Skip to Main Content
In order to improve the forecasting precision of traditional grey model for short-term price in competitive electricity market, a novel grey model is presented in this paper based on period-decoupled price sequence. According to the interval time of market cleaning, the historical price data are divided into 24 sequences or 48 sequences. In the proposed grey model, two kinds of price sequences, called the main sequence (MS) and the reference sequence (RS), are defined. The correlation coefficient between price sequences of adjacent time intervals is analyzed, which is more than 0.9522 obtained from the Nordpool data in 2007. Therefore, it is determined that the MS is composed of prediction-period price data, and the RS is composed of hour-before-period price data. Furthermore, considering the limitation of the least square method (LSM) used in the traditional grey model for identification the developing coefficient a and the grey input b, the Particle Swarm Optimization algorithm (PSO) is adopted instead of LSM. Thus the PSOGM (1,2) forecasting model to short-term price is founded. The historical data from the Nordpool power market is used for computing, and the numerical results demonstrate that the MAPE of PSOGM (1,2) model for short-term price rolling prediction is 5.0626% and 7.5491% for continuous forecasting, raising 3%~20% compared with traditional grey model.