Automatic evaluation of EDM electrode wear via integration of image segmentation and 3D registration (IS-3DR)

Electrode wear (EW) during the electrical discharge machining (EDM) process determines surface topography accuracy. In this paper, an automatic evaluation method is proposed by integrating image segmentation and three-dimensional (3D) registration (IS-3DR) to evaluate the EW more accurately. First, the original two-dimensional (2D) image and the 3D point cloud were derived from a laser microscope image, and the 2D worn area was determined by automatic recognition and segmentation. Then, 3D registration between the 3D point cloud and the 2D worn area was utilized to qualitatively define the 3D EW area. Finally, the 3D EW volume was obtained quantitatively through micro-element calculation based on the 3D EW area and the shape fitting of the original electrode. The results showed that the intersection over union (IOU) of image segmentation was higher than 97.5%, compared with the manually labeled image segmentation method. Compared with IS-3DR, the random error of the two-dimensional evaluation method could reach 27.08%, indicating higher reliability and stability of IS-3DR. The EDM experiment was carried out to investigate the effect of parameters on EW. The experimental results showed that the effects of current and servo voltage were most significant, followed by pulse width.