An H∞ Synthesis of Robust Current Error Feedback Learning Control

Abstract

Conventionally, iterative learning control law is updated using past error information. As a result, the controller is effectively open-loop, apart from being unable to control unstable systems, it also does not share the robustness properties of feedback systems. It is proposed that current error feedback should be used to update the learning control law instead. We present a systematic design procedure based on the H∞ control theory to construct a robust current error feedback learning control law for linear-time-invariant, and possibly unstable, systems. The optimal design will ensure: 1 That the closed-loop system is stable; 2 that the convergence rate is optimum about the nominal plant; 3 robustness in the presence of perturbed or unmodeled dynamics, or nonlinearity.