A statistical moments based approach for the dynamic identification of civil structures

Abstract

In recent years, interest in developing identification techniques that are valid in the case of unmeasured input has increased. In this field some interesting parametric approaches have been proposed. Nevertheless, the improvement of the available techniques or the formulation of new techniques is desirable. In this paper a time domain dynamic identification approach based on the statistical moments of the response of civil structures under base random excitations is discussed. Two types of models are used: the classically damped models characterized by mass proportional damping and the so-called “potential models” which are non linear in damping and stiffness. By applying the Itô differential stochastic calculus to some functions of the response, algebraic equations that depend on the above statistical moments can be obtained. These equations can be used for the dynamic identification of the mechanical parameters characterizing the structural model, in the case of unmeasured input as well, and the identification of the input itself. The above equations reveal the possibility of identifying the dissipation characteristics independently from the input characteristics.