[Back]

P. Rosko:
"Identification of structural matrices and model optimization";
Talk: 5-th World Congress of Structural and Multidisciplinary Optimization, Jesolo; 2003-05-19 - 2003-05-23.

The aim of the paper is to identify structural matrices - stiffness, mass and damping matrices. The structure is experimentally investigated using modal analysis. The modal testing results frequency response functions as ratio between output (acceleration) and input (force) in the mesh nodes. In the next step, there are vibration modes, frequencies and damping parameters obtained. They serve as input data for numerical model of the structure. The mode shape vectors from experiment don’t satisfy the orthogonality condition. The least square correction procedure uses correction term for mode shape vectors derived on the base of Lagrange multiplier method. The preliminary determination of stiffness matrix uses higher number of measurement results as the number of unknown matrix elements. Considering the same measurement error distribution the Gauss - Markov algorithm is applied. In the next step, accuracy is improved using the modified least square method in regard to different measurement data reliability.