one by one according to the component identifier in the component set. The spatial transformation relationship in the target registration transformation is applied to the spatial position, boundary range, and point cloud of the component to obtain the transformation position, transformation boundary, and transformation point cloud of each component under the action of the target registration transformation. The component identifier, transformation position, transformation boundary, and transformation point cloud are then organized into the same record to form the target transformation component record. Based on the spatial position correspondence, boundary coverage, and point cloud occupancy relationship between the target transformation component record and the current time phase point cloud, the component identifiers in the target transformation component record are judged one by one to form a component completion status record. When forming a component completion status record, a position correspondence comparison is performed based on the transformation position in the target transformation component record and the point cloud distribution position in the current time phase point cloud. A boundary coverage comparison is performed based on the transformation boundary in the target transformation component record and the point cloud coverage range in the current time phase point cloud. A point cloud occupancy comparison is performed based on the transformation point cloud in the target transformation component record and the point cloud occupancy range in the current time phase point cloud. The position correspondence results, boundary coverage results, and point cloud occupancy results are written into the corresponding record of each component identifier to form a component completion status record. According to the component identification in the component completion status record and the task mapping relationship in the planned task set, the task correspondence is performed. Combined with the task start time, task end time and task sequence constraints in the planned task set, the progress deviation is calculated on the task correspondence results to form the construction progress deviation record. When forming the construction progress deviation record, the task mapping relationship is retrieved in the planned task set according to the component identification in the component completion status record to determine the task identification corresponding to each component identification. Then, the task start time, task end time and task sequence constraints corresponding to the task identification are compared with the execution time position and sequence position of the completion judgment result in the component completion status record. The comparison results are organized into the task identification correspondence record to form the construction progress deviation record. Based on the task identifiers, schedule deviation values, and task sequence constraints in the construction schedule deviation record, the task start time, task end time, and task sequence are adjusted for the planned task set. When generating the task rearrangement result, the task start time adjustment amount and task end time adjustment amount are determined based on the schedule deviation value in the construction schedule deviation record, and the task identifier's position in the planned task set is determined based on the task sequence constraints in the construction schedule deviation record. The task start time adjustment amount, task end time adjustment amount, and position adjustment results are written into the corresponding record of the task identifier to generate the task rearrangement result. When merging the task rearrangement results, component completion status records, and construction progress deviation records according to task identifiers to form the construction progress optimization results, the task rearrangement results are aggregated according to the same identifier, the component completion status records are aggregated according to the same identifier, and the construction progress deviation records are aggregated according to the same identifier. The task rearrangement results, component completion status records, and construction progress deviation records corresponding to the same task identifier are then organized into the same result unit to form the construction progress optimization results.

[0026] Example 1: To verify the feasibility of this invention in practice, it was applied to a building construction project that included a main structure construction area and an electromechanical installation area. During the project construction, there were problems such as a large number of planned tasks and components, a high proportion of repetitive components on standard floors, and partial occlusion of the monitoring point cloud due to scaffolding and temporary material stacking. Traditional construction progress management methods mainly rely on manual inspection records and comparison with the static BIM model, which can identify obvious delays but is difficult to continuously determine the completion status of components. While ordinary point cloud registration methods can achieve spatial alignment between the point cloud and the model, traditional methods are not feasible. However, false matching and jumps are prone to occur in areas with repeated distribution of beams and columns, boundary occlusion areas, and areas with changes across monitoring cycles, resulting in unstable progress deviation identification results and discrepancies between task adjustment results and actual on-site conditions. During the project testing phase, a total of 128 planned tasks, 4260 component objects, and 4260 sets of component model point clouds were included. Point clouds were continuously collected for 8 monitoring phases, with the number of effective point clouds in each phase maintained between 850,000 and 1,020,000 points. Repeated components accounted for approximately 38% of the total number of components, and occlusion areas accounted for approximately 16% of the effective observation area, which can better reflect the actual technical problems that this invention intends to solve.

