核电厂概率安全评价中先进故障树分析方法研究

发布时间：2018-06-30 19:25

本文选题：核安全 + 概率安全评价　；参考：《中国科学技术大学》2014年博士论文

【摘要】：核能是目前一种非常重要的能源类型,而核安全是核能与核技术利用事业发展的生命线。有效的核安全分析方法能够帮助提高核电厂的安全性,以故障树分析(FTA)为主要技术手段的概率安全评价(PSA)方法在核能安全领域发挥着重要作用,已经成为核安全分析的主要方法之一。在核电厂PSA中,往往会产生规模非常大的故障树。大规模故障树的高效分析是一个非常困难的问题。本文结合核电厂实时风险分析的特点,从大规模故障树的预处理与定性分析入手展开深入研究,提出了一套基于改进型零压缩二元决策图的快速FTA方法,以满足核电厂实时风险分析的要求。在这套方法中,快速、充分的故障树规范化、化简与模块化算法,以及多种高效的零压缩二元决策图基本事件排序方法及并行计算方法得以发展,较大程度的提高了FTA的效率和速度。在中科院核能安全技术研究所.FDS团队前期开发的RiskA工作基础上,本文的主要工作与贡献之处在于：(1)提出了用于处理高阶K/N门的降阶组合算法和二分组合算法,使得高阶K/N门能被快速处理；(2)提出了一种基于故障树分页存储的故障树化简方法,使得在内存占用较小的分页存储情况下也能收到较好的化简效果,与改进前RiskA相比节省了30%-70%内存,计算速度提高了1-10倍；(3)提出了一种动态调整化简策略以提高模块化程度的方法,对同一个故障树实例提高模块数量0.5-1倍；(4)提出了一种基于割集信息零压缩二元决策图基本事件排序方法,与改进前RiskA相比运行时间约减少50%以上,内存占用降低约20%：(5)提出了一种基于割集信息与故障树结构的混合排序方法,牺牲了少量内存方面的性能,使得计算速度基于割集的排序方法进一步提升约1-2倍；(6)提出了一种基于零压缩二元决策图的故障树分层并行求解算法,效率明显高于串行算法,与改进前RiskA相比提速达到4-6倍。此外,基于上述方法开发了一个可进行快速FTA、具有很强实用性的工业级软件包RiskAT.所发展的快速FTA算法能解决核电厂实时处理大规模故障树存在的问题,促进FTA方法在核能安全领域的进一步应用和发展。RiskAT在秦山第三核电厂的事故序列定量分析中得到了成功应用,计算结果正确可靠,尤其是在计算速度方面明显优于同类商业软件,为秦山第三核电厂快速高效的进行分析工作提供了重要的技术支持。
[Abstract]:Nuclear energy is a very important type of energy at present, and nuclear safety is the lifeline of the development of nuclear energy and nuclear technology utilization. Effective nuclear safety analysis method can help to improve the safety of nuclear power plants. Probabilistic safety assessment (PSA) method, which takes fault tree analysis (FTA) as the main technical means, plays an important role in the field of nuclear safety. It has become one of the main methods of nuclear safety analysis. In nuclear power plant PSA, a very large fault tree is often produced. Efficient analysis of large-scale fault trees is a very difficult problem. Based on the characteristics of real time risk analysis of nuclear power plants, a set of fast FTA method based on improved zero compression binary decision graph is proposed in this paper, which is based on the pre-processing and qualitative analysis of large-scale fault trees. To meet the nuclear power plant real-time risk analysis requirements. In this set of methods, fast and sufficient fault tree normalization, simplification and modularization algorithms, as well as a variety of efficient zero-compression binary decision graph basic event sequencing methods and parallel computing methods have been developed. The efficiency and speed of FTA are greatly improved. Based on the work of RiskA developed by the Institute of Nuclear Safety Technology of the Chinese Academy of Sciences, the main work and contributions of this paper are as follows: (1) A reduced order combination algorithm and a binary combination algorithm for dealing with higher-order K / N gates are proposed. The high order K / N gate can be processed quickly. (2) A fault tree simplification method based on fault tree paging storage is proposed, which can get better simplification effect in the case of smaller memory footprint. Compared with RiskA before the improvement, RiskA saves 30-70% memory, and the calculation speed increases 1-10 times. (3) A dynamic reduction strategy is proposed to improve the modularization degree, and the number of modules is increased 0.5-1 times for the same fault tree instance. (4) A method for sorting basic events based on zero-compression binary decision graph based on cut-set information is proposed. The running time of RiskA is reduced by more than 50% compared with that of RiskA. The memory footprint is reduced by about 20%: (5) A hybrid sorting method based on the cut-set information and the fault tree structure is proposed, which sacrifices the performance of a small amount of memory and further improves the speed of the sorting method based on the cut-set by about 1-2 times; (6) A fault tree hierarchical parallel algorithm based on zero-compression binary decision graph is proposed, which is more efficient than the serial algorithm, and the speed is 4-6 times higher than that of RiskA before the improvement. In addition, based on the above method, a fast FTA-capable industrial class software package, RiskAT. is developed. The developed fast FTA algorithm can solve the problem of real-time processing of large-scale fault trees in nuclear power plants. To promote the further application and development of FTA method in the field of nuclear energy safety. RiskAT has been successfully applied in the quantitative analysis of accident sequence of Qinshan No.3 Nuclear Power Plant. The calculation results are correct and reliable. Especially, it is superior to the similar commercial software in computing speed, which provides an important technical support for the rapid and efficient analysis of Qinshan No.3 Nuclear Power Plant.
【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM623.8

【参考文献】