Microphone Arrays for Acoustic Detection during Laser Welding of Cold-Rolled Steel Strip

In order to achieve the quality control for the cold-rolled steel strip, it was required to monitor the laser welding processing. Acoustic signal detecting based on microphone is wildly applied in the welding processing. However, traditionally single microphone was limited by hostile background noise in the realistic industry environment. In this paper, a microphone array consisting of 7 microphones was employed to sample the acoustic signal generated by the laser welding. In order to reduce the background noise and extract the clean welding acoustic signal, the Constant Modulus Algorithm was used to the array beam former. Meanwhile, power spectrum density estimation and wavelet packet transform methods were applied to analyze the original acoustic signal and the beam former output signal. By comparing the two signals, it was obviously shown that the beam former output signal could effectively reduce the power spectrum density of the background noise at lower frequency bands. And the energy contribution was concentrated in the 15 kHz frequency range. Meanwhile, the frequency characteristic of the acoustic signal during the penetration welding processing by the modulated laser beam was explained.