Prediction of spot market prices of electricity using chaotic time series

In the deregulated power systems, pricing is an important issue in a market environment. But there is always a dilemma of how to predict spot prices to generation companies (GenCos). This paper is concerned with the prediction of the spot prices in the electricity market using the method of nonlinear auto-correlated chaotic model associating with neural network and wavelet theory. Data information including the weather and day-ahead electric prices are preprocessed through the Fourier wave filter. A new wavelet based on neural network study programming, in which the Sigmoid function in the ANN is substituted by wavelet function, is presented to solve this problem. Through the approach, GenCos can make accurate decisions on scheduling generators and provide high quality power services to customers. The results of simulation through this new method demonstrate that the accuracy of prediction is greatly improved.