Journal of Computations & Modelling

Matlab automation algorithm for performing global sensitivity analysis of complex system models with a derived FAST method

• [ Download ]
• Times downloaded: 20110

• Abstract

  This study presents a Global Sensitivity Analysis Tool, named GoSAT, which is a ready to use Matlab automation algorithm of a derived Fourier Amplitude Sensitivity Test (FAST) method that automatically ranks, by decreasing order in a bar diagram, both the first order effects and the order one interaction effects of a model factors. To perform a global sensitivity analysis with the proposed algorithm, only the Matlab code of the surveyed model, all factors' names, their respective nominal values and ranges of variation, and their distinct attributed frequencies are necessary to be known as inputs. To test the efficiency of the proposed procedure, it was applied on a ten factors Sobol function and the expected outcome was obtained, that is, the factors related to the least valuable Sobol parameter are the most influent ones. With the proposed algorithm, modelers do not have to wait for other researchers’ subsequent work completion to have feedbacks on data input and factors of their own models but can directly perform global sensitivity analysis right after developing the computing codes related to these models.