Using neural networks to model and forecast solar PV power generation at Isle of Eigg

Optimizing small-scale power systems, microgrids, involves modeling their components to include renewable energy (RE) sources, like solar. Obviously, RE generation is dependent on the weather conditions and because of this, solar photovoltaic (PV) power generation models are inherently nonlinear. This paper explores applicability of artificial neural networks (NN) to model solar power generation as a part of standalone microgrids found on remote islands. Specifically, it applies NN tools to the generation, weather, and irradiance time series data for the Isle of Eigg, Scotland, and shows a potential for further investigation including the predictive modeling based just on weather data that can then be used for optimization of the microgrid and microgrid performance.