BP神经网络在油浸式变压器过热性故障诊断中的应用

针对充油变压器绝缘故障诊断的三比值法的局限性,建立以变压器油中溶解气体含量为样本数据,对不同的隐含层数目进行仿真分析,通过比较确定了适用于变压器绝缘故障诊断的BP神经网络模型。研究表明,这种方法提高了神经网络的收敛速度,符合电力变压器故障诊断系统的实际情况,准确率高。

Application of BP Neural Network to Overheating Fault Diagnosis for Oil-immersed Transformer

BP neural network model for transformer insulation fault diagnosis is established,according to the limitations of three-ratio method used for the insulation fault diagnosis of oil-immersed transformer.The model is based on the sample data of the content of dissolved gas in transformer oil,and the analysis is based on the number of hidden layers.The results shows that the proposed method can improve the convergence rate of neural network with higher accuracy,which accords with the actual situation of the fault diagnosis system of power transformer.