Open and Closed Loop Neural Identification of Uncertain Systems with Transient Time Adjustment

This paper addresses the online identification problem of uncertain systems. By using a neural identification model with feedback and a weight law based on Lyapunov theory, an online identification algorithm is proposed to make the residual state error arbitrarily small and related to two design matrices. In addition, it is shown that the transient can be controlled by other parameters which are not related to the residual state error. Hence, it is possible to decouple the transient and the steady state error. To validate the theoretical results, the identification and control of a chaotic system was accomplished and the influence of the main design parameters in the performance of the algorithms was examined.