基于LM-Elman神经网络的电力变压器故障诊断  

Fault diagnosis method of transformer based on LM-Elman neural network

作　　者：孙婧雅[1] 倪远平[1] 李凌[1] 刘守盎[1] 

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650051

出　　处：《云南大学学报（自然科学版）》2009年第S2期50-54,共5页Journal of Yunnan University(Natural Sciences Edition)

摘　　要：为更加准确的检测变压器状态和及时发现变压器可能存在的故障,将Levenberg-Marquardt算法与Elman神经网络相结合,应用于变压器故障诊断中,并与BP网络及三比值法的故障诊断率进行了比较.结果表明该方法的诊断率相比其他2种方法具有一定的优势.In order to detect the state of power transformers more accurately and find any possible failure earlier,in this paper,the method which combined Levenberg-Marquardt algorithm and Elman neural network is used in fault diagnosis of power transformer.It is compared the BP neural network and three ratio method.The results show that the diagnostic accuracy based on LM-Elman neural network is higher than the other two methods.

关 键 词：ELMAN网络 LEVENBERG-MARQUARDT算法 三比值法 电力变压器 故障诊断 

分 类 号：TM41[电气工程—电器]