A perturbation technique for interval model updating is proposed.
RBF neural network-based surrogate models are used for representing complex models.
The proposed method can identify parameter intervals with very good accuracy.
The proposed method is practical and feasible for complex structures.