Feedforward element design using learning controller for precision control of linear synchronous motor with nonlinear characteristics
Kaiji Sato
Abstract
This study presents a time-invariant feedforward (FF) element design for the high-speed and high-precision tracking control of an ultrahigh-acceleration, high-velocity linear synchronous motor (LSM). The linear motor can generate an acceleration greater than 70 G (= 686 m/s2) and move at a velocity above 10 m/s. To take advantage of this performance and realize high response, the design and usage of suitable FF elements is crucial. However, as the LSM includes highly nonlinear characteristics, it is difficult to provide an exact dynamic model for FF design. To overcome this problem, a control system with a learning controller (LC) as the FF element has been designed previously, demonstrating high-precision and high response motion. However, the motion performance can be achieved only with sufficient pre-learned motions. The integrator and the disturbance observer that were effective in suppressing disturbances were removed from the control system. In addition, the control system has some FF time-invariant elements along with the LC. This study proposes a design method for easy design of all FF elements using an LC. The designed FF elements are time invariant and are used with an integrator and a disturbance observer, without pre-learning. Using the proposed method, two sets of time-invariant FF elements are designed. The performances of two control systems, which include a set of time-invariant FF elements for each, and a simple disturbance observer are experimentally examined and compared with two previously designed control systems. Experimental results demonstrate that the performance of one of the control systems with a set of time-invariant FF elements designed in this study and a disturbance observer is good and almost comparable with that of the previously designed control system with high-precision and high response motion.