Multi-objective framework for structural modeling consistent with data.

A generalized multi-objective framework for structural model updating is proposed and demonstrated using experimental data from a small scale laboratory structure. Multiple Pareto optimal structural models are obtained that are consistent with the measured data and the norms used for reconciling finite element models with data. The relation between the mul-ti-objective identification framework and conventional weighted residuals methods that mini-mize a weighted sum of the residuals between the structural model and the measured data is investigated. The Pareto models contain the optimal models obtained from conventional weighted residuals methods. Theoretical and computational issues involved in estimating the Pareto optimal models are addressed. Conventional methods use arbitrary assumptions to select a single optimal model among the multiple alternative ones. The proposed multi-objective framework and corresponding computational tools provide the whole spectrum of optimal models and can thus be viewed as a generalization of the available conventional methods. Multi-objective and conventional single-objective model updating methods are com-pared and their effectiveness is demonstrated using experimental results from a three-story laboratory structure. The results clearly indicate that there is wide variety of Pareto optimal structural models consistent with the measured data and the norms used for reconciling finite element models with data. It is shown that the response and reliability predictions from these data-consistent Pareto optimal models can vary considerably. The size of observed variations depends on the information contained in the measured data, as well as the size of model and measurement errors always present in structural modeling and data processing techniques.