Abstract
It is important to forecast photovoltaic (PV) power output for the stable and efficient grid operation, intelligent household energy management, and so on. In this paper, a new forecasting method using real PV outputs power data of multipoint PV systems is proposed. In this method, first, the output power data of respective PV systems are normalized. Then, the future motion of the geographic distribution of the normalized values is estimated. Finally, future normalized values are re-converted to future output of respective PV power systems. We assess this method by analyzing errors between actual PV output and predicted PV output.