[0027] During application, construction plan data and BIM models first undergo task mapping to form a set of planned tasks, a set of components, and a point cloud of component models. Then, the semantic index module performs spatial association organization and process time binding on the component set to form a BIM semantic constraint index, enabling component identification, spatial association, and process time binding to directly participate in subsequent registration. After temporal arrangement, the current temporal point cloud sequence is extracted from the monitoring phase point cloud sequence. The registration transformation of the previous monitoring phase, the set of completed components, and the boundary of the unfinished area are used together to construct a temporally continuous corridor, limiting the point cloud registration of the current cycle to the range of continuous construction changes. The four-point base construction stage does not use the pure geometric four-point combination method of ordinary Super4PCS. Instead, it first uses the BIM semantic constraint index to screen out semantic constraint point records that satisfy the consistency of component identification, spatial association, and process time binding. Then, combined with the spatial coverage and spatial extension direction in the temporally continuous corridor, it establishes a point cloud four-point combination and a model four-point combination, and forms a candidate four-point base set based on the opposite side length relationship, diagonal relationship, and intersection point segmentation ratio relationship. In the transformation solution stage, candidate solutions that do not satisfy the spatial coverage consistency relationship and spatial extension consistency relationship are eliminated through candidate transformation pruning and rollback verification. In the evidence ranking stage, the visible coverage rate, voxel occupancy rate, occlusion uncertainty and process matching degree are merged into a completion evidence vector set, and then jointly ranked with the candidate transformation geometric consistency. The target registration transformation is output. Finally, the component completion status is identified based on the target registration transformation, the construction progress deviation is calculated, and the planned task set is rearranged.

[0028] The present invention was compared with the static BIM comparison method combined with manual inspection and the ordinary Super4PCS method. The results are shown in Table 1: Table 1 Comparison of Construction Progress Optimization Effects

[0029] As shown in Table 1, the present invention significantly outperforms the comparative methods in terms of the number of candidate four-point bases, registration error, inter-period jump rate, completion status recognition accuracy, and progress deviation recognition accuracy. This indicates that the improved Super4PCS semantic constraint four-point base construction can effectively compress the invalid search space, the temporally continuous corridor can significantly reduce unreasonable registration jumps across monitoring cycles, and the joint ranking of the completion evidence vector set and the geometric consistency of candidate transformations can make the target registration transformation more in line with the construction status determination requirements. After eight consecutive monitoring periods, the construction progress optimization results output by the present invention include adjustments to the start time of 29 tasks, adjustments to the end time of 11 tasks, and adjustments to the order of 7 tasks. The adjustment results, after on-site verification, maintain a high degree of consistency with the actual construction status, indicating that the present invention can stably achieve component completion status recognition and construction progress optimization in construction scenarios with many repetitive components, strong occlusion interference, and significant inter-period changes.

[0030] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A BIM-based construction schedule optimization system based on big data analysis, characterized in that, include: The task mapping module is used to perform task mapping on construction plan data and BIM models to form a set of planned tasks, a set of components, and a point cloud of component models. The semantic indexing module is used to perform spatial association organization and process time binding on the component set to form a BIM semantic constraint index; The corridor construction module is used to perform temporal arrangement on the monitoring time phase point cloud sequence, extract the current time phase point cloud, and combine it with the previous monitoring time phase registration transformation, the set of completed components and the boundary of the unfinished area to form a temporally continuous corridor. The four-point basis construction module is used to perform improved Super4PCS semantic constraint four-point basis construction on the current phase point cloud and component model point cloud under BIM semantic constraint index and temporal continuous corridor constraints, forming a candidate four-point basis set; The transformation solution module is used to perform conformal matching search, rigid body transformation solution, candidate transformation clipping, and rollback verification based on the candidate four-point basis set to form a candidate transformation set. The evidence sorting module is used to apply the candidate transformation set to the component set to form the completed evidence vector set, and to perform joint sorting on the geometric consistency of the completed evidence vector set and the candidate transformations, and output the target registration transformation. The optimized output module is used to identify the completion status of components and construction progress deviations based on target registration transformation, and to perform task rearrangement on the planned task set to generate construction progress optimization results.

