Regularization for symmetric and almost symmetric systems in model updating

The occurrence of mechanical systems with symmetries in the context of problems with parametric response dependencies can cause a range of problems such as ill-conditioning, singularities, qualitative changes of behaviour. Common methods using parameterized response model representations include the techniques of model updating. The main driver of these algorithms is often the sensitivity matrix corresponding to a given parameterization. Using only the sensitivity matrix in cases of systems with symmetries will result in erratic behaviour of the updating algorithms, numerical problems and can even cause the algorithms to fail to converge to physically acceptable results. The goal of this paper is to investigate the potential of regularization conditions to stabilize the behavior of problems with symmetries. The use of regularization is demonstrated in a numerical case study with singularities and illconditioning present in the first order approximation representation. © 2008 by the Katholieke Universiteit Leuven Department of Mechanical Engineering All rights reserved.