An Optimization-based Approach for Design of Iterative Learning Controllers with Accelerated Rates of Convergence

In this paper, a new technique for designing iterative learning controllers has been proposed. The control update law is based on the minimization of a quadratic cost function. The control input update law is time varying. It is shown that the proposed controller has monotonic super-linear convergence. A systematic robustness and performance analysis has been presented to evaluate the effectiveness of the controller. The effect of different design parameters on the closed loop system performance, robustness, learning rate is investigated. The relationship between three critical indices for evaluation of ILC's - performance, rate of learning and robustness - has been studied and inferences drawn about the trade-offs. Numerical simulations verify the results.