Development of statistical time series models for solar power prediction

Abstract

The increasing use of renewable energy sources necessitates accurate forecasting models for generation scheduling. Amongst the renewable sources, solar and wind have gained acceptance and are being increasingly used in distributed generation. The main problem with these sources is the dependence of their power output on natural environmental parameters at a given point of time. This paper proposes time series models for short-term prediction of solar irradiance from which solar power can be predicted. The predictions are done for 1 day ahead using different time-series models. For each model, these predicted values are compared with the actual values for the next day and graphs are plotted. Basic time-series models such as moving average and exponential smoothing were tested. The decomposition model is proposed, where the measured data is decomposed into seasonal and trend patterns and each of them predicted separately. The model was developed for different durations of data, to identify the best possible set of data. It is observed from the results that the prediction with decomposition model for 2 months data gave the best result with around 9.28% error.