Neural network compensation for micro-gap weld detection by magneto-optical imaging

A BP neural network was proposed to compensate detection accuracy of micro-gap weld seam (seam width less than 0.1 mm). The butt welding of low carbon steel was carried out with laser welding. The magnetized weldments were detected by using a magneto-optical sensor and the magneto-optical images of weld seam were captured. By using BP neural network and processing weld seam magneto-optical images with low contrast and strong magnetic field noises, the weld seam center position could be extracted accurately. Experimental results at different welding speeds indicated that the absolute mean error of weld seam is about 0.015 mm, and the error measured by BP neural network decrease about 28% than that detected directly by magneto-optimal imaging. The compensation technique for magneto-optical imaging by BP neural network can be applied to detect the micro-gap weld seam accurately. It provides a novel approach for automatic identification and tracking of the micro-gap weld during the laser welding.