RESEARCH ON GRAY OPTIMIZATON COMBINATION POWER LOAD FORECASTING BASED ON MULTIVARIATE EXPONENTIAL WEIGHTING

Because power loads possess the features of increment and seasonal fluctuation and are influenced by various factors, so the change of power load presents complicated and non-linear combination character. When traditional forecasting models are applied to such a complicated load series, the forecasting accuracy is always not satisfied. To improve the forecasting accuracy, the authors proposed a new multivariable exponential weighting based gray optimal combination model to forecast the power loads, in which the gray optimal combination model was used to forecast the power loads with non-linear increasing trend and the multivariable exponential weighting method was used to solve the fluctuation of historical load data. Case calculation results show that the proposed method, in which various features of power loads are considered, can; effectively improve the accuracy of load forecasting.