1992 Fiscal Year Final Research Report Summary
Fuzzy Adaptive Control of Weld Pool in Pulsed MIG Welding
| Project/Area Number |
03555159
|
|---|
| Research Category |
Grant-in-Aid for Developmental Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
溶接工学
|
|---|
| Research Institution | Saitama University |
|---|
Principal Investigator |
OHSHIMA Kenji Saitama University, Department of Engineering, Professor, 工学部, 教授 (70026087)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KANJOU Yoshihiro NKK, Engineering Research Center, 応用技術研究所
SUGITANI Yuuji NKK, Engineering Research Center, 応用技術研究所, 溶接研究室室長
KUBOTA Takefumi Himeji Institute of Tachnology, Assistant Professor, 工学部, 助教授 (40047585)
|
|---|
| Project Period (FY) |
1991 – 1992
|
| Keywords | Pulsed MIG Welding / Butt Welding / Back Bead Control / Weld Pool / Penetration Shape / Fuzzy Theory / Adaptive Control / Neuro Network |
|---|
| Research Abstract |
It is important to construct the intelligent welding robots by means of the numerical model of the plant, the fuzzy controller, and the neural network, in order to obtain a good quality of the welding results. The authors have been studied the construction of the intelligent welding robots with CCD cameras to apply to the pulsed MIG welding. The realization of the back bead control can be tried by using the result of the research.
A new method is developed to estimate the penetration depth of the weld pool from the information of the welding side, i.e., the surface shape of the weld pool, the heat inputs, and state of the groove gap. First, the observation of both the weld pool surface and the groove gap is realized with the CCD cameras. Next, the relationship between the pulsation phase of the current and the shutter timing of the CCD cameras is discussed to obtain the clear image of the groove gap and weld pool. While the shutter opens, let the welding current be decreased to avoid the affection of the arc length. The computer processes the image to measure the pool surface shape and the width groove gap.
The method of controlling the back is constructed by using the numerical model which represents the state of a plant. The authors propose the neural networks to describe the state of the plant. The welding current is controlled with the fuzzy controller so as to keep the desired penetration depth constant. Since it takes the time to process the image and to determine the welding current, the plant has the time delay. The method is proposed for designing the fuzzy controller from the knowledge of the modern control theory. The validity of the neural network and the fuzzy controller is verified from the experimental results.
|
