The influence of the wavelet filter in the parameters extraction for signal classification: An experimental study.

Abstract

This paper treats the influence of the wavelet filters on the parameter extraction
for signal classification. Therefore, a database with vibration signal failures of an induction motor in stationary case was utilized. The classification is done by techniques based on the wavelet packet transform and dimension reduction, using the principal component analysis (PCA). Different wavelet filters with different supports are used for comparison. The use of signals from a validated experimental bench shows that the results of the classification model has a high precision.