ENTRIQ: Smart Energy Optimization with Explainable AI

Abstract

The increasing integration of renewable energy in smart buildings introduces challenges in efficient energy coordination due to the intermittent nature of solar power and dynamic load demand. Conventional rule-based control systems lack adaptability and fail to optimize energy usage under varying environmental conditions. ENTRIQ presents an LSTM-based multi-output intelligent energy optimization framework for solar–grid integrated smart buildings. The system leverages real-time IoT sensor data to predict optimal energy control actions and incorporates Explainable Artificial Intelligence (XAI) to interpret model decisions. Experimental results demonstrate stable and adaptive decision-making with improved energy utilization compared to conventional static approaches. The proposed framework provides a practical solution for intelligent energy management in IoT-enabled smart building environments. In addition, a safety-aware validation layer is integrated to prevent unstable control actions during fluctuating operating conditions. The framework also enhances transparency by identifying the key sensor features influencing energy management decisions.