Many dynamical systems exhibit complex behaviors dominated by low-dimensional structures, even though they possess a large degree of freedom. Dynamic Mode Decomposition (DMD) is a recent development in the post-processing algorithm for extracting low-dimensional governing features in nonlinear dynamical systems with large dimensions, which can be applied equally well to data from simulations and experiments. Unlike conventional modal decomposition techniques such as the Proper Orthogonal Decomposition (POD), DMD identifies characteristic growth rates, frequencies, and their corresponding spatial patterns. Moreover, the fact that the DMD algorithm is an approximation of the Koopman spectral analysis provides a firm mathematical foundation for its application. DMD has been utilized to analyze systems with large degrees of freedom such as power systems, fluid flows, and heat flows in the building. There also exist many other systems to be analyzed such as microelectromechanical systems (MEMS), which may be a promising direction in the DMD applications. In this study, the DMD analysis is performed on data sets obtained from numerical simulations of MEMS. Complex dynamical behaviors on attractors, including chaos, are successfully decomposed into their characteristic modes, which oscillate with individual fixed frequencies.
各種検索
サポート
ヘルプ