Data‐driven system parameter change detection for a chain‐like uncertainties embedded structure

Abstract

A data-driven identification technique is used to implement an effective change detection approach for uncertain multi-degree-of-freedom chainlike systems. The information about mass properties of system is not required in the process of identification, but only the excitation information and the corresponding structural dynamic response are needed to obtain a stochastic representation of estimated changes in stiffness-like and damping-like structural coefficients. The validity and reliability of the data-driven technique in uncertain chain-like systems are further verified by using shaking table experimental data from a five-floor shear structure. The results of this study show that this method can not only accurately detect the existence of physical structural changes but also accurately locate the region of the changes and determine the degree of structural changes. Additionally, a finite element model of the test shear structure was developed and simulated to verify the effectiveness of this change-detection approach.