|Budget Amount *help
¥3,800,000 (Direct Cost : ¥3,800,000)
Fiscal Year 1998 : ¥1,400,000 (Direct Cost : ¥1,400,000)
Fiscal Year 1997 : ¥2,400,000 (Direct Cost : ¥2,400,000)
In this year, a robust control system for recognizing and controlling the penetration of the weld in butt welding of thin steel plates was constructed. In this system, the recognition of the molten pool shape and detection of peculier frequency of molten pool oscillation were performed. Furthermore, in order to obtain sound weld regardless change of welding conditions, such as groove shape, gap of groove and so on, system to diagnose the welding phenomena during welding and to control welding conditions were constructed.
Main results obtained are summarized as follows.
(1) An intelligent welding robot system with vision sensors that observe the shape and dimensions of the weld molten pool, including length, width and area was constructed.
(2) An adaptive control system to optimumly control the penetration of weld by controlling the dimensions of the weld molten pool was constructed.
(3) An adaptive control system using Artificial Neural Network was proposed. lt has an input layer of six units for length, width, area, welding speed, welding current and root gap of welding line, an output layer of seven units that indicates change of welding current, and an hidden layer of eight units.
(4) Applying pulse current as welding current and detecting oscillation of arc voltage during base current time, the frequency of the oscillation of molten pool can be detected.
(5) A system to control penetration welding using the principle to detect the peculiar frequency of molten pool was constructed.
(6) Mild steel plate welding experiments using TIG arc welding method were performed. The effectiveness of the constructed welding control systems was confirmed.