Prediction of the Chaotic Response of Electrochemical Oscillators

Abstract

A prediction method is proposed for one-step and multi-step forecasting of the chaotic response. The method is based on the determination of the transition matrix from experimental data by a linearization procedure. The procedure is applied to various chaotic time series obtained from the electrochemical oscillator Fe/H₂SO₄. The one-step prediction results are good in the case of smooth time series. Regions of poor prediction are analyzed and it is found that the forecasting is limited due to the presence of a high frequency component in the time series causing a virtual increase of the measuring interval, Δ t. The method is also applied successfully for multi-step prediction. The performance is increased by implementing a false neighbors detection routine.