| Summary: | 碩士 === 華梵大學 === 機電工程研究所 === 87 === This thesis discuss the design of iterative learning controller for discrete time nonlinear system and propose a new technique for stability analysis. As we know ,λ-norm approach is widely used to prove the convergence and robustness for traditional iterative learning systems. However , convergence in the sense of λ-norm can not guarantee a reasonable error convergence in the final iterate. The first contribution of this thesis is to propose a new sup-norm approach for analysis of convergence and robustness , which can ensure the reasonable performance in the final iterate. The convergence of learning systems highly depends on the design of learning gain. In most of the works , we know that the choice of learning gain will depend on input-output coupling matrix of the nonlinear system. However , how to get the formation of input-output coupling matrix is not discussed in most of the works. The second contribution of this thesis is to apply the Adaptive Network Based Fuzzy Inference System (ANFIS) for identification of input-output coupling matrix. It is shown via computer simulation that the convergence performance can be greatly improved by using the ANFIS based iterative learning controller.
|
|---|
