Moment-independent Importance Measure Analysis Method Based to Point-estimate

Abstract

Abstract：The moment independent importance measure can reflect the influence on the output characteristics by the uncertainty of input parameters. The traditional numerical simulation method needs a large number of samples, thus is of low efficiency, which limits its application to engineering problems. To solve this problem, a nesting point estimate method is proposed to calculate two moment independent importance measures. In this method, the traditional double loop is converted to a single loop, which greatly enhances the computational efficiency. This method immune from the distribution type of random variables, and calculate the importance measures with selected points and the corresponding weights. The proposed method is of good applicability and high efficiency. The proposed method is applied to several practical problems to demonstrate its applicability in the engineering.

Hybrid multiplicative dimension reduction method for uncertainty analysis of engineering structures

“…In order to estimate the first four moments of outputs, a three-point estimation method is introduced. 30–32 The basic idea of the three-point estimation method is to use the selected points and the corresponding weights to approximate the integral of the function. The original n -dimensional response function is approximated as the product of a series of one-dimensional functions, so the original n -dimensional integral is approximated by employing one-dimensional integrals.…”