Indoor-outdoor interactive thermal response in public building:onsite data collection and classification through cluster algorithm

Abstract

Efficient thermal environment control in large public buildings with atria is critical for reducing energy consumption and carbon emissions. This study investigates the indoor thermal environments of two distinct atria in Suining, China, within a Hot Summer and Cold Winter (HSCW) climate zone. A novel methodology combining K-means clustering algorithms and multiple linear regression (CAMLR) was employed to analyze extensive thermal data from a transitional atrium and a traditional atrium. CAMLR significantly outperformed Traditional Building Thermal environment analysis Methods (TBTM), achieving a 33.8 % higher R2 and reducing the residual sum of squares by 62.11 %. Key findings include: 1) A strong correlation (R2 = 0.9555) between outdoor temperature and indoor thermal conditions in the transitional atrium, highlighting a critical outdoor temperature threshold of 30 °C that leads to prolonged indoor heat retention. 2) Solar radiation plays a dominant role in the traditional atrium (R2 = 0.5584), influencing rapid temperature fluctuations. These insights underscore the critical role of atrium design in optimizing thermal performance and energy efficiency. Compared to TBTM, CAMLR provides faster, more accurate, and highly visualized data analysis, making it particularly suitable for complex thermal environment studies. This study establishes CAMLR as a robust tool for advancing thermal management strategies in large public buildings, offering practical guidance for energy-efficient architectural designs in similar climate zones.