Adaptive substructure identification for shear structures with virtual control system

Abstract

Abstract Improving structural identification accuracy plays a vital role in enhancing the accuracy of structural damage detection. In the previous study, the authors proposed a cross power spectrum density (CPSD) based substructure identification method for a shear structure. The CPSDs of a two-story substructure’s responses are used to formulate a recursive identification problem to estimate the structural parameters. The in-depth identification error analysis revealed that the CPSDs of two key substructure responses near the substructure frequency are critical factors to determine the identification accuracy. In this paper, a virtual control system (VCS) is introduced into the substructure identification process to improve the identification accuracy. The VCS is a self-balanced system, composed by a control device and the self-balanced forces, which offset the actions that the control device apply on the structure. The control device is combined with the structure to form a controlled structure system, which is used to replace the original structure in the substructure identification. The self-balanced forces are treated as the known excitations to the controlled structure. Although the VCS is self-balanced and does not change the original structural responses, it can adjust the dynamic characteristics of the controlled structure via tuning the VCS’s parameters, which makes the original structural responses become favorable to the identification of the controlled structure and, thus, improves the identification accuracy. To make full use of the VCS’s ability to enhance the identification accuracy, an adaptive design procedure is proposed to obtain the optimal parameters of the VCS. Furthermore, a numerical example of an eight-story shear structure is used to demonstrate the effectiveness of the VCS to improve the identification accuracy. Finally, shake table tests are performed on a five-story bench-scaled structural model to verify experimentally the efficacy of the proposed method.