Optimal Time-Varying ILC Design to Monotonically Minimize Converged Error
This paper presents a design method for Iterative Learning Control (ILC) algorithms using time-varying Q-filters. The design of an optimal bandwidth profile for a given plant model is formulated as a constrained minimization problem. The resultant time-varying ILC algorithm generates the lowest converged error norm possible while guaranteeing monotonic convergence. The time-varying ILC background, problem setup to optimize the time-varying Q-filter bandwidth, as well as results obtained using computational methods are presented. A simulation example is used to demonstrate the potential benefits of the algorithm in comparison with LTI ILC. Lastly, experimental validation is provided by application of the ILC algorithm developed here on a Microscale Robotic Deposition system for precision motion control.