基于时间溯因推理的电网诊断报警方法  

Temporal Abductive Reasoning Based Diagnosis and Alarm for Power Grid

作　　者：康泰峰[1] 吴文传[1] 张伯明[1] 李鹏[2] 张文峰[2] 

机构地区：[1]电力系统及发电设备控制和仿真国家重点实验室(清华大学电机系),北京市海淀区100084 [2]中国南方电网公司,广东省广州市510623

出　　处：《中国电机工程学报》2010年第19期84-90,共7页Proceedings of the CSEE

基　　金：国家自然科学基金项目(50707013);中国南方电网有限责任公司总部科技项目(K0812)~~

摘　　要：电网发生故障时有大量报警产生,这些报警信息通常具有时序约束关系,对电网故障诊断具有重要意义。溯因推理是基本的推理方式之一,根据已知事实或规律性知识,推断产生这一结果的原因。溯因推理不仅可以揭示故障原因,还可以得到故障过程,在电网诊断报警领域中有良好的应用前景。将溯因推理方法引入电网诊断报警中。首先,采用时间约束和溯因推理规则来描述报警信息之间的时序关系和逻辑关系;其次,提出基于报警信息的时间溯因推理方法求得源事件和故障事件链。所提出的方法可识别出断点报警、错误报警和缺失报警,减少报警结果的不确定性。该方法可以作为单独的报警模块,也可以与其他故障诊断方法相结合进行深入诊断。When faults occur in power grid, a lot of alarms are generated. There are temporal constraints among these alarms and such temporal information is essential for fault diagnosis. As one of basic inference methods for diagnosis problems, abductive reasoning is known as using domain theories to infer about reasons or explanations of observations and results. Abductive reasoning based inference methods can find causes and evolution processes, which is useful for power grid diagnosis and alarm problems. In this paper, abductive reasoning was introduced into power grid diagnosis and alarm. Firstly, the temporal and logical relationships of faults and alarms in power grid were formulated as temporal constraints and inference formulas. Then a suit of temporal abductive reasoning based algorithms were proposed to solve fault evolution process using observed alarms. The proposed algorithms can identify break alarms, exception alarms and missing alarms which can reduce the uncertainty of diagnosis results. These algorithms can be developed as a function of intelligent alarm method for power grid, or used for further fault diagnosis accompanying with other methods.

关 键 词：溯因推理 时间约束 电力系统 故障诊断 智能报警 

分 类 号：TM71[电气工程—电力系统及自动化]