2001 年 32 巻 1 号 p. 21-26
Brake mounted pneumatic cylinders are used for the purpose of transferring machinery for many industries because of their low cost and their cleanliness. It is important to improve the positioning accuracy of pneumatic cylinder for matching to the demand of every industry. Therefore many researchers have studied learning algorithms to improve accuracy and to estimate the overrun error by changing the operating time of the valves. However some trials are needed to learn the characteristics, and load change is not taken into account at all. In this paper a new learning algorithm based on the braking model is introduced and is confirmed to realize not only high learning speed but quick adaptation to load change as well. It is found through various field tests that good positioning has been achieved and that this method is practical and effective.