Enhancing Energy Efficiency with Explainable AI: A Study in Smart Building Heating Load Prediction

Abstract

Artifical intelligence (AI) is a powerful tool for Achiving sustainable development goes. However, the complexity of any AI models possess the significant changes in the locals such as energy management. This paper expores the application of outstanding articulated intellectuality (xAI) techniques to enter the transparency and resolution of the machine learning models used for predicting energy consumption in smart buildings. Using a database containing structural and physical features of buildings, a pre-active model was developed based on the Random Forest regression algorithm. To interpret and explore the model’s preferences, two xAI methods - LIME and SHAP - were applied. The results demonstrate that both xAI techniques effectively identify the key features influencing load projections,thereby providing eligible insights into energy-saving strategies. Furthermore, the combination of machine learning and xAI methods forters trust and accounting in AI-driven energy management systems.