Implementation algorithms for self-identification of adaptive structures with variable geometric properties

Abstract

On-orbit vibration testing is one of the issues for in-orbit construction of large space structures because of structural variations during assembly. Variable geometry truss structures would be the candidates as the tools for the in-orbit construction. This paper proposes structural identification algorithms for implementing self-identification of variable geometry truss structures. Using the implementation algorithms, the variable geometry truss determines quasi-static changes of their own geometric parameters to enhance the lack of vibration modes with limited excitation locations for on-orbit vibration testing. The proposed algorithms are numerically examined through the identification of the stiffness matrix of a two-dimensional variable geometry truss with a single variable length member and a single excitation location. We assumed that all the nodes of the truss structure were fully instrumented for measuring mode shape and modal frequency. In the numerical experiments, unknown stiffness matrix of finite element model corresponding to the variable geometry truss is identified under the assumption that exact mass matrix is given using finite element model. It is found that increasing the number of geometry changes can enhance the lack of vibration modes, and the identification error of stiffness matrix is reduced accordingly. The sensitivity of identification accuracy to the modeling error in given mass matrix is also addressed. Furthermore, we discuss a method for selecting sufficient number of geometry changes by considering the convergence of the norm of identified stiffness matrix.