Long and Medium Term Power Load Forecasting with Multi-Level Recursive Regression Analysis

Abstract

A lot of exogenous factors and the sequential variance held by the power system itself should be considered in the medium and long term load forecasting of power system. In view of these characters, the study on long and medium term load forecasting is worked by the means of multi-level recursive regression in this paper, and the model is established based on it. Forecasting is divided into two parts: the forecasting of time-varying parameters and the forecasting of power load based on the former one. The forecasting precision is improved by precise prediction of time-varying parameters and sufficient consideration of the important function of the related factors. In this paper, the yearly demand energy of a region in China from 1996 to 2003 is taken as data for the modeling. Simulation result shows that the average error of the forecasted power load of 2004 to 2006 is 1.5%. So the applicability and advantage of the method is verified.