Data-driven prediction and optimization of residential building performance in Singapore considering the impact of climate change

Abstract

Research on the performance of tropical residential buildings requires careful consideration of the effects of the changing weather conditions. Based on machine learning and multi-objective genetic optimization algorithms, this study proposes an integrated framework for predicting and optimizing the performance of residential buildings in Singapore. The framework was used for two Singapore residential building types, the point block building and the slab block building, to examine their optimal performance in terms of daylight performance, energy efficiency, and thermal comfort under future climate conditions in the short term (2021–2040), middle term (2041–2060), and long term (2061–2100). The results show that the XGBoost algorithm and transfer learning method perform well in building performance prediction, specifically in the source domain (point block buildings) with R2 = 0.95 and in the target domain (slab block building) with R2 = 0.87. The general trend in the optimal design parameters for residential buildings is to enhance the thermal insulation of building walls and envelopes to better cope with future warming climate conditions, while providing adequate indoor lighting. This study contributes to the field of climate-resilient residential buildings in Singapore. This is closely linked to the Singapore government's quest for sustainable development and to enhance the well-being of its residents.