模糊输入的概率神经网络在变压器故障诊断中的应用  

Application of Probabilistic Neural Network for Fault Diagnosis of Power Transformer Based on Fuzzy Input

作　　者：段慧达[1] 刘学军[1] 刘文斌[1] 

机构地区：[1]北华大学电气信息工程学院,132021

出　　处：《电气应用》2007年第7期28-30,共3页Electrotechnical Application

基　　金：吉林省科技厅项目(200505263)

摘　　要：将概率神经网络引入到电力变压器的故障诊断中,并提出了利用模糊数学的知识对神经网络输入进行预处理的方法。采用改良的三比值方法来实现故障诊断,建立了基于模糊输入的神经网络故障诊断模型。通过对神经网络模型的实例训练及仿真,证明了此方法有较好的收敛性,提高了故障诊断的有效性和准确性。Probabilistic Neural Network is used in fault diagnosis of power transformer. A method using fuzzy math theory to deal with the data of the neural network input is proposed. Fault diagnosis of power transformer based on improved three ratio method is realized. And a fault diagnosis model of neural network based on fuzzy input is constituted. The simulation diagnosis demonstrates the effectiveness, astringency and veracity of the proposed algorithm.

关 键 词：概率神经网络 模糊 故障诊断 气相色谱分析法 

分 类 号：TM407[电气工程—电器]