NARX prediction of some rare chaotic flows: Recurrent fuzzy functions approach

• A new method is proposed for prediction of chaotic time series.
• This method is based on novel recurrent fuzzy functions (RFFs) approach.
• Some rare chaotic flows are used as test systems.
• The new method shows proper performance in short-term prediction.
• It also shows proper performance in prediction of attractor's topology.

The nonlinear and dynamic accommodating capability of time domain models makes them a useful representation of chaotic time series for analysis, modeling and prediction. This paper is devoted to the modeling and prediction of chaotic time series with hidden attractors using a nonlinear autoregressive model with exogenous inputs (NARX) based on a novel recurrent fuzzy functions (RFFs) approach. Case studies of recently introduced chaotic systems with hidden attractors plus classical chaotic systems demonstrate that the proposed modeling methodology exhibits better prediction performance from different viewpoints (short term and long term) compared to some other existing methods.