Data enabled predictive energy management of a PV-battery smart home nanogrid

This paper proposes a data-enabled predictive energy management strategy for a smart home nanogrid (NG) that includes a photovoltaic system and second-life battery energy storage. The key novelty is utilizing data-based forecasts of future load demand, weather conditions, electricity price, and power plant CO₂ emissions to improve the NG system efficiency. Specifically, a load demand forecast model is developed using an artificial neural network (ANN). The forecast model predicts load demand signals for a model predictive controller (MPC). Simulation results show that the data-enabled predictive energy management strategy achieves 96%-98% of the optimal NG performance derived via dynamic programming (DP). Its sensitivity to the control horizon length and load demand forecast accuracy are also investigated.