A neural network and multiple regression method for the characterization of the depth of weld penetration in laser welding based on acoustic signatures

The need for the control of the depth of weld penetration has been and remains of a long term interest in the automated welding process. In this study, the relationship between the depth of weld penetration and the acoustic signal acquired during the laser welding process of high strength steels is investigated. The acoustic signals are first preprocessed by the spectral subtraction noise reduction method and analyzed both in the time domain and frequency domain. Based on this analysis, two algorithms are developed to acquire the acoustic signatures. The acquired acoustic signatures are then used to characterize the depth of weld penetration by using a neural network and a multiple regression analysis. The results show that the acoustic signatures can characterize and predict the depth of weld penetration well under different laser welding parameters.