摘要
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.
作者
王英洁
曹铁男
WANG Ying-jie;CAO Tie-nan(Research Institute,China Southern Power Grid Co.Ltd.,Guangzhou 510663,China)
出处
《沈阳工业大学学报》
EI
CAS
北大核心
2021年第3期254-258,共5页
Journal of Shenyang University of Technology
基金
国家自然科学基金项目(61372071)
中国南方电网有限责任公司专项课题(ZBKJXM20170060).
关键词
ELMAN神经网络
核主成分分析算法
变压器
故障诊断
参考向量
特征参数
数据集
测试集
Elman neural network
kernel principal component analysis
transformer
fault diagnosis
reference vector
characteristic parameter
data set
test set
