Forecasting Indoor Temperature in a Passive Solar Building Under Semi-Arid Climate

Abstract

Buildings are increasingly exposed to severe climatic conditions, particularly large temperature fluctuations in arid regions, which create significant challenges for maintaining indoor thermal comfort and energy efficiency. This thesis investigates passive design strategies suitable for such climates, with a focus on improving the thermal performance of insulated buildings. The study proposes an integrated system that combines effective thermal insulation and shading techniques to enhance indoor thermal stability and reduce the impact of external climatic variations. In parallel, an artificial intelligence approach based on Artificial Neural Networks (ANNs) was developed to accurately predict indoor temperature using both indoor and outdoor climatic data obtained from an experimental setup at the Applied Renewable Energy Research Unit in Ghardaïa (32.36 ° N, 3.51 °E). The results demonstrated that the combination of passive design and ANN-based predictive system provides reliable and efficient thermal management. The developed ANN model achieved high predictive accuracy. With a correlation coefficient of approximately 98% between measured and predicted temperatures, indicating strong agreement and model robustness. These findings confirm the potential of integrating passive building strategies with intelligent predictive models as a sustainable and effective solution for improving thermal comfort and energy performance in harsh climatic environments.