Optimization of the Number of Eigenvectors Used for Nonlinear Reduced Order Models

Abstract

Substructuring techniques have wide applications in various engineering fields. The systems are built with several degrees of freedom (DOF), which are highly complex and experience considerable computational costs. To increase the efficiency and to simplify engineering problems, a technique was developed by Przemieniecki in 1963. Many researchers have since improved on this technique. The behavior of structures under various conditions can be expressed by the superposition of their eigenmodes. In this paper, a genetic algorithm, a type of global optimization technique, is used to determine dominant eigenvectors to reduce the DOF of geometrically nonlinear beams. Thus, a complex structure can be simplified while retaining its own characteristics. To demonstrate the accuracy of the proposed approach, beams with different boundary conditions are taken as test beds. The results obtained are then compared to those of a full system analysis.