基于建筑信息模型(BIM)的建筑空间功能分类方法  

A CLASSIFICATION METHOD UPON BUILDING SPACES BASED ON BIM

作　　者：金超逸 徐旻洋 周向东[1] Jin Chaoyi;Xu Minyang;Zhou Xiangdong(School of Computer Science,Fudan University,Shanghai 200433,China;Shanghai Xian Dai Architectural Design(Group)Co.,Ltd.,Shanghai 200041,China)

机构地区：[1]复旦大学计算机科学技术学院,上海200433 [2]上海现代建筑设计(集团)有限公司,上海200041

出　　处：《计算机应用与软件》2018年第11期220-225,共6页Computer Applications and Software

基　　金：国家自然科学基金项目(61370157);上海市科技创新行动计划项目(17DZ1203600)

摘　　要：提出一种基于BIM(Building Information Modeling)数据的建筑空间功能分类方法。基于非监督学习技术对BIM建筑模型中的相关属性和拓扑信息进行挖掘;基于图的信息传播过程得到每个功能空间的特征表达;通过支持向量机(SVM)实现对建筑空间功能的自动分类。主要贡献是提出一种基于IFC(Industry Foundation Class)标准的BIM度量特性和拓扑结构特性的提取方法,并采用经典的分类器技术获得了较好的分类准确率。在真实的BIM数据集上进行了充分的实验,实验结果验证了方法的有效性。We presented a method to classify building spaces based on building information modeling(BIM)data.An unsupervised learning technique was used to mine related attributes and topological information in BIM building model.The feature expression of each functional space was obtained on the basis of graph information propagation process.Support vector machine(SVM)was applied to realize automatic classification of building space functions.The main contribution was to propose a method for extracting BIM metrics and topology features based on industry foundation class(IFC)standard.And the classical classifier technology was adopted to get better classification accuracy.Experiments on real BIM datasets show that the proposed method is effective.

关 键 词：图传播 无监督学习 分类 BIM 数据挖掘 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]