The variability of solar and wind generation poses challenges for grid stability and efficiency. This paper presents a smart energy management system (SEMS) with an AI-driven software architecture to optimize renewable energy use in buildings and microgrids. The platform integrates IoT sensor data, machine learning forecasts (e.g. Support Vector Regression) for generation/load prediction, and a predictive scheduling algorithm for battery storage and load control. A microservices architecture with cloud/edge deployment is used for scalability and fault tolerance. In simulation with realistic profiles, the SEMS improves renewable self-consumption and reduces grid dependency, consistent with results in similar studies.
Bothra et al. (Tue,) studied this question.