Abstract: The problem of interval model updating, with test structure variability is formulated. Constructing the objective function of the optimization problem, which is the residuals of natural frequencies of structure with weighting factors, based on radial basis function (RBF) and genetic algorithm, to infer the method of model uncertainty quantification and propagation from the aspect of theoretical perspective, and an iterative procedure of the interval model updating approach is given. Then, the method is validated by a three degree of freedom mass-spring system and bolt connection structures. Results show that the updated hypercube of uncertain parameters is in good agreement with the true hypercube, and the predictive space of responses, which is simulated by using Monte Carlo simulation in parameter rang, have very slight difference with the test one, they are demonstrated the accuracy and practicality of this method. In addition, the prediction range would be more conservative and reliable by using interval updating with irreducible uncertain measured data, since the noises are unavoidable in the process of test.

Key words: Interval analysis, Model updating, Prediction, Radial basis function model, Uncertainty analysis