A simple and reliable PV forecasting method for local area energy management

Various forecasting methods for photovoltaic (PV) generations have been proposed so far. However, the conventional methods cannot be widely used in various situations because they require sophisticated data that cannot easily be obtained. Furthermore, the prediction accuracy of such methods tend to deteriorate especially due to lack of data. This paper proposes a simple and reliable PV forecasting method for local energy management. The proposed method uses only public weather forecasting data that is easily obtained. The method maintains high accuracy by using the real time correlation data between the target and neighboring areas. Multiple neural networks are effectively used based on a weather clustering technique. It has been confirmed that the proposed method shows the robustness in the prediction accuracy when used for local area PV prediction.