Data-driven approach to iterative learning control via convex optimisation

Abstract

A new data-driven iterative learning control methodology is presented which uses the frequency response data of a system in order to avoid the problem of unmodelled dynamics associated with low-order parametric models. A convexoptimisation problem is formulated to design the learning filters such that the convergence criterion is minimised. Since thefrequency response data of the system is used in obtaining these filters, robustness is ensured by eliminating the uncertainty inthe modelling process. The effectiveness of the method is illustrated by considering a case study where the proposed designscheme is applied to a power converter control system for a specific accelerator requirement at CERN.