基于粗糙集的智能变电站保护设备仿生故障诊断方法  

A rough set-based bio-inspired fault diagnosis method for smart substation protection equipment

作　　者：梁文武 朱维钧 李辉 严亚兵 董国琴 王璇 龚锦霞 LIANG Wenwu;ZHU Weijun;LI Hui;YAN Yabing;DONG Guoqin;WANG Xuan;GONG Jinxia(State Grid Hunan Electric Power Co.,Ltd.Research Institute,Changsha 410000,China;Chenzhou Power Supply Branch,State Grid Hunan Electric Power Co.,Ltd.,Chenzhou 423000,China;Shanghai University of Electric Power,Shanghai 200090,China)

机构地区：[1]国网湖南省电力有限公司电力科学研究院,湖南长沙410000 [2]国网湖南省电力有限公司郴州供电分公司,湖南郴州423000 [3]上海电力大学,上海200090

出　　处：《电力系统保护与控制》2021年第21期132-140,共9页Power System Protection and Control

基　　金：国家自然科学基金项目资助(51607112);国家电网公司科技项目资助“第三代智能变电站一次设备就地模块运检关键技术研究及应用”(521A5180016)。

摘　　要：变电站保护设备报警信息的不精确性和不确定性会严重影响变电站故障诊断结果的准确性和可靠性。针对这一问题,提出了一种基于粗糙集的变电站保护设备仿生故障诊断方法。该算法由变电站子区域划分法、粗糙集属性约简算法、二元推理脉冲神经膜系统(BRSNPS)和并行推理算法4个关键部分组成。具体来说,采用变电站子区域划分方法和粗糙集约简算法,为每个子区域寻找约简的故障产生规则集,简化了问题的复杂性并可处理故障报警信息的不确定性。然后,提出了BRSNPS及其推理算法,实现了故障信息的展示和分析,获得了准确的故障诊断结果。由于粗糙集和spike神经系统的协作,不需要历史统计和专业知识,问题的规模减少。最后,基于真实的110 kV和750 kV变电站,对所提出的算法进行了实验验证,结果表明该方法优于其他方法。The inaccuracy and uncertainty of alarm information for substation protection equipment will seriously affect the accuracy and reliability of fault diagnosis results.A rough set-based bionic fault diagnosis method for substation protection equipment is proposed.The algorithm consists of four key parts:substation sub-area division method,rough set attribute reduction algorithm,Binary Reasoning Pulse Neural Membrane System(BRSNPS)and parallel reasoning algorithm.Specifically,the sub-region division method of the substation and the rough set reduction algorithm are used to find the reduced fault generation rule set for each sub-region.The complexity of the problem is simplified and the uncertainty of the fault alarm information can be handled.Then,BRSNPS and its reasoning algorithm are proposed.These realize the display and analysis of fault information and obtain accurate fault diagnosis results.Because of the collaboration of rough set and spike nervous system,historical statistics and professional knowledge are not required,and the scale of the problem is reduced.Finally,based on real 110 k V and 750 k V substations,the algorithm proposed is verified by experiment,and the results show that this method is better than other methods.

关 键 词：智能变电站 故障诊断 脉冲神经膜系统 粗糙集 

分 类 号：TM63[电气工程—电力系统及自动化]