分类: 物理学 >> 核物理学 提交时间: 2025-01-17
摘要: The Monte Carlo Simulation based method requires considerable computational efforts for the estimation of functional reliability analysis. Efficient sampling techniques can be adopted for performing robust estimations with limited number of samples and associated with computational time. In order to solve the problem of multi-dimensional uncertainties and small functional failure probability in passive system reliability analysis, an innovative optimized line sampling was presented. In the presented method, genetic algorithm was employed to solve the nonlinear constrained minimization problem for identifying the optimal important direction of line sampling, and then the failure probability can be evaluated by line sampling with high efficiency. Taking the reliability estimation on the capacity of natural circulation cooling in reactor care of Xi’an Pulsed Reactor for example, the uncertainties related to the model and the input parameters were considered in this paper. Natural circulation functional failure probability was calculated under middle loss of coolant accident (MLOCA). The numerical results show that the presented method has the high computing efficiency and excellent computing accuracy compared with traditional probability analysis methods. In addition, this method demonstrates the efficiency and feasibility for functional reliability analysis of implicit nonlinear limit state function with high dimensional variables and small failure probability.