基于学习矢量神经网络的电力变压器故障诊断

Research on transformer fault diagnosis based on LVQ network

提出了一种基于学习矢量化(Learning Vector Quantization)神经网络分类器结合油中溶解气体分析(DGA)的变压器故障诊断方法,把故障信号的训练样本输入到LVQ网络中进行训练,利用网络的竞争性将分类信息转变成使用者所定义的类别。训练和测试结果表明了该方法的有效性。

A novel specific transformer fault diagnostic method based on LVQ （ Learning Vector Quantization ） network classifiers with gas-in-oil analysis （DGA） is presented in this paper. Training samples are input the LVQ network, and then the network outputs the scale vectors. Becausse the scale distributions of neurons of the competitive level are different, transformer fault can be ascertained and diagnosed. Finally, the analytical result of practical sample verifies the accuracy of the proposed idea.