A graph- and feature-based building space recognition algorithm for performance simulation in the early design stage
model recognition, early design stage, performance-oriented design, graph- and feature-based method
This paper presents a graph- and feature-based building space recognition algorithm for a boundary representation format (B-rep) geometric model, which can identify the building element type and space. The flow of the algorithm is described in detail, including the construction of a building geometric topology relation graph (BTG), the recognition of building element type, and the extraction of building space based on graph and local feature. The algorithm can be applied to the design of a building scheme; it can quickly identify and transform the geometric model into the input model required by the performance simulation software. This is a key step in realizing a performance-oriented design in the early design stage. We implemented this algorithm using SketchUp for testing its performance. Through the case study, it is proved that the algorithm can recognize the model and extract all the building spaces accurately. There is linear correlation between the recognition time and number of faces. Moreover, at the time of analysis, a model composed of 500 spaces and 3001 faces did not exceed 1.69 s, which meets the requirements of most applications well. Compared to previous works, this algorithm performs well in both recognition accuracy and time efficiency simultaneously, and can better serve the actual demand of automatic real-time building performance feedback in the early design stage. Finally, the future work regarding performance-oriented design based on model recognition is proposed.
Tsinghua University Press
Hongzhong Chen, Ziwei Li, Xiran Wang et al. A graph- and feature-based building space recognition algorithm for performance simulation in the early design stage. Build Simul, 2018, 11(2): 281–292.