Electricity reference price forecasting with Fuzzy C-means and Immune Algorithm

A new hybrid training method for radial basis function (RBF) neural network is presented in this paper. The proposed methodology produces RBF neural network models based on specially designed fuzzy C-means (FCM) and fuzzy immune algorithm (FIA), which are used to auto-configure the structure of networks and obtain the model parameters. With the proposed method, the number of hidden layer neurons and cluster centers are automatically determined according to the given data; both the output weight values and cluster radii are calculated by fuzzy immune algorithm. Meanwhile, the wavelet de-noising technique is introduced to ensure the neural network performance. This learning approach is proved to be effective by applying the optimized RBF neural network in predicting of Mackey-Glass chaos time series and forecasting of Queensland electricity reference price from Australian National Electricity Market.