Analysis of acoustic signals detected from different locations during laser beam welding of steel sheet

Abstract

Acoustic emission is one of the most promising techniques to be recently developed for in-process monitoring of laser beam welding. However, signal analysis is still an area that requires further investigation in order to enhance the potential adoption of acoustic emission. The main issue is the reliability of this technique because the weld quality cannot be described by a simple function of acoustic emission. In other words, we need to study whether or not the acoustic signals contain sufficient information to identify or classify the welds. Air-borne acoustic signals from various location around the welding spot have been detected. Information measure (entropy) has been applied to the acoustic signals in the frequency domain and the frequency-based pattern recognition concepts have been used in analyzing these acoustic signals in an effort to distinguish signal sources of optimal welds from those of bad welds due to partial penetration and overheated. Analysis results show that acoustic signals detected from different locations give different accuracy of classification of welds which is predicted by the information measures.Acoustic emission is one of the most promising techniques to be recently developed for in-process monitoring of laser beam welding. However, signal analysis is still an area that requires further investigation in order to enhance the potential adoption of acoustic emission. The main issue is the reliability of this technique because the weld quality cannot be described by a simple function of acoustic emission. In other words, we need to study whether or not the acoustic signals contain sufficient information to identify or classify the welds. Air-borne acoustic signals from various location around the welding spot have been detected. Information measure (entropy) has been applied to the acoustic signals in the frequency domain and the frequency-based pattern recognition concepts have been used in analyzing these acoustic signals in an effort to distinguish signal sources of optimal welds from those of bad welds due to partial penetration and overheated. Analysis results show that acoustic signals detecte...