Abstract: To track seam automatically during underwater flux-cored arc welding(UFCAW), a vision-based seam tracking system has been studied. This seam tracking system consists of a vision sensing system, an image processing and seam recognizing system, a control system and executive unit. The edge of the weld line is precisely detected by a wavelet transform method. The seam position error is calculated by a seam position recognition algorithm based on the binary image. Based on the theoretic analyze of the structure model of the vision-based seam tracking system, a control rule self-adjust fuzzy controller is designed. The factors of decreasing the precision of the seam tracking system are analysised, and some measures are adopted to improve the tracking precision. The tracking experiments results of the UFCAW confirm that the control rule self-adjust fuzzy controller meet the demand of the tracking precision. In the tracking trials of sidelong, reflex and curve weld line, the system can all meet the demand of the tracking precision. With the improved mini drain water cap, the tracking experiments of the underwater local dry welding are carried. The results indi-cate that this process can not only meet the demand of improve welding quality but also satisfy the tracking results.

Key words: Fuzzy control, Image processing, Mini drain cap, Seam tracking, Underwater welding, Visual sensing