ARTICLE ANALYSIS

Analysis of article using Artificial Intelligence tools





Id 2805
Author Huang C.; Zhang G.; Yao J.; Wang X.; Calautit J.K.; Zhao C.; An N.; Peng X.
Title Accelerated environmental performance-driven urban design with generative adversarial network
Reference

Huang C.; Zhang G.; Yao J.; Wang X.; Calautit J.K.; Zhao C.; An N.; Peng X. Accelerated environmental performance-driven urban design with generative adversarial network,Building and Environment 224

Keywords Architectural design; Climate models; Environmental management; Genetic algorithms; Numerical methods; Numerical models; Design-process; Environmental performance; Genetic optimization; Outdoor environment; Performance-driven; Performance-driven design; Surrogate modeling; Thermal; Urban blocks; Urban design; computer simulation; numerical model; surrogate method; urban design; urban development; urban planning; Generative adversarial networks
Link to article https://www.scopus.com/inward/record.uri?eid=2-s2.0-85143054602&doi=10.1016%2fj.buildenv.2022.109575&partnerID=40&md5=8f47cb67ad2cf13812333007afefe274
Abstract The morphological design of urban blocks greatly affects the outdoor environment. Currently, performance-based urban and building design relies on a time-consuming numerical simulation process, hindering performance optimization early in the design process. This paper proposes an automated design process that applies generative adversarial network (GAN) as a surrogate model to accelerate environmental performance-driven urban design. Parameterized urban blocks are designed for random sampling and constructing a numerical simulation database. The GAN model was trained to predict pedestrian level wind (PLW), annual cumulative solar radiation (Radiation) and Universal Thermal Climate Index (UTCI) in real-time. The GAN-based surrogate model is combined with a multi-objective genetic algorithm to achieve real-time optimization of urban morphology. The results show that on the test set, the pix2pix model using a specific encoding method predicts the R2 of 0.70, 0.86 and 0.80 for PLW, Radiation and UTCI, respectively, while the method can speed up 120–240 times compared to the numerical simulation method. The optimization results show that NSGA-II combined with global averaging pooling achieves the best optimization results. When the number of optimized samples exceeds 174, the proposed method has a time advantage over traditional methods for outdoor environment optimization in urban design. © 2022 Elsevier Ltd

Metodology

Technique

Keyword Find research methods used
Tentative Keyword Show Candidate Transition Variables for article (AI method)
Categories Find category for article (AI method)
Crossover theme Find social impact for article (AI method)
Wordcloud Show WordCloud from article (AI method)
Find semantically similar articles Find semantically similar articles (Semantic search)