2. The BIM construction schedule optimization system based on big data analysis according to claim 1, characterized in that, The task mapping module is specifically as follows: Read the task identifier, task start time, task end time, task area identifier, and task sequence constraints from the construction plan data; read the component identifier, component spatial location, component boundary range, and component geometric information from the BIM model to form task records and component records. The task records and component records are compared by region affiliation to screen out candidate mapping records where the task region identifier covers the spatial location of the component. Then, the start time, end time, and task sequence constraint are compared for the candidate mapping records to form task mapping records. Task mapping records are grouped by task identifier to form a planned task set, and task mapping records are grouped by component identifier to form a component set; Geometric sampling and point cloud transformation are performed on the geometric information of the components along the boundary range of the corresponding components in the component set to form the component model point cloud, and the component model point cloud is written into the component set according to the component identifier.

3. The BIM construction schedule optimization system based on big data analysis according to claim 1, characterized in that, The semantic indexing module specifically comprises: Expand the component identifier, component spatial location, component boundary range and task mapping relationship in the component set according to the component identifier to form a component index record. Expand the task identifier, task start time, task end time and task area identifier in the planned task set according to the task identifier to form a task index record. Based on the continuity between the spatial locations of components and the continuity between the boundary ranges of components, adjacency determination and pairing and merging are performed on the component index records to form component adjacency records. Based on the task mapping relationship and task area identifier, the area of ​​the component adjacency records is adjusted to form spatial association records. Bind the task identifier, task start time and task end time in the task index record to the component identifier in the spatial association record to form a process time binding record; Spatial association records and process time binding records are collected by component identifier to form a BIM semantic constraint index that includes component identifier, spatial association relationship and process time binding relationship.

4. The BIM construction schedule optimization system based on big data analysis according to claim 1, characterized in that, The corridor construction module is specifically as follows: The monitoring time phase point cloud sequence is expanded according to the monitoring time phase order to form a time phase point cloud record, and time phase arrangement is performed according to the monitoring time phase order to form a time phase arrangement sequence; Extract the time phase point cloud record located in the current monitoring time phase from the time phase arrangement sequence to form the current time phase point cloud, and extract the time phase point cloud record located in the previous monitoring time phase from the time phase arrangement sequence; The registration transformation of the previous monitoring time phase is applied to the set of completed components and the boundary of the unfinished area to form the mapping results of the completed components and the mapping results of the unfinished boundary. Based on the spatial connection relationship between the current temporal point cloud, the mapping results of completed components, and the mapping results of unfinished boundaries, corridor connectivity is performed to form a temporally continuous corridor.

5. The BIM construction schedule optimization system based on big data analysis according to claim 1, characterized in that, The four-point base construction module is specifically as follows: The current phase point cloud is expanded according to spatial location to form point cloud point records. The component model point cloud is expanded according to component identifier and spatial location to form model point records. Based on the component identifier, spatial relationship and process time binding relationship in the BIM semantic constraint index, semantic correspondence filtering is performed on the point cloud point records and model point records to form semantic constraint point records. The semantic constraint point records are mapped to the temporal continuous corridor. Based on the spatial coverage and spatial extension direction in the temporal continuous corridor, the semantic constraint point records are subjected to position constraint filtering to form corridor constraint point records. Improved Super4PCS semantic constraint four-point basis construction is performed on the corridor constraint point records. Point cloud four-point combinations and model four-point combinations are established according to the spatial extension direction in the temporally continuous corridor. The opposite side length relationship, diagonal relationship and intersection point segmentation ratio relationship in the point cloud four-point combination are calculated. The opposite side length relationship, diagonal relationship and intersection point segmentation ratio relationship in the model four-point combination are also calculated. The point cloud four-point combinations that satisfy the component identification consistency relationship, spatial association consistency relationship, process time binding consistency relationship, opposite side length consistency relationship, diagonal consistency relationship and intersection point segmentation ratio consistency relationship are paired with the model four-point combinations to form four-point basis candidate records. Perform component identification consistency judgment, spatial association consistency judgment, and process time binding consistency judgment on the four-point basis candidate records to form a candidate four-point basis set.

