基于不对称自回归条件异方差模型的短期负荷预测  

Short-Term Load Forecasting Based on Asymmetric Autoregressive Conditional Heteroscedasticity Models

作　　者：陈昊[1] 

机构地区：[1]江苏电力公司南京供电公司,江苏省南京市210008

出　　处：《电网技术》2008年第15期84-89,共6页Power System Technology

摘　　要：研究了负荷时间序列的自回归条件异方差效应,提出了一种基于不对称自回归条件异方差模型的短期负荷预测方法。建立了广义误差分布假设下的不对称广义自回归条件异方差模型,借助模型的不对称参数,分析了不同冲击下的不对称机制,比较了各种广义自回归条件异方差模型的预测能力。其中,幂指数广义自回归条件异方差-广义误差分布模型的预测效果尤为突出。最后通过实际算例验证了上述方法的可行性和有效性。In this paper on the basis of research on autoregressive conditional heteroscedasticity （ARCH） effect of load time series, a feasible short-term load forecasting method based on asymmetric ARCH models is proposed. The asymmetficARCH, such as EGARCH, TARCH, PARCH and TCGARCH, models under the assumption of generalized error distribution （GED） are built. By means of asymmetric parameters of the models, the asymmetric mechanism under different impacts is analyzed. The forecasting abilities of various asymmetric generalized ARCH models are compared, in which the performance of power ARCH-GED model is far better than others. The feasibility and effectiveness of the proposed method is validated by the results of practical calculation example.

关 键 词：负荷预测 不对称自回归条件异方差模型(ARCH) 逆杠杆效应 厚尾 广义误差分布(GED) 

分 类 号：TM715[电气工程—电力系统及自动化]