Monthly electric load forecasting with seasoanal decomposition based grey prediction method

In this study, first of all, load forecasting and the importance of load forecasting are explained and load forecasting techniques are discussed. Analyzing the studies in the literature the position of this study is specified. Data used in the implementation is the monthly maximum electricity loads of Ankara central metropolitan area for the years 2004-2007. Electricity loads are considered as time series and it is determined that these loads have a seasonal fluctuation. Seasonal decomposition method is preferred. Applying additive and multiplicative decomposition approaches involved by this method, individually, trend series obtained by removing seasonality are predicted by linear regression and grey prediction techniques. Then, integrating seasonal effects to predicted trend series, load predictions are obtained. Errors between obtained load prediction series and actual load series are analyzed for all four prediction models and the best model is chosen by some error criterions? calculation. Monthly load forecasting for the year 2008 is calculated with this model.