Model-less FIR repetitive control with consideration of uncertainty

Repetitive control (RC) is used to track and reject periodic exogenous signals by including a model of a periodic signal in the feedback path. The performance of RC can be improved by also including an. The accuracy of this filter is the main limitation to the RC bandwidth. The bandwidth is typically limited with a robustness filter - often a low-pass filter which attenuates the model at high-frequencies where the model-mismatch often occurs. In this paper, two robustness filter designs are compared. The first design is a brick-wall low-pass filter commonly used in the literature. The second design is based on the uncertainty between the inverse plant response filter and the measured response of the system. Experimental results demonstrate that the proposed method outperforms the traditional brick-wall filter approach.