l1鲁棒辨识渐近性态的研究

On the Asymptotic Properties of l 1 Robust Identification

基于ｌ１拓扑空间的局部紧性，研究了ｌ１鲁棒辨识的渐近收敛性质，据此提出了一个具体的辨识算法。

The goal of control oriented system identification is to provide a method with a form compatible with robust control. This is a new research field in control theory. The asymptotic properties of control oriented identification algorithms are carefully investigated. The asymptotic properties of l 1 robust identification are studied based on local compactness in l 1 topological space. It has been proved that there exist l 1 robust identification algorithms for BIBO stable systems and to any bounded input u(t) and the bound of disturbance ε ,the asymptotic identification of the algorithms is less than the worst case error bounds. A new identification algorithm is proposed and the l 1 norm methic bound of identification error is discussed.