含置信区间的改进ARIMA电价预测曾鸣.pdf

第 37 卷 第 18 期 电力系统保护与控制 Vol.37 No.18 2009 年9 月16 日 Power System Protection and Control Sep. 16, 2009 含置信区间的改进ARIMA电价预测 曾 鸣，刘 玮，汪晓露 （华北电力大学工商管理学院，北京 102206） 摘要：电力市场环境下，电价已日益成为市场参与者关注的焦点。准确的电价预测能为各参与者提供重要的参考。国内外对 于电价预测的方法有很多。引用国内外相关理论对现有ARIMA预测模型进行了改进。在现有ARIMA电价预测模型的基础上， 采用游程检验法判断电价序列的平稳性，采用AIC准则确定ARMA模型的阶数，在误差预测中，借助统计学的区间估计理论， 给出了一种含置信区间的电价预测方法，更好地确定了电价序列的上下限，为发电商的竞价提供参考。 关键词: 电价预测；ARIMA；置信区间；预测模型；误差预测 An improved ARIMA approach on electricity price forecasting with confidence interval ZENG Ming, LIU Wei, WANG Xiao-lu (North China Electric Power University, Beijing 102206, China ) Abstract: In the power market, electricity price has become the focus of attention of market participants. Accurate price forecast for all participants can provide important reference. As there are a lot of ways for price forecasting, this paper improves the existing theory of ARIMA model. On the basis of the current ARIMA price forecasting model, it uses run-length test to determine the sequence of price stability and the AIC criterion to determine the order of ARMA model. In the error forecasting, with an estimated range of statistical theory, this paper gives a confidence interval of the electricity price forecasting to better determine the sequence of the upper and lower price for power generations’ bid for reference. Key words: price forecasting; ARIMA; confidence interval; forecasting model; error forecasting 中图分类号： TM73; F123.9 文献标识码：A 文章编号： 1674-3415(2009)18-0025-07 采用自相关系数和偏自相关系数来判断模型的阶 0 引言 数，通过判断托尾和截尾情况来确定模型的阶数。 在当前的电力市场环境下，电价已成为电力系 与上述方法相比，AIC准则