Expected value analysis of multiparameter systems

摘要

This paper proposes a new method, called EVA (expected value analysis), to analyze the effect of the uncertainties of the parameters on the output function of systems described as differential equation initial value problems. Starting from the model equations, the associated EVA differential equations are shown to be readily handled through available computer codes. A quantitative analysis of the EVA approximation is presented along with a comparison to local SSA (statistical sensitivity analysis) and global FAST (Fourier amplitude sensitivity test).