Intelligent Control System Design for Precision Positioning

• Main Authors: Pai-Yi Huang, 黃百毅
• Other Authors: Yung-Yaw Chen
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/48373080487197249456

博士 === 國立臺灣大學 === 電機工程學研究所 === 89 === This dissertation is focused on the design problem for precision positioning system. For experimental requirements, two ballscrew table systems were implemented for investigations. Integrated by IBM-PC personal computer due to their open and flexible architecture, the experimental systems were suitable for versatile advanced controller design practices. Mechanical devices usually come with undesirable nonlinearities such as frictions, backlashes and saturation. Unfortunately, these nonlinear characteristics become more important as the positioning resolution increased. To perceive more detailed nonlinear behaviors and feedback information, high-resolution laser scales were attached to the ballscrew table for capturing sub-micron movements. With nonlinear friction behaviors, common identification algorithms based on linear theories are not qualified. The dissertation proposes a modified identification algorithm works for mechanical system with friction nonlinearities. The accuracies of the identified mathematical model are greatly improved. Secondly, Intelligent control design methodologies based on genetic algorithms were studied. Not knowing any system knowledge, the learning processes gain experiences from trials guided by genetic algorithms. Based on given performances goal, the control parameters were optimally justified. From experimental results, the proposed method is quite successful. The positioning resolution achieved 10nm (0.01mm), same as the resolution for laser scale, and no steady state errors were observed.