摘要
研究了结合贝叶斯网络的动态因果推理链故障诊断模型。已有基于推理链的电网故障诊断方法存在推理子链个数较多、构造较复杂、未考虑保护和断路器缺失、错误等不足,针对每个疑似故障元件根据保护逻辑建立原始的动态推理链,对误报、漏报的报警信息进行纠正,形成改进后的动态推理链。对推理链中各保护和断路器动作的可信度进行评估,将各保护和断路器动作信息从二值逻辑的0或1模糊化为0,1之间的可信度值。针对每个疑似故障元件的推理链,建立相应的贝叶斯网络模型,通过贝叶斯反向推理得到元件的故障概率。典型电网的多个故障案例验证了所提方法的有效性。
A fault diagnosis method of dynamic causalreasoning chain model combined with Bayesian network is proposed. Existing power system fault diagnosis method based on reasoning chain have large number of reasoning sub-chains with more complex structure, not considering deficiency and abnormity of protections and breakers. In this paper original reasoning chain is established for each suspected fault component according to the protection logics. After correcting missed and abnormal alarming messages, the reasoning chains are improved. By evaluating operation credibility of protections and breakers, their operation status in reasoning chain are changed from binary 0 or 1 into fuzzy values [0,1]. Finally, for each reasoning chain of suspected fault component, a corresponding Bayesian network model is created. By reverse Bayesian reasoning, fault probability is obtained. Effectiveness of the method is verified by multiple fault cases of a typical power grid.
出处
《电网技术》
EI
CSCD
北大核心
2017年第4期1315-1323,共9页
Power System Technology
基金
国家自然科学基金项目(51377137)~~
关键词
电力系统
故障诊断
因果推理链
可信度
power system
fault diagnosis
causality inference chain
credibility
