Iterative learning control of high-acceleration positioning table via sensitivity identification

Through the convergence analysis of the iterative learning control, it can be seen that the inverted model of the sensitivity can be used as the update law of iterative learning controller. However, due to the little knowledge in modeling uncertainty and in feedback controllers embedded in drivers with hard-to-obtain parameters, it is difficult to directly calculate the transfer function of the sensitivity. To solve this problem, this article presents sensitivity identification–based iterative learning controller for a high-acceleration positioning table actuated by the voice coil motors. In the sensitivity identification process, a frequency sweep signal was exerted to the closed-loop system as the equivalent disturbance, and the perturbation of the tracking error was gathered as output data. Then, the autoregressive model of the sensitivity was gained conveniently. Finally, the proposed sensitivity identification–based iterative learning controller was implemented in the experimental setup with the maximum acceleration of 5.3 g. The experimental results show that the tracking error of the X- and Y-axes decreased from 530 and 2480 µm to 6 and 18 µm, respectively, after the third iteration of the proposed iterative learning controller method, and the identified sensitivity can be easily designed in the iterative learning controller to greatly improve the tracking performance.