研究电力市场系统边际电价(system marginal price,SMP)条件方差的变化规律及残差的统计分布特征,据此引入广义自回归条件异方差(generalized auto-regressive conditional heteroskedasticity,GARCH)模型,并建立了基于加权双高斯(weighed double Gaussian,WDG)分布假设的GARCH模型(GARCH-WDG)对系统边际电价的变化规律进行研究。美国PJM市场和澳大利亚NSW市场的实际数据表明,GARCH模型对电价的估计和预测均有良好的效果,GARCH-WDG模型则进一步改善了GARCH模型的性能。 The variance of system marginal price (SMP) and the statistical distribution of residuals are studied. Based on this, a generalized auto-regressive conditional heteroskedasticity (GARCH) model is introduced and a The GARCH-WDG model based on the weighed double Gaussian (WDG) distribution hypothesis is used to study the variation law of the system marginal price. The actual data from the PJM market in the United States and the NSW market in Australia show that the GARCH model has good effect on the estimation and prediction of the electricity price and the GARCH-WDG model further improves the performance of the GARCH model.