Interval strategy-based regularization approach for force reconstruction with multi-source uncertainties

doi:10.1016/j.cma.2023.116679

Return

Interval strategy-based regularization approach for force reconstruction with multi-source uncertainties

2024-02-01

PRE

陈阳 (Chen Yang) *

DOI：10.1016/j.cma.2023.116679

Original

Original request for help

Save

Abstract

En 中文

A force reconstruction method for uncertain structures instead of direct measurement is proposed in this paper. Considering the multi-source uncertainties as unknown-but-bounded (UBB) parameters including unknown structural dynamics and measurement errors, this study proposed a novel force reconstruction method using an interval strategy-based regularization approach with truncated singular value decomposition (TSVD). To overcome the drawbacks of incompleted models and limited information, the overall modelling and inversion of the interval force are reconstructed based on the non-probabilistic uncertainty theory. First, The interval uncertaintybased kernel function for force inversion is deduced using the second-order interval perturbation method, in which the uncertainties of the reconstructed force originate from the uncertain singular values and measurement errors. Based on the regularization of the TSVD, the interval bounds of the singular values can be accurately obtained. Combining the two typical regularization approaches, a novel interval strategy-based regularization approach is proposed to optimize the best regularization parameter based on the uncertain interval distribution of singular values. This novel strategy can not only balance the two typical regularization approaches but also reduce the burden of force inversion. Two numerical examples including a truss and space capsule are applied to access the proposed method, and the estimated accurate force can prove the effectiveness.

Keywords:

Force reconstruction

Second-order interval analysis

Interval of singular value

TSVD-based regularization

Interval strategy for regularization parameter

Multi -source uncertainties