Research On Data Consolidation Query And Detection Method Of Civil Engineering Supervision BIM Based On Ontology

Civil Engineering Supervision(CES)process management includes management of progress,quality,materials,personnel,vehicles,safety,etc.The processes involved are complicated,the amount of data is large,and there are many participants.It is difficult to coordinate and feedback in time.Moreover,there are many sources of data,whose formats are not uniform,and they have multiple representations.All of these are the problems that need to be solved urgently in the process of informatization in the field of CES.This thesis analyzes the whole life cycle activities of CES by used data-driven thinking,and proposes the top-level data organization organization mode of CES according to the supervision process.The data organization mode of multi-angle classification is proposed based on analyzing the CES data from various angles,such as the comprehensive supervision process,interaction unit and the users.By further analysis of CES data,the TDTM four-element data organization mode is proposed.At the same time,a integration algorithm of CES data model and CES data integration specification are given to realize the CES data from the outside to the inside,from the coarse to the thin,from the clutter to the specification integration,and the data collection and the relationship between them are standardized in order to avoid conflicts,reduce redundancy,and ensure data consistency.On this basis of the above research,the unified representation of CES data is studied,and a formal representation method is given.Then an ontology model of CES data is given based on ontology technology.Based on RDF and OWL,the representation of CES data based on ontology model is studied.In addition,the reasoning mechanism is discussed to support for the subsequent storage,query and detect.Aiming at the storage and retrieval of civil engineering supervision data,the CES Storage Model(CESSM)of civil engineering supervision ontology data is constructed,and the CES ontology data files are stored in a persistent non-relational database to improve the store and query performance for the CES ontology data.Some store methods are compared to show its advantage.At the same time,the query algorithm for CES ontology data is given,and the overall framework of ontology-based CES access query is further improved.For the CES video,the R-tree is uesd as the basic index structure of the key frame and annotation data of the CES video.In order to achieve query and efficient retrieval of CES video,the hierarchical semantic model of CES video are studied.The preprocessing algorithm,the node splitting algorithm based on spectral clustering of CESVS-R tree and the ontology data R tree retrieval algorithm based on the above algorithms are proposed.In addition,in order to solve the problem of vehicle detection based on surveillance video at construction site,a multi-scale CES video target detection method based on deep learning is proposed,which could distinguish the vehicle type while detecting vehicles.Based on the analysis of the integration,representation,storage,query and on-site vehicle detection of ontology-based CES data,the prototype of CES BIM system is developed,and the data conversion and interface with typical BIM are given,which verifies the availability of research results.The application examples were given for the proposed method,and the experiment analysis for the proposed algorithms are also.The examples and experimental results show the feasibility of the proposed method,and the algorithm has obvious advantages in main performance.