Short-Term Electricity Price Forecasting Using Optimal TSK Fuzzy Systems

Since all financial transactions in restructured power markets are based on electricity prices, it is necessary that the price of electric power be predicted precisely. Some particular features such as: nonlinearity, non-stationary behaviors, as well as volatility of electricity prices make such a prediction a very challenging task. In this paper, a new structure of TSK fuzzy systems is presented that provides high order TSK fuzzy systems from lower orders which have capability of modeling and forecasting chaotic time series. The method used for optimization of fuzzy systems is the Interior point method. Using this method for forecasting electricity price is useful because of its chaotic behavior. The results are compared with RBF neural network and TSK fuzzy system presents better results.