On-Line Neuro Identification of Uncertain Systems Based on Scaling and Explicit Feedback

摘要

This paper focuses on the identification problem of uncertain systems. Based on a neural identification model with feedback, scaling and Lyapunov-based weight adjustment law, an identification algorithm is proposed to make the ultimately bounded on-line state error. The relevance of this work is also the formalization of the fact that the scaling of unknown nonlinearities, prior to the neural approximation, and the introduction of an explicit feedback in the neural model has a positive impact on performance and application of the algorithm. To validate the theoretical results, the identification of three chaotic systems is performed...