Generation and evaluation of photovoltaic forecasts based on freely available weather data

Abstract

Photovoltaic forecasts are necessary in many fields of the energy industry. The rapid expansion of photovoltaic systems within the last years poses a growing problem especially for grid operators. New innovative concepts like smart storage solutions for reducing the grid load often need a generation forecast in order to be operated properly. In this work, a photovoltaic forecast based on only freely available cloud data is developed and compared to a commercially available one by evaluating their accuracy. Therefore, both forecasts are adapted to a 14 kWp photovoltaic system in Upper Bavaria, Germany. An Analysis of the results shows that both forecasts yield comparable results, but the one based on freely available data shows extreme inaccuracies in case of difficult prediction conditions like changeable cloud movement.