Towards the Automatic Tuning of Linear Controllers Using Iterative Learning Control Under Repeating Disturbances

In this paper, iterative learning control (ILC) is proposed as an alternative approach that can be used to simplify the design of linear controllers for rejecting repetitive disturbances affecting a system. Since ILC is a data-driven method, its usage allows rejecting repetitive disturbances without a priori knowledge of their nature. The benefits of ILC appear especially when a system is subject to complex repetitive disturbances. This is simply because it would require a control designer to spend much more efforts to obtain similar results applying a state augmentation by building models for the same disturbances (Internal-Model-Principle based control). Accordingly, this paper first shows the equivalence of performing state augmentation and applying ILC in case of a simple sinusoidal repetitive disturbance. Next, a workflow named Learning Based Controller Tuning (LBCT) is proposed to simplify the parameter tuning of linear controllers under repetitive disturbances. The feasibility of LBCT is analysed by testing the ILC on a system subject to complex repetitive disturbances in two different forms: linearly combined sinusoidal signals and non-linearly combined sinusoidal signals. The results demonstrate that ILC can successfully learn the required controller parameters to reach a good rejection performance against the repetitive disturbances without a need of modelling them. This supports the fact that the tuning of linear controllers may be automated by integrating an ILC to the system.