A Chaotic Time Series Prediction Method Based on Fuzzy Neural Network and Its Application

Abstract

An approach based on chaos theory and fuzzy neural network (FNN) is proposed for chaotic time series prediction. Firstly, C-C algorithm is applied to estimate the delay time of chaotic signal. Grassberger-Procaccia (G-P) algorithm and least squares regression are employed to calculate the correlation dimension of chaotic signal simultaneously. Considering the difficulty in determining the number of input nodes of FNN, minimum embedding dimension obtained from chaotic time series analysis is used to design FNN. It was proved from two study cases that the proposed model is efficient in the practical prediction of chaotic time series.