Machine Learning Driven Approach to Adaptive Energy Re-routing for Photovoltaic Smart Homes

Abstract

The growing prevalence of photovoltaic (PV) systems in smart homes thus requires sophisticated energy
management techniques that will use energy optimally while reducing grid dependence. This paper presents an
intelligent energy optimization framework integrated with various machine learning (ML) models to forecast weather
and predict energy demand within Photovoltaic-powered smart homes with battery storage. The system enables dynamic
energy source allocations based on real-time analyses of solar generation, battery status, and forecasted demand to
ensure efficiency and sustainability in energy use. It adopts a strategic approach of utilizing battery management,
charging with excess solar energy, and reserving the stored energy for low solar radiation periods. At the same time, it
switches between solar, battery, and grid power to minimize energy wastage, and therefore, reduce the dependence on the
grid. This work demonstrates the capability of ML-driven energy systems to improve smart home resilience, thus
supporting sustainable and self-sufficient energy practices