Adaptive importance sampling method for estimation of reliability sensitivity

On the basis of the simulated annealing algorithm for global optimization, an adaptive importance sampling method is presented for evaluation of reliability sensitivity. In the presented method, the simulated annealing optimization is employed to seek, in the failure region gradually, the most probable failure point used as the sampling center of the adaptive importance sampling function. The samples generated by the adaptive importance sampling functions are utilized to estimate the reliability sensitivity. The reliability sensitivity evaluation, the variance and the variation coefficient of the evaluation are derived for the adaptive importance sampling based on the reliability sensitivity method. Compared to the Monte-Carlo method through the demonstration of examples, it can be observed that the presented method is more efficient.