A time-frequency method for nonlinear system classification in presence of noise

This paper presents a method for the classification of nonlinear systems through the study of the free oscillations in the time-frequency plane, when the measured data are affected by noise. Nonconservative SDOF (Single Degree Of Freedom) oscillators described by a nonlinear second order differential equation are considered. The nonlinearity is due to a nonlinear function of the state variable, which produces free oscillations with a time-variant spectrum. The method used for the classification is a substantial modification of a basic algorithm proposed by the same authors for noise-free data. In presence of noise improved performances are obtained with the new algorithm.