Abstract：For the reliability sensitivity estimation with correlative normal variables, the Transform Method (TM) based on Monte Carlo simulation have been established firstly through transforming the correlative normal variables into independent ones, and its variances is analyzed also. In order to improve the efficiency of the reliability sensitivity analysis, the TM based on Monte Carlo simulation are combined with the adaptive radial-based importance sampling (ARBIS) method respectively, and the ARBIS-based TM is established for the reliability sensitivity analysis with correlative normal variables. Using information provided by the required samples for the reliability sensitivity estimation, the optimal radials of the ARBIS based method can be determined by gradual iteration. Furthermore, in the determination of the optimal radial, due to the interpolation constructed by the most probable point (MPP) in the failure domain of the each iteration, the robustness and the accuracy of the ARBIS based method are improved greatly. Since the universality and the robustness of the Monte Carlo simulation and the high efficiency of the radial-based importance sampling are propagated to the ARBIS-based TM, the established method is strongly applicable to the highly non-linear implicit limit state equation, systems with multiple failure modes in series, in parallel or in mixed states, and the multiple MPPs. The results of the illustrations adequately demonstrate the robustness, efficiency, accuracy and universality of the established ARBIS-based method.
2010, Vol. 31 
