A universal learning controller for a class of nonlinear system

Abstract

A novel universal adaptive learning control approach based on Fourier series is proposed for a class of general time-varying nonlinear systems. The closed-loop system. consists of two loops. One loop is the inner loop, in which the feedback control actions are up-dated at each sampling period. The inner loop is used to stabilize the open-loop system and provide robustness. The other loop is an outer loop, in which the Fourier series-based learning control actions are up-dated at each pass. The outer loop is designed to overcome unmodeled dynamics. The global stability condition of the closed-loop nonlinear control system has been given. Numerical simulation is carried out to illustrate the effectiveness of the proposed controller.