Experimental Identification of Nonlinear Vibratory Systems by Neural Networks

Abstract

Recently identification techniques using neural networks for nonlinear vibratory systems attract interests of engineers. The identification techniques developed so far can determine the input-output relation of the objective vibratory system as a whole. They cannot determine the parameters of the system separately. In this report, we propose a new experimental identification technique which can determine the linear parameters as well as nonlinear terms of the system. The applicability of the technique is confirmed by numerical simulation and experiment.