Mid-long-term Power System Load Forecasting Based On Gray Theory

Mid-long term load forecasting in power system is the basis of power system planningand scheduling. In particular, the renewable energy and a lot of uncertainty load to accesssystem. The prediction technology is facing new challenges, therefore, in the power system,the research of middle and long term load prediction technology has important practical valueand theoretical significance. The traditional GM(1,1) model is a low accuracy model. In orderto reduce the prediction errors of GM(1,1) model caused by sudden load change, the grayprediction theory is used to research the medium-and long-term power system loadforecasting.To improve the accuracy of GM(1,1) model on the medium and long term loadforecasting with mutant historical load data, this article make a correction on forecastingresults of GM(1,1) model based on numerical approximation algorithm principle. Combinewith the pretreatment on historical load data, the improved GM(1,1) gray model for themedium and long term load forecasting is established. The results of the new model show itshigher accuracy of the medium and long term load forecasting.Concerning the non-optimal coefficient given by GM (1,1) grey model in the process ofcreating the model of load forecasting actually, establish the NGM (1,1, k) gray model withcorrected coefficient by adding correction amount to coefficients and combining withpretreatment of the buffer operator to historical load data. The improved gray model has beenused in medium and long term load forecasting, and the results shows that the improved graymodel can effectively enhance the prediction accuracy.