System identification of mechanical structures by a high-order multivariate autoregressive model

This paper presents an algorithm for a time series analysis—the high-order multivariate autoregressive model M-AR(P). The theory of the model and the application in modal parameter identification of structures excited by natural random forces is described. A water transmission tower is taken as a real example to illustrate the application procedure of this algorithm. Good results in natural frequencies, damping ratios, mode shapes, as well as power spectra and coherence functions identified by this model prove the advantages and usefulness of the algorithm.