基于ARMA模型对我国电力消费量的预测

对于我国电力消费需求量时间序列自身规律的认识有助于电力消费量的预测，从而可以避免电荒和电力生产能力的过度投资。以1980年-2007年期间我国年发电量的时间序列数据为基础，通过建立指数回归一模型来拟合时间序列数据。运用残差序列趋势和残差序列相关图、偏相关图的分析，对模型的五种形式的参数估计结果进行比较分析，最后选择模型来拟合发电量时间序列数据，并对其模型拟合的残差进行了单位根检验，检验结果表明残差序列是平稳的，采用模型是合理的。

China's Electricity Consumption Forecast Based on the ARMA Model

Understanding of the laws that China＇s demand for electricity consumption in their own time series will help forecasting electricity consumption, and then avoiding electrical scarcity and the excessive investment in power production capacity. Based on China＇s generating capacity time-series annual data from 1980 to 2007, using the index regression -ARMA （ p, q） model to fit the time-series data, by analyzing of residual sequence trends plot residual serial correlation plot and partial correlation plot, as well as analyzing comparatively the five forms of the estimated parameters results of the ARMA （ p, q） model, we finally choose ARMA （2,1） model to fit the data, process the unit root test at the same time. The test results show that the residual sequence is stable. It is reasonable to use the ARMA（2,1 ） model.