Application of Sobol Sensitivity Analysis with Surrogate Model of Analytical Function

Abstract

The Sobol sensitivity analysis method enables identifying the sensitivity factors that significantly impact the output results of mathematical models from a large number of uncertainty factors and analyzing and measuring the degree of their impact and sensitivity on the output results. However, Sobol sensitivity analysis is challenging to perform due to the specific analytical functions of mathematical models in the industry that are difficult to obtain frequently. In this paper, the required functional relationship was fitted by combining heat transfer theory with Lasso regression for the relationship between the influence of temperature changes in a piece of equipment in an outdoor environment and external environmental factors, and a more satisfactory result was obtained by performing a sensitivity analysis in this way. The study results prove that this method could provide convenience for industrial tests and has application value.