Contribution of the Proper Orthogonal Decomposition Modes to Accuracy of Parameter Identification in Flexible Multibody Systems

EasyChair Preprint 13375

Abstract

Acquisition of precise parameters is of key importance for accurate numerical simulation for multibody systems. However, it is generally difficult to obtain all parameters required for the simulation by only direct measurements.  The authors have presented a parameter identification technique based on the adjoint method. It incorporates the proper orthogonal decomposition (POD) into a cost function, in order to consider model uncertainties. More specifically, the proposed method uses data samples decomposed into the proper orthogonal decomposition modes (POMs) for the cost function. According to works by the authors, the cost function given by relatively higher POMs leads precise estimated values, even though such higher POMs have quite lower contribution ratios. This study investigates the relation between the POMs and the accuracy of the parameter identification. In order to derive the equations of motion, we employ the floating frame of reference formulation for the description the elastically supported beam. Then, the time series data for the beam displacements are divided into the components of the POMs. Each components of POMs is analyzed by the Fourier analysis. After that, we compare the frequency components for the component of POMs with the analytical values of natural frequencies and discuss the contribution to the accuracy of the parameter identification.

Keyphrases: Flexible multibody systems, Proper Orthogonal Decomposition, adjoint method, model uncertainties, parameter identification

@booklet{EasyChair:13375,
  author    = {Kensuke Hara and Nozomi Kitamura and Ryoya Nishi and Hiroshi Yamaura},
  title     = {Contribution of the Proper Orthogonal Decomposition Modes to Accuracy of Parameter Identification in Flexible Multibody Systems},
  howpublished = {EasyChair Preprint 13375},
  year      = {EasyChair, 2024}}