基于LS-SVM方法的变压器铁心振动状态监测  

Monitoring of transformer core vibration based on LS-SVM

作　　者：林聪 何兆磊[1] 朱梦梦 赵静[1] Lin Cong;He Zhaolei;Zhu Mengmeng;Zhao Jing(Metering Center of Yunnan Power Grid Co.,Ltd.,Yunnan Kunming,650011,China;Key Laboratory of Electric Power Measurement of China Southern Power Grid,Yunnan Kunming,650217,China;Yunnan Electric Power Research Institute,Yunnan Power Grid Co.,Ltd.,Yunnan Kunming,650217,China)

机构地区：[1]云南电网有限责任公司计量中心,云南昆明650011 [2]中国南方电网公司电能计量重点实验室,云南昆明650217 [3]云南电网有限责任公司电力科学研究院,云南昆明650217

出　　处：《机械设计与制造工程》2023年第11期73-76,共4页Machine Design and Manufacturing Engineering

基　　金：云南电网有限责任公司科技项目(YNKJXM20191481)。

摘　　要：为了节省布置数据传输线的成本,利用无线射频信号技术(RFID)设计了RFID传感器。将振动信号产生的Hilbert边际谱作为故障特征,以总谐波畸变(THD)为状态指数,通过基于最小二乘支持向量机(LS-SVM)故障诊断模型对绕组与铁芯故障进行诊断,并以某110 kV变压器为例,选择铁芯松动、铁芯形变、绕组松动以及绕组形变4种典型的机械故障类型进行了故障诊断预测。结果表明,基于LS-SVM的变压器故障诊断算法具有良好的泛化能力,能够可靠地存储和快速查询海量的状态数据,从而大大提高状态评估和故障诊断的准确性;在系统允许的误码率范围内能够实现高精度的故障诊断,从而为变压器故障诊断和预测提供参考。In order to save the cost of arranging data transmission lines,RFID sensors are designed by using Radio Frequency Signal Technology(RFID).Taking Hilbert marginal spectrum generated by vibration signal as fault feature and total harmonic distortion(THD)as state index,winding and iron core faults are diagnosed by fault diagnosis model based on least squares support vector machine(LS-SVM).Taking a single-phase transformer and three-phase transformer in a mining area as examples,four typical mechanical fault types are selected:iron core looseness,iron core deformation,winding looseness and winding deformation.Eight examples are tested for fault diagnosis and prediction.The results show that the transformer fault diagnosis algorithm based on LS-SVM has good generalization ability,It can reliably store and quickly query massive state data,so as to greatly improve the accuracy of state evaluation and fault diagnosis.It can realize high-precision fault diagnosis within the allowable bit error rate range of the system,and provide reference for transformer fault diagnosis and prediction in mining area.

关 键 词：最小二乘支持向量机 变压器 绕组铁芯 状态监测 

分 类 号：TM855[电气工程—高电压与绝缘技术]