基于频谱分析的发电机转子系统故障诊断  

Fault diagnoses of generator rotor system based on frequency spectrum analysis

作　　者：周政新[1,2] 孙秋冬[2] 朱刚[2] 李蓓蓓[2] 毛建民 

机构地区：[1]东华大学信息学院,上海200051 [2]上海第二工业大学电子与电气工程学院,上海201209 [3]外高桥第二发电厂,上海200131

出　　处：《电子测量技术》2011年第2期80-83,共4页Electronic Measurement Technology

基　　金：上海市教委重点学科建设项目资助基金(J51801)

摘　　要：通过对发电机转子线圈的电磁特性分析,得出转子绕组主磁场变化特性,以及转子匝间短路在定子绕组并联支路出现的电势差和环流。采用频谱特征向量作为学习样本,建立径向基神经网络的频谱特征和故障类型的映射关系。应用表明,采用该方法能有效地诊断出发电机转子系统的故障。Through the analysis of the electromagnetic properties for the rotor coil of the generator,the main magnetic field variation characteristics for the rotor winding and the short circuit fault between rotor inter-turns causing the potential difference and circulation between the parallel branches was obtained.The spectral feature vector was treated as learning sample,the Radial Basis Function neural network can reflect the mapping relations between spectral features and fault types.The practical application showed that the spectral analysis method and the RBF neural network can effectively improve the diagnostic accuracy and efficiency.

关 键 词：发电机 监测与诊断 频谱分析 RBF神经网络 

分 类 号：TM85[电气工程—高电压与绝缘技术]