Approximate entropy as acoustic emission feature parametric data for crack detection

Abstract

This paper addresses a new feature parametric data of acoustic emission (AE) signals for crack monitoring based on the approximate entropy (ApEn). ApEn is a ‘regularity statistic’ that quantifies the unpredictability of fluctuations in a time series such as an instantaneous AE time series. Recently, ApEn has been developed as a statistic suited to quantify regularity and complexity, with potential applications to a wide variety of physiological and clinical time series data. Measures of AE variability and nonlinear complexity could be used as indications of crack initiation and/or propagation. A typical AE signal was simulated as a damped sinusoidal function with supplementary noise. ApEn values of such simulated signals were calculated and compared. The results show that AE time series containing many repetitive patterns have relatively low ApEn values. Simulated results were compared with those obtained experimentally through AE measurements obtained from cracks in steel tube tests. By analysing the complexity of these AE signals, we can conclude that ApEn can obviously distinguish the AE signals. Therefore, the ApEn values can contribute as prognostic parameters to complement traditional AE parameters.

Keywords:

• acoustic emission
• approximate entropy
• feature value