Modeling Of Underwater Wet Welding Process Based On Visual And Arc Sensor

Bo Chen,Jicai Feng

DOI: https://doi.org/10.1108/IR-03-2014-0315

2014-01-01

Abstract:Purpose - The purpose of this paper was to use visual and arc sensors to simultaneously obtain the underwater wet welding information, and a weld seam-forming model was made to predict the weld seam's geometric parameters. It is difficult to obtain a fine welding quality in underwater welding because of the intense disturbances of the water environment. To automatically control the welding quality, the weld seam-forming model should first be established. Thus, the foundation was laid for automatically controlling the underwater welding seam-forming quality.Design/methodology/approach - Visual and arc sensors were used simultaneously to obtain the weld seam image, current and voltage information; then signal algorithms were used to process the information, and the back propagation (BP) neural network was used to model the process.Findings - Experiment results showed that the BP neural network model could precisely predict the weld seam-forming parameters of underwater wet welding.Originality/value - A weld seam-forming model of underwater wet welding process was made; this laid the foundation for establishing a controller for controlling the underwater wet welding process automatically.