ADAPTIVE CONTROL METHOD FOR VARIOUS PLASTIC CHARACTERISTIC USING FUZZY CLUSTERING

摘要

In the automation of proficient skills, automation equipment is required to adapt to varying conditions over long use. We propose a new in-line learning algorithm utilizing fuzzy clustering. In order to achieve precise straightness, steel bars are reformed using bending machines. Bending machine have been operated by experts. In this case, highly precise plastic deformation to micron orders is required. In this paper, we propose a new plastic deformation control method based on n-th power hardening law. This new method calculates the deformation load in accordance with average plastic characteristics such as n-value and F-value. However the disadvantage is that adaptability for each work piece is not satisfactory. The mechanical characteristics of steel bars vary slightly for each work piece. And over the long term, material processing conditions will vary, changing the characteristics sharply. This proposed learning algorithm can adapt for differences in plastic characteristics for each work piece by getting and analyzing abundant data collected automatically on the production line. The results in simulation tests have been highly satisfactory. Also, a reformation precision experiment yielded values within the target range.