Abstract: Based on the moment estimation of limit state function for calculation of failure probability, a new reliability sensitivity analysis method is presented. The partial differential of the moment of the limit state function to distribution parameters of basic variables is derived. By use of the relationship of the failure probability and the moment of the limit state function, the partial differential of the failure probability to the distribution parameters of basic variables is derived furthermore. Hereby, the reliability sensitivity is obtained. Neither design point nor gradient of the limit state function to the basic variable is required for the presented method, therefore it is more suitable for the implicit limit state equation comparing with the reliability sensitivity based on the first order and second moment method. Some illustrations are used to demonstrate the rationality and the precision of the presented reliability sensitivity analysis method.

Key words: reliability, probabilistic analysis, first-order reliability method, Monte-Carlo simulation, parameter sensitivity, moment method