Markov chain simulation based line sampling method for reliability sensitivity

For reliability sensitivity analysis of implicit non-linear limit function with high dimensionality and small failure probability, which is difficult to be solved in engineering, a novel line sampling method was presented on the basis of Markov chain simulation. In the presented method, Markov chain simulation was employed to generate samples in failure domain rapidly. These failure samples were used to determine the important direction, and they were also used as samples of the line sampling for estimation of the reliability sensitivity. After the concept and the implementation of the presented method were explained, the variance of the reliability sensitivity estimation was analyzed approximately. At last, several examples were given to demonstrate the presented method. The results of the examples show that the presented method is valid for calculating the reliability sensitivity with high precision and high efficiency.