基于BP神经网络的转子匝间短路故障识别方法  

Fault Identification of Rotor Windings Inter-turn Short Circuit Based on BP Neural Network

作　　者：王晓华[1] 李永刚[1] 樊静[1] 刘玉飞[2] 

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003 [2]邯郸建筑设计有限责任公司,河北邯郸056002

出　　处：《电力科学与工程》2010年第2期5-8,共4页Electric Power Science and Engineering

基　　金：国家自然科学基金资助项目(50677017)

摘　　要：分析了发电机转子绕组发生匝间短路后的电磁特性,得到了匝间短路后的特征参数。在确定的运行状态下,发电机励磁磁动势故障前后维持不变而励磁电流增大。据此选取故障样本,将BP神经网络应用于发电机匝间短路故障识别。BP神经网络不依赖于发电机的数学模型及其结构参数。最后实测了故障发电机的特征信号,与理论分析结果基本吻合。The electromagnetic characteristic of turbine generator were analyzed in details when rotor windings inter-turn short circuit has happened. This paper revealed that exciting magnetic force Ff is constant in a fixed condition whereas the exciting current If increases in case of rotor inter-turn fault. Based on the theory, BP （back propagation） neural network can be adequately trained and diagnosis rotor winding inter-turn short circuit. BP neural network is independent on mathematic models and parameters of turbine- generator. Finally practically acquired dynamic experiment data of generator, the results of verification showed that the theory analysis is right.

关 键 词：发电机 匝间短路 磁动势 神经网络 

分 类 号：TM311[电气工程—电机]