Genetic Programming with Rough Sets Theory for Modeling Short-term Load Forecasting

The accurate and robust short-term load forecasting (STLF) plays a significant role in electric power operation. The accuracy of STLF is greatly related to the selected the main relevant influential factors. However, how to select appropriate influential factor is a difficult task because of the randomness and uncertainties of the load demand and its influential factors. In this paper, a novel method of genetic programming (GP) with rough sets (RS) theory is developed to model STLF to improve the accuracy and enhance the robustness of load forecasting results. RS theory is employed to process large data and eliminate redundant information in order to find relevant factors to the short-term load, which are used as sample sets to establish forecasting model by means of GP evolutional algorithm. The presented model is applied to forecast short-term load using the actual data from GuiZhou power grid in China. The forecasted results are compared with BP artificial neural Network with RS theory, and it is shown that the presented forecasting method is more accurate and efficient.