In-process monitoring and control of weld pool geometry during laser welding of magnesium and titanium alloys

In the laser keyhole welding process, the weld pool behavior is important to determining welding quality, especially with use of assisted gas. To better monitor and control the welding process, accurate extraction of the weld pool boundary and penetration depth is required.  This work presents a weld pool monitoring system for detecting welding pool boundary and penetration depth based on an off-axial green illumination laser and a coaxial visual monitoring system and a robust estimator for keyhole depth.  The proposed edge detection algorithm is derived from the maximum grey-value gradient searching approach according to the difference of illumination reflecting features in captured images, which efficiently reduces the effects of imaging noises. Using the radial basis function based observer, the keyhole depth is effectively estimated and used to monitor the stability of key hole welding process.  Experiments and image processing are conducted under various welding conditions during laser welding of magnesium and titanium alloys with a fiber laser.  Extracted geometries of weld pool are validated by comparisons with a three-dimensional transient model developed and validated to investigate the dynamics of weld pool and keyhole.   Finally, a robust control strategy is developed and implemented to control the keyhole depth during the welding process and is shown to greatly improve the quality and consistency of the weld.