关键词: 灰色模型, 平滑处理, 预期影响, 拟合精度, 预测精度

Abstract: The medium-long term power load forecasting is fundamental to system regulation, power plant and network construction and demand side management. Based on the conventional medium-long term grey forecasting model, the moving average, transformation of power functions and the expected impacts are introduced to improve the data preprocessing of the traditional grey model. It is the first time that the expected impact adjustment component is applied to power load forecasting. Through deeper mining and sufficient utilization of information, the prediction accuracy can be improved. The methods for evaluating the fitting and prediction accuracy of the model quantitatively are also introduced. At last, based on the power loads in China between 1999 and 2011, the fitting and prediction accuracy of the improved methods are compared in two scenes. And, the power loads of 2012 and 2013 are also predicted with the improved methods.

Key words: grey model, smoothing, expected influence, fitting accuracy, prediction accuracy