Experimental study of quality monitoring system integrated with a microphone array in laser microlap welding

In microlap welding, a real-time welding quality monitoring system is crucial to the identification of low strength joints caused by the unreliable contact between two layers of stainless metal sheets. In this study, a multispacing configured microphone array filter was designed and applied to a sound-based quality monitoring system for laser microlap welding, and the filter's performance was evaluated to improve the reliability of the developed monitoring system when collected sound signals are contaminated by the noises generated around a welding site. In the experimental setup, joint strength was modulated by controlling the clamping conditions of the fixture and changing the welding location. The results indicated that noises contaminated the signals obtained from single microphones and reduced classification rates by up to 25% when time domain features were adopted. Through the application of the proposed microphone array in the developed monitoring system, the classification rate of weld quality can be improved to a level resembling that observed when artificial noise is not applied to the system.