Electricity price forecasting methods combining ARIMA and GARCH with confidence intervals

Base on research on existing methods for electricity price forecasting,a new combined method with confidence intervals is proposed.The run test was used to judge the price series stationarity and the AIC criterion method was applied to determine the order,so the subjectivity can be avoided and the forecasting accuracy can be improved.Moreover,the ARIMA and GARCH were effectively combined though heteroscedasticity determination for electricity price error series.Also,the price forecasting method with confidence intervals is proposed,which overcomes the shortcomings of single-point forecasting and increases the algorithm flexibility to enable the participators to choose the price fluctuation range according to their expectations on price forecasting accuracy.The model is proved effective by data on PJM market.