Projeto de observadores não-lineares com propriedades de robustez e capacidade de aprendizado

Abstract

This paper presents the development of a robust nonlinear observer with learning capabilities. The observer is designed to be input-to-state stable (ISS) to exogenous disturbances (which may also represent modeling errors), as well as to an auxiliary signal (generated, e.g., by a neural network) accountable for improving the estimation performance of the observer. Finally, numerical simulations demonstrate the potentials of the proposed observer applied to the state-of-charge estimation of a lithium-ion battery (LIB).