基于LSTM网络的同步电机励磁绕组匝间短路故障预警  

Field Winding Short Circuit Fault Warning of Synchronous Motor Excitation Winding Based on LSTM Network

作　　者：李俊卿[1] 陈雅婷 LI Junqing;CHEN Yating(School of Electrical and Electronic Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《电力科学与工程》2020年第6期37-42,共6页Electric Power Science and Engineering

摘　　要：构建了一种LSTM故障预警模型对同步发电机励磁绕组匝间短路早期故障进行预警。以发电机组正常情况下的数据训练LSTM网络,求得正常情况下因变量(励磁电流,定子振动)与自变量(定子三相电压,定子三相电流,励磁电压,有功功率,无功功率)之间的函数关系。将2个输出量进行数据融合,计算正常情况下的总偏移距离,定义预警阈值,当总偏移距离大于预警阈值时判定为故障。实验证明,该模型可以实现励磁绕组匝间短路早期故障的预警。A fault warning model of LSTM is built to predict the early fault of short circuit between turns of excitation winding of synchronous generator.The LSTM network was trained with the data of generator set under normal conditions,and the functional relationship between dependent variables(excitation current,stator vibration),and independent variables(stator three-phase voltage,stator three-phase current,excitation voltage,active power,reactive power)under normal conditions were obtained.The two outputs were fused to calculate the total deviation distance under normal conditions,and the warning threshold was defined.When the total deviation distance was greater than the warning threshold,the fault was determined.Experiments show that the model can be used to warn the early fault of short circuit between turns of excitation winding.

关 键 词：故障预警 LSTM网络 同步发电机 励磁绕组匝间短路 

分 类 号：TM341[电气工程—电机]