Deterministic Learning From State Observation
【摘要】：正 This paper studies deterministic learning of a class of nonlinear dynamical systems with only output measurement. Following an earlier result of deterministic learning using full-state measurement, it is shown that by using the high-gain observer technique, the "deterministic learning" ability can still be implemented. Specifically, a partial PE condition is satisfied, and accurate identification of system dynamics along a estimated state trajectory is achieved. The significance of the paper is that it shows that the estimated state information can be effectively used to accurately modelling the underlying dynamics in a nonlinear observer problem. Simulation studies are included to demonstrate the effectiveness of the approach.