船舶发电机定子绕组匝间短路故障预警  

Warning of Interturn Short-circuit Fault in Stator Winding of Ship Generator

作　　者：姜景强 施伟锋[1] 刘宇航 谢嘉令 JIANG Jingqiang;SHI Weifeng;LIU Yuhang;XIE Jialing(Logistics Engineering College,Shanghai Maritime University,Shanghai 201306,China)

机构地区：[1]上海海事大学物流工程学院,上海201306

出　　处：《电力系统及其自动化学报》2024年第6期145-151,共7页Proceedings of the CSU-EPSA

基　　金：上海市科技计划项目(20040501200)。

摘　　要：为保障船舶的安全航行,需要对船舶发电机定子绕组早期匝间短路故障进行预警。首先,通过同步发电机的数学模型对正常运行和不同程度的匝间短路故障进行仿真,选取负序电流分量和电磁转矩作为故障特征量并进行信息融合;然后,使用CNN-BiLSTM网络对特征量进行回归预测得到网络预测的残差;最后,根据残差的变化趋势和设定的阈值来判断故障的发展情况。结果表明,建立的模型可以实现定子绕组匝间短路故障的预警,并且基于自适应原理设置的动态阈值会比固定阈值提前确认故障。To ensure the safe navigation of ships,a warning of interturn short-circuit fault in the stator windings of ship generators is required.First,a synchronous generator in the cases of normal operation and different degrees of interturn short-circuit fault is simulated by its mathematical model,the negative-sequence current components and electromag-netic torque are selected as fault feature quantities,and information fusion is performed.Then,a convolutional neural network-bidirectional long short-term memory(CNN-BiLSTM)network is used for the regression prediction of feature quantities,thus obtaining the residuals of network prediction.Finally,the development of fault is judged according to the trend of residuals and the set threshold value.Results show that the established model can realize the warning of in-terturn short-circuit fault in stator winding,and the dynamic threshold value which is set based on the adaptive princi-ple will confirm the fault earlier than the fixed threshold value.

关 键 词：船舶发电机 匝间短路 阈值 故障预警 卷积神经网络 双向长短期记忆网络 

分 类 号：U672[交通运输工程—船舶及航道工程]