本文针对改革开放以来即1978年到2010年我国一次性能源生产量数据,首先运用单位根检验及差分方法来处理该一次性能源生产量时间序列的非平稳性,然后在此基础上进行模型识别、定阶、参数估计和适应性检验,最终选择出ARMA(1,3)模型。最后,根据该模型对最近五年的一次性能源生产量进行预测并与实际值进行对比分析。 In this paper, from 1978 to 2010, China’s primary energy production data since the reform and opening up, the first unit root test and the difference method to deal with the one-time energy production time series of non-stationary, and then on the basis of model identification , Order, parameter estimation and fitness test, finally choose ARMA (1,3) model. Finally, based on the model, one-off energy production in the last five years is predicted and compared with the actual value.