Pages 429-430
This paper deals with sensing of weld pool condition from reverse side and control of penetration in fixed pipe welding. In order to realize full-automatic welding of fixed pipes under high pressure atmosphere, it is important to control the back bead width properly. Accordingly, a method controlling the back bead width from the reverse side is proposed. As it is difficult to measure the back bead width directly, it is estimated by analyzing the molten pool shape and the dimensions. Furthermore, Neural Network is applied so as to estimate the relations among the parameters of the weld pool shape, welding conditions and the penetration of weld. The back bead width is controlled by controlling the welding current estimated from the output of the Neural Network. As the result of welding control experiments. The effectiveness of the proposed system in the penetration control of fixed pipes is verified.