Portal de Periódicos da CAPES

Feature Extraction of Acoustic Signals Based on Complex Morlet Wavelet

2011; Elsevier BV; Volume: 15; Linguagem: Inglês

10.1016/j.proeng.2011.08.088

1877-7058

Ping He, Pan Li, Sun Hui-qi,

Grey System Theory Applications

Abstract This article studies feature extraction of acoustic signals based on complex Morlet wavelet. Since, parameter optimization is the important and difficult point of complex Morlet wavelet application. In this article, a new 2-parameter process optimization algorithm is proposed, i.e., ascertaining a parameter, getting the optimized value of the other parameter, and then repeating the process until the system performance index is satisfied. Based on the optimal wavelet scalogram obtained from applying wavelet transform to typical acoustic emission (AE) signal, the region segmented location method is designed to judge exactly the number of AE signal feature frequencies whose exact values are then calculated. By this way, the error induced by misjudgment and misreading can be avoided effectively. Finally, the optimal complex Morlet wavelet obtained from 2-parameter process optimization algorithm is compared with the extraction results of the traditional Morlet wavelet on acoustic signal feature. Simulation results show that such methods can improve precision of acoustic signal feature extraction and have good engineering value.