Non-Linear System On-Line Identification using Dynamic Neural Networks

Abstract

ABSTRACTWe analyze nonlinear on-line identification using a dynamic neural network, with the same state space dimension as the system. We assume the system space state completely measurable. We formulate the identification error, and determine stability conditions for this error by means of a Lyapunov-like analysis. As our main original contribution, we establish a bound for the identification error. Applicability is illustrated via an example.