基于改进Elman神经网络的变压器故障诊断方法

Transformer fault diagnosis method based onimproved Elman neural network

针对变压器故障诊断困难的问题,提出了一种基于改进Elman神经网络的变压器故障诊断方法.利用核主成分分析算法对Elman神经网络进行改进,通过引入累积贡献率计算得到改进Elman神经网络最优输入特征参数,基于数据集和测试集将改进Elman神经网络应用于变压器的常见故障诊断中,并将试验结果与其他故障诊断方法的诊断结果进行对比.结果表明,所提方法诊断准确率超过90%,与支持向量机和BP神经网络诊断方法相比,该方法的诊断准确率更高.

Aiming at the difficulty of transformer fault diagnosis,a transformer fault diagnosis method based on improved Elman neural network was proposed.The kernel principal component analysis algorithm was used to improve the Elman neural network.The optimal input characteristic parameters for the improved Elman neural network were calculated and obtained by introducing the cumulative contribution rate.In addition,the improved Elman neural network was applied to the diagnosis of common transformer faults based on data set and test set,and the test results were compared with the diagnostic results by other fault diagnosis methods.The results show that the diagnostic accuracy of as-proposed method is more than 90%.Compared with the diagnosis methods of support vector machine and BP neural network,the diagnostic accuracy of as-proposed method is extremely higher.