基于特征相似度的同步发电机绝缘故障诊断  

Insulation fault diagnosis of synchronous generator based on characteristic similarity

作　　者：崔小鹏[1] 王公宝[1] 林城美 饶金[1] 

机构地区：[1]海军工程大学舰船综合电力技术国防科技重点实验室,武汉430033

出　　处：《海军工程大学学报》2014年第5期14-19,共6页Journal of Naval University of Engineering

基　　金：国家自然科学基金资助项目(51077129)

摘　　要：以同步发电机为分析对象,利用粗糙集可有效地处理和分析不精确的信息和知识的优点,从历史信息和决策中进行了特征量选取以及属性约简。对于无法处理的属性约简,采用条件限制得到能够有效反映属性的重要程度特征量,并以此计算特征量权重系数,消除主观因素的影响。在此基础上,针对励磁绕组绝缘故障,选择与设备健康状态直接相关特征作为判断对象,与同机组历史数据进行相似度分析,计算得到了其故障因子。设计出快速判断相关状态的算法,给出了判定准则并进行了编程实现。实验验证表明:该诊断算法能够对设备故障进行有效判定。With the synchronous generator as a research object, this paper shows the characteristic quantity from the historical information and decision-making data by using the rough set to effectively process and analyze the inaccurate information and knowledge. For the attribute that is unable to be simplified, certain conditions should be constrained so as to reflect the importance of the attribute effectively. Then, the weighting factor of the characteristic quantity is calculated to eliminate the impact of subjective factors. On this basis, as far as the field winding insulation failure is concerned, the selection of the features which are directly related to the healthy status of the device as a judgment target and the similarity analysis of the historical data in the same conditions lead to the calculated failure factor. Thus, the algorithm that quickly determines the relevant state can be designed and the criteria can be given for programming. The experiments verify that the diagnostic algorithm can effectively deal with faults in the equipment.

关 键 词：粗糙集 属性约简 故障诊断 相似度分析 

分 类 号：TM359[电气工程—电机]