Improving moment method for fuzzy reliability sensitivity analysis with correlative normal variables

After reviewing past research papers such as Refs.1 and 2, we propose making further improvements. Section 2 of the full paper explains in some detail the moment method that includes our improvements. It is divided into four subsections. Subsection 2.1 is: the basic idea. Subsection 2.2 is: the discretization of fuzzy region. Subsection 2.3 is: transforming the fuzzy reliability sensitivity of the fuzzy failure probability to the parameters in a membership function. Our method essentially is: by splitting the integral region of the fuzzy failure probability into a set of integral subregions, the fuzzy reliability sensitivity with correlative normal variables is transformed into random reliability sensitivity with correlative normal variables. The random reliability sensitivity can be solved by our improved moment method, in which the correlative normal variables are transformed into noncorrelative normal variables equivalently. Section 3 gives three numerical examples of reliability sensitivity analysis. The analysis results, presented in Tables 1 through 3, show preliminarily that our improved moment method is much more efficient than the Monte-Carlo numerical simulation method.