Particle swarm optimization based GM (1, 2)method for short-term electricity price forecasting with predicted error improvement

Based on the comprehensive analysis of change law of stock price in electricity market, a particle swarm optimization based GM (1, 2) method for short-term electricity price forecasting with predicted error calibration is proposed, in which the moving average method is used to process the raw data series to build the particle swarm optimization grey background value based GM (1, 2)model, and then ARMA in the time series analysis is used to establish ARMA predicted model for grey residual error series, and finally ARMA model's forcasting value is used to calibrate the forecasted results of price. The numerical example based on the historical data of the PJM market shows that the method can reflect the change law of electricity price more accurately and the forecasting accuracy can be improved virtually compared with the conventional GM (1, 2) model, and it can be used by electricity market participants to prepare their bidding strategies.