Research on identification of the corner point of 90° weld based on multi-sensor signal fusion technology

It is of great economic benefit to realize the automatic welding of 90° welds, and difficulty of automatic welding of 90° welds is the machine identification of the corner points of 90° welds. The math model of welding-torch pose was built, and the identification algorithms of wire extension, welding deviations, and tilt angles of welding torch have been studied respectively. In this paper, three pattern recognition algorithms are studied, and they are used to identify the corner points of 90° welds based on variation characteristics of wire extension, welding deviations, and tilt angles of welding torch, respectively. At the same time, accuracy values of three recognition algorithms have been researched respectively. On this basis, the recognition method of the corner points of 90° welds based on multi-sensor signal fusion technology has been studied in three-dimensional directions, and the calculation method for the total accuracy of corner point identification has been analyzed. At last, the corner point recognition experiments of 90° welding seam were carried out, and the change curves of wire extension, welding deviations, and tilt angles have been analyzed respectively. The welding results indicated that the corner point of 90° welding seam could be identified stably, accurately, and quickly by using the methods in this article.