NARMAX-Model-Based Time Series Modeling and Prediction: Feedforward and Recurrent Fuzzy Neural Network Approaches

The nonlinear autoregressive moving average with exogenous inputs (NARMAX) model provides a powerful representation for time series analysis, modeling and prediction due to its strength to accommodate the dynamic, complex and nonlinear nature of real time series applications. This paper focuses on the modeling and prediction of NARMAX-model-based time series using the fuzzy neural network (FNN) methodology with an extention of the model represention include feedforward and recurrent FNNs. This paper introduces and develops a efficient algorithm, namely generalized fuzzy neural network (G-FNN) learning algorithm, for model structure determination and parameter identification with the aim of producing improved predictive performance for NARMAX time series models. Experiments and comparisons demonstrate that the proposed G-FNN approaches can effectively learn complex temporal sequences in an adaptive way and outperform some well-known existing methods.