摘要
针对核动力装置系统参数密集且相互影响、故障征兆随故障发生的位置、程度以及警报阈值的不同而发生变化、难以获得确定的故障诊断规则的问题,在采用多层流模型描述核动力装置的运行特性的基础上,引入了影响因子来描述故障征兆之间的因果关系强度,进一步结合警报分析方法和贝叶斯理论进行不确定性故障推理。建立了压水堆主冷却剂系统不确定性故障诊断模型,通过仿真证明了该方法可以合理有效地识别系统故障,解决了由于诊断规则的不确定性而可能造成的误诊和漏诊,可有效辅助操纵员进行深层次决策。
Nuclear power plants have large significantly interacting parameters. The symptoms of a fault may show up well away from the fault's location with missleading degree of severity,while escaping pre-set alarm thresholds.It is very difficult to diagnose faults by definite rules. In this paper,Multilevel Flow Models were used to describe the operational characteristics of nuclear power plants where impact factor was introduced to quantify the strength of causality between fault symptoms. An uncertain fault diagnosis approach based on alarm analysis and Bayes theory was explored. That is,an uncertain fault diagnosis model of the primary coolant system was developed. The soundness of the proposed approach was verified through simulations. The proposed approach can identify faults logically and effectively. It can solve misdiagnosed and undiagnosed problems missed by the misleading certainty of rules and can assist operators in their deep level decision-making.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2015年第10期1421-1426,共6页
Journal of Harbin Engineering University
基金
国家十二五核能开发基金资助项目(L20110489)
关键词
故障诊断
多层流模型
贝叶斯理论
不确定性推理
核动力装置
fault diagnosis
multilevel flow models
Bayes theory
uncertain reasoning
nuclear power plants