Local filtering of noisy nonlinear time series

We investigate the use of different local nonlinear modelling and nonlinear filtering techniques to clean a noisy time series obtained from a deterministic chaotic systems. The methods are tested on data from the Ikeda map and the Mackey-Glass delay differential equation. We test the results of the filtered times series using the correlation dimension statistic and SNR gain. In all cases we see that local filtering has produced a new time series which is more consistent with the original clean time series. (C) 1998 Published by Elsevier Science B.V.