摘要
为了能更准确地诊断出发电机转子绕组匝间短路故障,基于改进的双层动态均值聚类分析的径向基神经网络对转子绕组匝间短路故障进行了诊断。同时,通过对同步发电机转子绕组故障信号进行分析,并把从中提取的故障信号的特征量作为学习样本,通过改进的径向基神经网络的训练,使构造的径向基神经网络能够反映样本的特征向量和转子绕组匝间不同程度的短路类型之间的映射关系,从而达到故障诊断的目的。仿真实验表明,该算法可以进行有效的故障诊断,精度优于传统的反向传播BP(back propagation)神经网络。
In order to more accurate diagnosis of generator rotor winding inter-turn short-circuit fault,a radial basis function neural network,which based on an improved two-tier dynamic means clustering analysis diagnoses the rotor winding inter-turn short circuit fault in this paper.At the same time,this paper analyses the synchronous generator rotor＇s winding fault signal,and extract the fault signal characteristic quantities as learning samples.Through the improved RBF neural network＇s training,we enable construction of radial basis function neural network can reflect the characteristics of the sample vector and the rotor winding inter-turn short circuit in varying degrees between the types of mapping relations,so as to achieve the purpose of fault diagnosis.The simulation results show that the algorithm can be of effective fault diagnosis and better accuracy than that of conventional BP（back propagation）neural network.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2011年第1期114-117,共4页
Proceedings of the CSU-EPSA
关键词
同步发电机
转子绕组
匝间短路
径向基神经网络
故障诊断
synchronous generator
rotor winding
inter-turn short-circuit
radial basis function neural network
fault diagnosis
