Interval model updating using perturbation method and Radial Basis Function neural networks

In recent years, stochastic model updating techniques have been applied to the quantification of uncertainties inherently existing in real-world engineering structures. However in engineering practice, probability density functions of structural parameters are often unavailable due to insufficient information of a structural system. In this circumstance, interval analysis shows a significant advantage of handling uncertain problems since only the upper and lower bounds of inputs and outputs are defined. To this end, a new method for interval identification of structural parameters is proposed using the first-order perturbation method and Radial Basis Function (RBF) neural networks. By the perturbation method, each random variable is denoted as a perturbation around the mean value of the interval of each parameter and that those terms can be used in a two-step deterministic updating sense. Interval model updating equations are then developed on the basis of the perturbation technique. The two-step method is used for updating the mean values of the structural parameters and subsequently estimating the interval radii. The experimental and numerical case studies are given to illustrate and verify the proposed method in the interval identification of structural parameters.