基于改进X-12-ARIMA的电煤需求预测模型与实证研究

Electrical Coal Demand Forecasting Model and Case Studies Based on Improved X-12-ARIMA

考虑中国春节、端午、中秋等移动假日效应,对美国人口普查局开发的X-12-ARIMA模型进行了改进和实证分析。结果表明,中国电煤消费量具有显著的季节性特征,每年11—12月为消费最高峰,7—8月为消费小高峰;基于改进X-12-ARIMA模型对2013年1、2和3月份的电煤需求预测精度分别为96.6%、95.1%和93.7%,具有较好的短期预测能力。

Considering the floating holiday effects of China＇ s Spring Festival, the Dragon Boat Festival and the Mid-Autumn Festival, the improvement and corresponding case studies are put forward on the X-12-ARIMA model developed by the U.S. Census Bureau. The case study results show that electrical coal consumption is characterized by apparent seasonal patterns. Each year the highest peak of the consumption happens in the period of ＂November-December＂ while the small peak appears in the period of ＂July-August＂. As for January, February and March in 2013, the forecasting accuracies of the improved model based on X-12-ARIMA are 96.6%, 95.1% and 93.7%, respectively, which demonstrates its satisfactory performance in short-term forecasting.