6. The BIM construction progress optimization system based on big data analysis according to claim 1, characterized in that, The transformation solution module is specifically as follows: Extract point cloud four-point combinations and model four-point combinations from the candidate four-point base set, establish point pairing relationships according to the order of the four points, and perform conformal matching search along the point pairing relationships to form pairing transformation records; Perform rigid body transformation on the point pairing relationship in the pairing transformation record, calculate the rotation and translation parameters of the four-point combination of the point cloud to the four-point combination of the model, and organize the rotation parameters, translation parameters and point pairing relationship into candidate transformation records; Based on the component identifiers, spatial relationships, and process time binding relationships in the BIM semantic constraint index, and combined with the spatial coverage and spatial extension direction in the temporally continuous corridor, the candidate transformation records are subjected to component identifier consistency judgment, spatial relationship consistency judgment, process time binding consistency judgment, spatial coverage consistency judgment, and spatial extension consistency judgment. Candidate transformation records that meet the judgment conditions are then subjected to candidate transformation clipping to form clipped transformation records. The rotation and translation parameters in the clipping transformation record are applied to the four-point combination of the point cloud to form the transformation point. The positional deviation between the transformation point and the spatial position in the four-point combination of the model is calculated, and the transformation point is returned to the original spatial position in the four-point combination of the point cloud according to the reverse mapping relationship. The clipping transformation record that meets the positional deviation requirements and the consistency relationship of the rollback position is retained to form a candidate transformation set.

7. The BIM construction schedule optimization system based on big data analysis according to claim 1, characterized in that, The evidence sorting module is specifically as follows: Apply each candidate transformation in the candidate transformation set to the component spatial location, component boundary range, and component model point cloud in the component set to form a transformation component record corresponding to each candidate transformation; Based on the spatial positional correspondence between the transformation component record and the current temporal point cloud, coverage and occupancy determination are performed on the transformation component record, the visible coverage and voxel occupancy rate corresponding to each candidate transformation are calculated, and occlusion determination is performed on the transformation component record based on the point cloud distribution in the current temporal point cloud to form an occlusion uncertainty record. Based on the task mapping relationship and process time binding relationship between the transformation component record and the planned task set, process correspondence analysis is performed on the transformation component record to calculate the process matching degree corresponding to each candidate transformation. The visible coverage, voxel occupancy rate, occlusion uncertainty and process matching degree are merged according to the candidate transformation to form a completion evidence vector set. Perform a joint sorting of the completed evidence vector set and the geometric consistency of the candidate transformations, and output the target registration transformation corresponding to the sorting result.

8. The BIM construction schedule optimization system based on big data analysis according to claim 1, characterized in that, The optimized output module is specifically: The target registration transformation is applied to the spatial location, boundary range, and point cloud of the component model in the component set to form the target transformation component record; Based on the spatial position correspondence, boundary coverage, and point cloud occupancy relationship between the target transformation component record and the current temporal point cloud, the component identifiers in the target transformation component record are judged one by one to form a component completion status record. According to the component identifier in the component completion status record and the task mapping relationship in the planned task set, the task correspondence is performed. Combined with the task start time, task end time and task sequence constraints in the planned task set, the schedule deviation is calculated on the task correspondence result to form a construction schedule deviation record. Based on the task identifiers, schedule deviation values, and task sequence constraints in the construction schedule deviation record, the task start time, task end time, and task sequence are adjusted for the planned task set to generate a task rearrangement result. The task rearrangement results, component completion status records, and construction progress deviation records are merged according to task identifiers to form the construction progress optimization results.

Images