Modal Parameter Identification of Linear Time Invariant System under Non-stationary Ambient Excitation

Abstract

In this paper, a new modal parameters identification method is presented based on NExT (natural excitation technique) and the response data of the structure under non-stationary ambient excitation the white noise excitation assumption for NExT is extended to a MA (q) model assumption to establish an extended NExT method. Then the assumption that the non-stationary excitation is composed of a (d-1)-th order polynomial and a MA(q) model is proposed. Taking d-th order difference of the non-stationary excitation yields a MA(q+d) model, which satisfies the assumption of the extended NExT method mentioned above. Thus, the extended NExT method can be used in the condition of non-stationary excitation. In order to reduce the error of difference calculation during identification implementation, the difference computation is performed in frequency-domain instead of time-domain. The identification results from the proposed method are compared with the standard NExT for a three DOFs (degrees of freedom) vibration system and the finite element (FE) model of Longtanhe Bridge. The results show that the method proposed in this paper can obtain better modal parameters of the system under non-stationary ambient excitation than the standard NExT.