中文摘要：

采用广义自回归条件异方差（GARCH）族模型分析了负荷时间序列波动性的动态时变结构,提出了模型系的概念,使用滚动数据窗技术估计了指数广义自回归条件异方差（EGARCH）和幂自回归条件异方差（PARCH）模型系;在研究动态显著性水平线（DSL）的基础上,探究了时序波动性的动态时变结构,讨论了贯穿各子样本空间的波动不对称效应。算例分析中,将所建模型应用于短期负荷预测,比较了GARCH族模型的预测能力,得到了较高的预测精度。

英文摘要：

Generalized Auto Regressive Conditional Heteroskedasticity（GARCH） model was employed to analyze the dynamic time-varying structure of volatility in load time series.The concept of model series was proposed,and by means of smoothing windows technique,the Exponential Generalized Auto Regressive Conditional Heteroskedasticity（EGARCH） and Power Auto Regressive Conditional Heteroskedasticity（PARCH） model series were estimated.Based on the study of Dynamic Significance Line（DSL）,the time-varying structure of volatility in time series was investigated,and the asymmetry effects over all the sub-sample space were indicated.In the calculation analysis,the proposed models were applied in short time load forecasting,and the forecasting ability of GARCH-type model was compared with that of the other models.It reveals that the proposed GARCH-type model holds higher forecasting accuracy.