Data-Driven Based Iterative Learning Control for Unknown Linear Discrete-Time System by Adaptive System Identification

Abstract

This paper considers an iterative learning control (ILC) scheme for unknown linear discrete-time system. ILC updates the system input using error information from previous iteration to sequentially improve the tracking performance. The asymptotic convergence may not be guaranteed when the system model information is uncertain. In this paper, we introduce the adaptive Fourier decomposition (AFD) algorithm to deal with the system identification problem. This adaptive approximation algorithm estimates a system representation using input/output data only. The learning gain of ILC satisfying the convergence condition is obtained according to the estimated system parameters. The effectiveness of system approximation and the tracking performance of ILC based on input/output measurements are verified with simulation results.