Thermal performance predictor for campus exterior spaces: a smart campus mobile application development

This study takes a subtropical university campus and its exterior spaces as a case study, exploring and predicting the spatial thermal performance via machine learning approaches. Subsequently, a novel smart campus mobile application to enhance the spatial cognition of students and faculty was developed.

The study leverages a suite of digital tools to conduct comprehensive simulations of the thermal performance of the campus exterior space. Based on the datasets from simulation, the real-time Multi-Layer Perceptron (MLP) prediction models for 33 spaces were realized, and their accuracy was validated through the field measurement. The developed application was designed to enhance students’ and faculty’s cognition of spatial thermal performance and the selection of appropriate spaces for various activities, in addition to facilitating management and the aggregation of post-occupancy evaluations.