An efficient sampling method for variance-based sensitivity analysis

In order to improve the efficiency, the accuracy and the robustness of the sampling-based methods for estimating the variance-based sensitivity indices, a new efficient method based on the combination of the unconditional expectation, the conditional expectation and the multiplication approximation of the response function is proposed in this paper. By the new equivalent forms of the variance-based sensitivity indices and the multiplication approximation of the response function, the proposed method can simultaneously estimate all the order effects by repeatedly making use of the same sample points. Through three typical test examples and an engineering application, the efficiency, the accuracy and the robustness of the proposed method are demonstrated in comparison with other existing sampling-based methods.