Annals of the University of Craiova: Economic Sciences Series (Dec 2016)
A SEASONAL AND MONTHLY APPROACH FOR PREDICTING THE DELIVERED ENERGY QUANTITY IN A PHOTOVOLTAIC POWER PLANT IN ROMANIA
Abstract
In this paper, we present solutions that facilitate the forecasting of the delivered energy quantity in a photovoltaic power plant using the data measured from the solar panels' sensors: solar irradiation level, present module temperature, environmental temperature, atmospheric pressure and humidity. We have developed and analyzed a series of Artificial Neural Networks (ANNs) based on the Levenberg-Marquardt algorithm, using seasonal and monthly approaches. We have also integrated our developed Artificial Neural Networks into callable functions that we have compiled using the Matlab Compiler SDK. Thus, our solution can be accessed by developers through multiple Application Programming Interfaces when programming software that predicts the photovoltaic renewable energy production considering the seasonal particularities of the Romanian weather patterns