Set-to-Set Iterative Learning Control for Non-Repetitive Disturbances

Iterative learning control (ILC) methods which track sets rather than a reference throughout an iteration, namely region-to-region (RTR) ILC and set-to-set (STS) ILC, have assumed that external disturbances are purely repetitive. However, in real-world applications, non-repetitive disturbances will likely be present, both on the state and output channels of the plant. These non-repetitive disturbances lead to challenges in ensuring that the sets for tracking will be achieved throughout each iteration. While STS ILC has been shown to outperform RTR ILC for purely repetitive disturbances, this letter develops a novel STS ILC architecture which incorporates feedback control and tightening of the sets to ensure that the sets are achieved despite the unknown, non-repetitive disturbances. A conservative bound for the necessary tightening of the sets is derived and a simulation-based case study demonstrates the novel STS ILC’s ability to achieve the sets while outperforming alternative ILC approaches.