基于模糊神经网络的变压器局部放电小波奇异性故障诊断

Wavelet Singularity Fault Diagnosis of Partial Discharge in Transformer Based on Fuzzy Neural Networks

变压器故障中除油变质劣化之外,另一种重要故障即为局部放电。将小波奇异性检测理论与模糊神经网络结合起来,对变压器局部放电时的信号先利用小波奇异性检测理论求出奇异性指数作为故障诊断的特征参数,然后将其模糊化作为神经网络的输入,输出为各种故障的隶属度,进而判断变压器存在何种故障的可能性。仿真结果表明此方法用于变压器局部放电故障诊断是行之有效的。

In transformer fault,except deterioration of oil,the other major fault shall be partial discharge.The wavelet singularity detection theory and fuzzy neural network was combined.Firstly,for the signal while transformer partial discharge,wavelet singularity detection theory was used to calculate singularity index and took it as fault diagnosis characteristic parameters.Secondly,it was fuzzed as the input of neural network,the output were various failures of the membership grade.And then,determining the possibility of transformer what fault there was.Simulation results show that the method used for transformer fault diagnosis of partial discharge is effective.