Building Information Modeling Spatial Topological Relations Extraction And Classification

Digitization of building information in the field of architectural design and urbanplan are widely used. In recent years, BIM(Building Information Modeling) plays avery important role in building’s design, construction, marketing, merchandising andoperations management stage. However, the BIM contains a large number of complexinformation, which is extremely valuable wealth of knowledge and experience. Inorder to overcome the phenomenon of "data rich and lack of knowledge", it is verynecessary researching building Information modeling spatial topological Relationsextraction and classification.Firstly, the spatial topological relations of building information modeling wouldbe a good character of the accessibility features in the interior space of a building.Firstly, there is rarely spatial topological relation extraction method for buildinginformation modeling. Secondly, the existing BIM software can providecomprehensive and large amount of building information parameters, but it can’tautomatically generate spatial topological relations inside the model as we want, so itis necessary to study the automatic extraction method, and apply the relations in modelclassification easily. It is needed to establish a reasonable description of the model torepresent the spatial topological relations. To this end, this paper intends to find anefficient and accurate spatial topological relation extraction method.Secondly, we build a classification of a combination of space syntax theory andclassification of SVM-based decision-making model. In this paper, the classification isdivided into feature extraction and SVM-based decision-making classification twolinks. The first establish RCARG model for building information modeling, and define a set of quantitative contrast components for the model’s inherent geometric feature;And then use the field property of building information model, combining the theory ofspace syntax, and use of syntactic variables rich spatial structure features of buildingmodel is a set of description to support classification model, this is largely improvedthe model classification of feature information, improves the precision of classification.Finally in the DAG-SVMS classification algorithm on the basis of increasedeigenvector equalization process, reduce misjudgment rate when decision classification,in order to realize high precision accuracy of classification result. Experimental resultsshow that the proposed method with KNN and improved compared to before the DAG-SVMS algorithm, has higher classification accuracy, and can be well applied to theclassification of building information model.