基于自适应协同克里金代理模型的铅铋冷却非能动系统可靠性分析

bstract Lead-bismuth cooling passive system plays a vital role in the safe operation of the reactor. However, there are many uncertain factors in the reactor, such as small driving force, large fluctuation of design parameters and high uncertainty of initial conditions, which may lead to the failure of thermal and hydraulic processes, thus affecting the normal realization of system functions. Therefore, the reliability analysis of passive lead-bismuth cooling system is a key step to ensure its effectiveness and safety. In this paper, the adaptive important sampling method and collaborative kriging model are deeply studied at first. Subsequently, the self-adaptive cooperative kriging agent model is verified by single failure region and multiple failure regions. Finally, this paper systematically discusses the sensitivity analysis and reliability analysis of TALL-3D passive system. The results show that the combination of adaptive cooperative kriging agent model is more accurate than the traditional Monte Carlo analysis method in fault probability estimation, which significantly improves the simulation efficiency. By optimizing the sample selection strategy, this method reduces the calculation of unnecessary areas and enhances the ability to explore the limit state function, thus greatly reducing the calculation required for the reliability analysis of complex systems. In addition, adaptive important sampling shows excellent performance in high-dimensional parameter space, which can flexibly adapt to different sample requirements and ensure more intensive sample collection in key failure areas. This characteristic makes the proposed method have strong feasibility and advantages, which provides strong technical support for the application of lead-bismuth cooling passive system and provides important reference value for the research in related fields.