Detection method of austenitic stainless steel laser welding quality based on the AE signal

Austenitic stainless steel is prone to defects during laser welding, which makes online monitoring essential for ensuring weld quality. Acoustic emission (AE) signals generated during laser welding provide real-time and highly sensitive information for weld quality detection. However, background noise, signal non-stationarity and multi-source interference greatly limit the direct application of AE technology for online defect detection. In this study, an AE testing platform for stainless steel laser welding was established to collect AE signals during the process. An adaptive resonance demodulation method based on the optimisation of variational mode decomposition (VMD) parameters using the sparrow search algorithm (SSA) is proposed to preprocess the AE signals. The time-domain waveform and spectrum of the extracted envelope signal reflect the distribution of laser energy during welding and can be used to evaluate weld formation quality. The sample entropy of AE signals was calculated, with higher entropy indicating poorer weld surface formation and lower entropy indicating better surface quality. Further destructive testing revealed that weld regions with higher sample entropy exhibited internal porosity. These results demonstrate that sample entropy effectively characterises laser welding quality. The proposed method offers an effective approach for the quantitative evaluation of laser welding quality in austenitic stainless steel.