Tight butt joint weld detection based on optical flow and particle filtering of magneto-optical imaging. (November 2017)

• Record Type: Journal Article
• Authors: Gao, Xiangdong; Mo, Ling; You, Deyong; Li, Zhuman
• Is Part Of: Mechanical systems and signal processing. Volume 96(2017)
• Page Start: 16
• Page End: 30
• Publication Date: 2017-11
• Subjects: Tight butt joint weld -- Magneto-optical image -- Optical flow method -- Particle filter; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2017.04.001
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: Micro-gap weld was detected by magneto-optical imaging (MOI) technology. Features of magnetic field distribution were obtained using optical flow method. Particle filter was integrated with optical flow for weld position prediction. Optical flow information of MOI is corresponding to actual weld deviation. Abstract: It is a challenge to detect the weld position during tight butt joint laser welding in that the tight butt joint is non-grooved and invisible. This paper proposes a novel method for tight butt joint weld detection based on magneto optical imaging. Two pieces of weldment were magnetized by an electromagnet so that they could show magnetic N and S polarity respectively. When a polarized light was projected on a magneto-optical film, it would deflect due to magneto-optical effect. In accordance with magneto field distribution, an image formed on the visual sensor. A transition zone of magnetic field distribution which corresponded to the butt joint could be shown in a magneto optical image of weldment. Variation features of magnetic field distribution were obtained by using image sequence optical flow method, and a particle filter was integrated to make an accurate prediction on weld position. Weld position was obtained by calculating the maximum value of optical flow intensity in the vertical direction, and a particle filter was used to realize the accurate prediction on weld position. Experimental results showed that the proposed method was effective in detection of weld and realizing weld seam tracking.