Hybrid ANN/GA/ANFIS Model for Very Short-Term PV Power Forecasting

The aim of this work is to develop a robust model for short-term prediction of Photovoltaics (PV) generation. The model is structured with algorithms that belong to the technical field of computational intelligence. This approach provides the potential to form a forecasting system with high flexibility, efficiency and customization. The paper examines various combinations of inputs, in order to fully investigate the influence of exogenous variables on the PV predicted time series. Simulation results indicate that the proposed model can be successfully implemented in the decision making process of retailers, distribution system operators, prosumers and others, to fully exploit the generation capacity of grid connected PV systems in day-ahead electricity markets. Out of the different combinations of inputs studied, the one that involves the panels' temperature together with historical PV power values lead to lower predictions errors.