Local eigenfunctions based suboptimal wavelet packet representation of contaminated chaotic signals

We report a suboptimal wavelet packet representation (SWPR) of signals emanating from a chaotic attractor contaminated by low levels of noise. Our method-geared towards choosing a suboptimal scaling function to parsimoniously represent the signal-involves extracting local eigenfunctions using artificial ensembles generated from a psendo-probability space, and using the extracted local eigenfunctions to develop a suboptimal scaling function. The application of our novel representation method to actual acoustic emission (AE) signals, sampled as time-series data (TSD) from the turning process, reveals the superiority of these methods over the existing signal representations.