摘要
煤炭属于重要的民用能源,对其消费量进行预测,可为合理安排煤炭生产提供依据,优化社会资源的配置。采用Box和Jenkins的ARIMA模型,对1953年以来我国煤炭消费量的年度数据进行分析。与结构性因果模型、自回归（AR）、移动平均（MA）、自回归移动平均（ARMA）模型等相比较,ARIMA模型不但适合于我国煤炭消费量的非平稳时间序列的特点,并且预测效果比较理想。结果表明,ARIMA（3,1,3）模型的预测效果良好,2002年~2005年平均预测误差仅为3.981%,可用于未来我国煤炭消费量的预测。
Coal belongs to one of the important civil energy sources. Forecast and analyses on coal consumption can help make reasonable arrangements in coal production and optimize the collocation of social resources. This paper uses the ARIMA model which was advanced by Box and Jenkins to analyze annual data of coal consumption in China. The original data are provided by China Statistics Yearbook. Unlike models such as the structure analysis of cause and effect, the autoregressive method （AR）, and the moving average method （MA）, the ARIMA model is not only suitable for the analysis of coal consumption in China, which is not a balanced time series, but also its forecasting effect is exact. This paper concludes that the forecast applying ARIMA （3, 1, 3）, which forecasting error only 3.981 percent from 2002 to 2005, is exact and the AMIMA model can be used to forecast short-term of coal consumption in China.
出处
《能源研究与信息》
2007年第2期117-123,共7页
Energy Research and Information
