ニューラルネットワークモデルによるTIG溶接溶融池幅と冷却時間の制御

抄録

A good quality of weld is obtained by controlling the weld pool width and the cooling time. The cooling time t815 (from 800 [°C] to 500 [°C]) is related to the hardness of welding metal. The welding metal is influenced by rapid heating and rapid cooling. Therefore, various kinds of crystalline structures are formed by cooling time. It is important how to select cooling time for gaining good crystalline structure and the proper hardness after the welding. The cooling time is influenced by heat input (arc voltage U, welding current I, welding speed S, heat input efficiency η).
The numerical simulations in the TIG welding of pipe are performed by using the finite difference method on the basis of heat conduction theory. The weld pool width and the cooling time have been calculated for various kinds of welding current and welding speed. The above data are used as the teaching data of neural network model. A weld pool model is constructed by neural network. The welding speed and the welding current are determined to obtain the desired pool width and cooling time by using the neural network model.