Modal analysis for randomly excited structural systems with unmeasured input

A new time series approach for the estimation of modal parameters of randomly excited structures is presented in this paper. Since previous methods in time series analysis techniques suffer from the phase ambiguity problem, the new method is proposed to circumvent this difficulty. It relies on the development of a vector first-order autoregressive model VAR (1) of the vibrating structure at the points of measurement and, in particular, on the utilization of velocity information to set up the cross-covariance matrix in the estimation procedures. The theoretical basis of the procedure is presented along with simulation results.