1

Abstract—Based on the characteristics of power load series such as periodic randomness and instability, A new method for short-term load forecasting was presented in this paper. Firstly, the power load series was decomposed into intrinsic mode functions (IMF). Then, according to the traits of each IMF, System Identification Toolbox of Mat lab 7.0 was used in selecting models (AR, MA or ARMA), setting orders, realizing forecasting and reconstructing the predictive values. The simulation was carried out with the data from a grid and the results showed that this method could achieve a high level of precision in forecasting. Keywords- EMD, Time series, short-term load forecasting, System Identification Toolbox I