Audio sensing and modeling of arc dynamic characteristic during pulsed Al alloy GTAW process
na lv,yanling xu,zhifen zhang,jifeng wang,bo chen,shanben chen
DOI: https://doi.org/10.1108/02602281311299680
2013-01-01
Sensor Review
Abstract:Purpose - The purpose of this paper is to study the relationship between arc sound signal and arc height through arc sound features of GTAW welding, which is aimed at laying foundation work for monitoring the welding penetration and quality by using the arc sound signal in the future. Design/methodology/approach - The experiment system is based on GTAW welding with acoustic sensor and signal conditioner on it. The arc sound signal was first processed by wavelet analysis and wavelet packet analysis designed in this research. Then the features of arc sound signal were extracted in time domain, frequency domain, for example, short-term energy, AMDF, mean strength, log energy, dynamic variation intensity, short-term zero rate and the frequency features of DCT coefficient, also the wavelet packet coefficient. Finally, a ANN (artificial neural networks) prediction model was built up to recognize different arc height through arc sound signal. Findings - The statistic features and DCT coefficient can be absolutely used in arc sound signal processing; and these features of arc sound signal can accurately react the modification of arc height during the GTAW welding process. Originality/value - This paper tries to make a foundation work to achieve monitoring arc length through arc sound signal. A new way to remove high frequency noise of arc sound signal is produced. It proposes some effective statistic features and a new way of frequency analysis to build the prediction model.