15th Triennial World Congress of the International Federation of Automatic Control
  Barcelona, 21–26 July 2002 
(N, ε) STABILITY ANALYSIS OF NONLINEAR SYSTEMS USING UNIVERSAL LEARNING NETWORKS
Kotaro Hirasawa* Jinglu Hu* Junichi Murata*
* Dept. of Electrical & Electronic Syst Engg., Kyushu University
Hakozaki 6-10-1, Higashi-ku, Fukuoka 812-8581, Japan
TEL: (+81)92-642-3907, FAX: (+81)92-642-3962
E-mail: hirasawa@ees.kyushu-u.ac.jp

This paper proposes a stability analysis method based on the higher order derivatives of ULNs. In the proposed method, the following are proposed. Firstly, if the absolute values of the first order derivatives of any coordinates of the original trajectory with respect to any initial disturbances approach zero at time infinity, then the trajectory is locally asymptotically stable. Secondly, the locally asymptotically stable region, where asymptotical stability is secured approximately, is obtained by neglecting the higher order derivatives until nth order with epsilon approximation.
Keywords: Nonlinear system, learning network, stability analysis, high order derivatives.
Session slot T-We-A04: Neural network analysis and learning/Area code 3e : Fuzzy and Neural Systems