Very Short-Term Photovoltaic Power Forecasting Considering the Stochastic Factors

This paper proposes photovoltaic (PV) power forecasting model, which is applied Gaussian blur algorithm filtering technique, for estimating power output and create stochastic forecasting model. As a result, affected power can be forecasted from stochastic factors with machine learning and artificial neural network. This work focuses on a very short-term forecast for 5 minutes ahead. As it uses only endogenous data, no exogenous data is needed. To evaluate the model, the result has been compared to the persistence model, which has a good accuracy for a short-term forecasting. It has been shown that the proposed PV forecasting considering the stochastic factors gained higher accuracy than the persistence model.