In the present article, the authors develop a new formulation for the problem of updating Finite Element models based upon results obtained from experimental tests. The process of updating is iterative and divided in two stages, (i) localization of the zones where the highest errors between the theoretical model and the experimental results occur and (ii) correction of the theoretical model in those zones. This paper focuses on the first stage, the error localization, the technique developed being based upon the definition of an error measure on the constitutive relation, which takes into account the forced response of a mechanical system and consequently enabling the incorporation of the frequency-response-functions obtained from experiments. In some previous works, the use of an error measure on the constitutive relation has been shown as highly efficient in the whole process of Finite Element updating, when free vibration was considered and only the experimental resonant frequencies accounted for. The work presented here is, somehow, an extension of those others. At present, the method only accommodates undamped or very lightly damped systems, future extensions being sought. A numerical example is given, where various situations are simulated, in order to validate the method.
展开▼
