An ANFIS model of electricity price forecasting based on subtractive clustering

Abstract

A day-ahead market clearing price forecasting method based on the Takagi-Sugeno model and the adaptive neuro-fuzzy inference system (ANFIS) is proposed. First, the structure of ANFIS is determined by subtractive clustering; then the premise parameters and consequent parameters of ANFIS are identified by the hybrid learning algorithm; finally, related factors that influence future daily electricity prices are input into the ANFIS to forecast next-day electricity prices. By use of the data of California Electricity Market in 1999, the forecasting model is constructed and the anticipated electricity prices of the next day are implemented. The forecasting results show that the forecasting model established by us is valid.