Nonlinear Time Series Prediction Based on Lyapunov Theory-Based Fuzzy Neural Network and Multiobjective Genetic Algorithm

This paper presents the nonlinear time series prediction using Lyapunov theory-based fuzzy neural network and multi-objective genetic algorithm (MOGA). The architecture employs fuzzy neural network (FNN) structure and the tuning of the parameters of FNN using the combination of the MOGA and the modified Lyapunov theory-based adaptive filtering algorithm (LAF). The proposed scheme has been used for a wide range of applications in the domain of time series prediction. An application example on sunspot prediction is given to show the merits of the proposed scheme. Simulation results not only demonstrate the advantage of the neuro-fuzzy approach but it also highlights the advantages of the fusion of MOGA and the modified LAF.