基于驱动电机电流与振动信号的有载分接开关故障诊断方法

Fault Diagnosis Method for On-load Tap Changer Based on Driven Motor Current and Vibration Signal

有载分接开关(on-load tap changer,OLTC)是电力变压器中唯一可以运动的机构,具有复杂的机械结构。为了解决有载分接开关机械故障诊断中样本库无法涵盖所有故障类型的问题,文中提出了一种基于驱动电机电流与振动信号的有载分接开关故障诊断方法。首先,文中研究了有载分接开关多源监测信号的特征提取方法,提出了驱动电机电流信号的持续时间、绝对值之和以及开关本体振动信号的经验模态分解能量等特征。接着,文中结合随机森林(random forest)和孤立森林(isolation forest)算法,提出了一种与异常检测相融合的有载分接开关故障诊断方法。最后,文中对VRG型有载分接开关进行了实验,采集了正常、护罩松动、连杆窜动、触头松动4类样本数据,实验结果表明,文中所提出的方法对已知故障的诊断准确率达97.78%,对未知故障的识别准确率达92.5%。

On-load tap change(rOLTC)with complicated mechanical structure is the only moving mechanism in pow-er transformer. In view of solving the problem that all fault types are not covered in the sample database in the me-chanical fault diagnosis of OLTC,a kind of fault diagnosis method for OLTC based on driven motor current and vi-bration signal is proposed in this paper. Firstly,the feature extraction method for multiple-source monitoring signal ofOLTC is studied,and such features as duration of current,sum of absolute current in driven motor and empiricalmode decomposition(EMD)energy of vibration are proposed. Then,in combination with Random Forest(RF)algo-rithm and the Isolation Fores(tIF)algorithm,a kind of fault diagnosis method integrating with abnormal detection isproposed. Finally,the experiment for type VRG on-load tap changer is performed and four types of sample data asnormal data,looseness of shield,movement of connecting rod and looseness of contact are collected for the VRG-OLTCs in the experiment. The experiment results show that the diagnosis accuracy of the proposed method for theknown and unknown faults is 97.78% and 92.5% respectively.