비선형 시스템의 적응 학습 제어

Abstract

In this thesis, we present adaptive learning control schemes for two classes of nonlinear systems. We analyze the stability of this control scheme and control parameter convergence. And then, computer simulation is used to show the validity of the control scheme. Firstly, we propose the adaptive iterative learning controller (AILC) for single-input single-output and multi-input multi-output uncertain nonlinear systems in normal form, the uncertainty of which is linearly parameterizable. Existing AILC only learns the system parameter, whereas proposed learning controller learns the system parameter as well as input gain parameter. Moreover, this controller estimates the desired input to drive the tracking error zero more rapidly. Proposed learning controller with parameter learning rules is designed to guarantee bounded error signals and achieve asymptotic tracking error convergence as repeating the task. The computer simulations establish the validity of the proposed controller. Also, we propose the adaptive ILC scheme for single-input single-output and multi-input multi-output uncertain feedback linearizable systems, the uncertainty of which is linearly parameterizable. Since only output signal is available for measurement, we design an high gain observer to estimate the unmeasurable state. Based on estimated state, we construct the AILC which learns the system parameter as well as input gain parameter. Proposed learning control scheme can guarantee the boundedness of error signals and achieve that mean-square tracking error is order of epsilon as repeating the task. The effectiveness of the learning control scheme is demonstrated by simulation examples.