智能电网中基于批标准化LSTM的互感器故障诊断技术  

Online monitoring and fault diagnosis technology of transformers based on the LSTM with batch normalization

作　　者：曹志强 陈洁[1] CAO Zhiqiang;CHEN Jie(School of Electrical Engineering,Xinjiang University,Urumqi 830046,China)

机构地区：[1]新疆大学电气工程学院,新疆乌鲁木齐830046

出　　处：《电力科学与技术学报》2023年第6期152-158,共7页Journal of Electric Power Science And Technology

基　　金：国家自然科学基金(61963034)。

摘　　要：互感器是高压电力系统中的必备设备之一,一旦互感器发生故障,将会导致保护装置拒动误动,造成电网瘫痪。传统的故障诊断和分类方法首先从原始过程数据中提取特征,然后采用特定的分类器进行诊断,缺乏对原始数据中动态信息的自适应处理。为了提高传统循环神经网络在诊断中的准确度,并考虑到长短记忆神经网络诊断时间较长的缺陷,提出一种基于批标准化的LSTM模型的故障诊断方法。该方法无需进行特征提取和分类器设计,直接对故障进行分类,并且能自适应学习动态故障数据。通过与其他故障诊断方法比较,该方法的诊断精度和诊断性能较高,在互感器故障诊断领域具有良好的应用价值。As one necessary equipment in the high-voltage power system,once the transformer fails,protection devices may refuse to operate and cause the malfunction of power grids.Traditional current transformer fault diagnosis and classification methods firstly extract features from the input operation data,and then use a specific classifier to diagnosis,which lacks adaptive update processing for dynamic input information.In order to further improve the accuracy of traditional recursive neural networks,the process efficiency of long short-term memory neural networks,this paper proposes a fault diagnosis method based on the LSTM model of batch normalization(BN).This method does not require feature extraction and classifier design steps,where the fault can be classified directly,and can also be updated adaptively.Compared with other fault diagnosis methods,this method has higher diagnostic accuracy and diagnostic performance,which validating its good application value in the field of current transformer fault diagnosis.

关 键 词：批标准化 LSTM神经网络 在线监测 故障诊断 智能电网 互感器 

分 类 号：TM933[电气工程—电力电子与电力传动]