A hybrid data mining driven algorithm for long term electric peak load and energy demand forecasting

Abstract

Load forecasting is one of the main required studies for power system expansion planning and operation. In order to capture the nonlinear and complex pattern in yearly peak load and energy demand data, a hybrid long term forecasting method based on data mining technique and Time Series is proposed. First, a forecasting algorithm based on the Support Vector Regression (SVR) method is developed. The parameters of the SVR technique along with the dimension of input samples are optimized using a Particle Swarm Optimization (PSO) method. Secondly, in order to minimize the forecasting error, a hybrid forecasting method is presented for long term yearly electric peak load and total electric energy demand. The proposed hybrid method acts based on the combination of Auto-Regressive Integrated Moving Average (ARIMA), Artificial Neural Network (ANN) and the proposed Support Vector Regression technique. The parameters of the ARIMA method are determined based on the autocorrelation and partial autocorrelation of the original and differenced time series. The proposed hybrid forecasting method prioritizes each forecasting method based on the resulted error over the existing data. The hybrid forecasting method is used to forecast the yearly peak load and total energy demand of Iran National Electric Energy System.