A dimensional reduction method for analysis of the impact of the epistemic uncertainty of distribution parameters on a structural system's variance of average performance output

To achieve effective control of the average performance output of a structural system, the influence of the epistemic uncertainty of distribution parameters on the average output variance was analyzed. In view of the fact the conventional Monte Carlo analysis method has the short comings of lower efficiency and large computational cost, a multiplicative version of the dimensional reduction method (M-DRM) was employed to compute the average output variance-based global sensitivity to reduce the functional function's usage greatly compared to the conventional Monte Carlo method, and the method of separating the aleatory and epistemic uncertainties, combined with the M-DRM method, was used to resolve impact of the epistemic uncertainty of contribution parameters on the average output variance to further improve the computional efficiency while keeping a higher precision.