Nonlinear Identification of Laser Welding Process

Abstract

It has been well recognized that the weld pool geometry plays a critical role in the welding process. In this study, a diode laser welding control system is established. The authors performed a series of open-loop experiments to investigate the interaction and the correlation among of the laser energy, the welding speed, and the weld pool geometry. A digital camera with a high speed shutter is equipped to take pictures of the weld pool in real time. Custom computer vision software and image processing programs are applied to acquire the surface width of the weld pool. Based on the experimental study, the authors propose an applicable method to identify a SISO nonlinear continuous model for the established diode laser welding process. The identified nonlinear model takes the reciprocal of the welding speed as the input and the surface width of the weld pool as the output. To validate the model, the authors conduct further experiments including step and PRTS (pseudo-random ternary signal) responses. In addition, the authors simulate the nonlinear model with same inputs and check the data agreements between the simulated results and the experimental data. The validation results confirm the applicability of the proposed nonlinear identification method and show that the model can successfully predict the surface width of weld pool for later use.