Multi-resolution analysis of passive cavitation detector signals

Abstract

Passive cavitation detectors are widely used for measuring acoustic emissions from cavitating bubbles. Acoustic emissions related to the dynamics of oscillating bubbles contain complex time and frequency domain information. Signal processing techniques traditionally used to analyse transient and stationary signals may be of limited value when analysing such acoustic emissions. This paper describes a multi-resolution approach developed for processing acoustic emissions data. The technique consists of the combination of a discrete wavelet transform and of the statistical and spectral analysis to extract cavitation features. These features include broadband emissions and harmonic, sub-harmonic and ultra-harmonic information. The implementation of the technique on experimental datasets demonstrates that this approach provides detailed information about key features of the acoustic signal, especially in complex situations where different types of cavitation occur simultaneously. Furthermore, statistical metrics used in this technique can provide a quantitative means for classifying signatures of cavitation, particularly the broadband segment of the spectrum created by inertial cavitation, which constitutes novel work.