基于BRBP神经网络的转子绕组匝间短路故障诊断方法  

A Rotor Winding Inter-turn Short-circuit Fault Diagnosis Method Based on BRBP Neural Networks

作　　者：李红连[1,2] 唐炬[1] 

机构地区：[1]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400044 [2]成都大学电子信息工程学院,成都610106

出　　处：《中国农村水利水电》2013年第2期152-155,158,共5页China Rural Water and Hydropower

基　　金：国家重点基础研究发展计划("973"计划)(2009CB724506);国家自然科学基金(11205022);重庆市自然科学基金(CSTC2008BB0327);四川省教育厅科技项目(12ZB170)

摘　　要：为了能更准确、容易地诊断出同步发电机转子绕组匝间短路故障,提出了一种基于贝叶斯正则化反向传播(BRBP)神经网络的故障诊断方法。该方法利用正常运行时不同工况下的机端电压、有功功率、无功功率和转子励磁电流来建立励磁电流的BRBP神经网络预测模型;利用该模型预测正常运行时所需励磁电流,与实测的励磁电流进行比较,相对误差超过阈值就诊断为发生匝间短路故障。通过微型同步发电机动模实验表明,该方法的精度优于BP神经网络法,并且参数设置简单、易于移植和训练速度快,对同步发电机转子绕组匝间短路故障的监测与诊断是有效的。In order to diagnose synchronous generator rotor winding inter-turn short-circuit fault more accurately and easily, a novel fault diagnosis method is put forward, based on Bayesian regularization back-propagation （BRBP） neural networks. Sample data un- der different fault-free operating conditions are measured and collected, including terminal parameters （voltage, active power, reac- tive power） and field currents, then a BRBP neural network model is established to predict field currents. Input to the model with measured terminal parameters, and a predicted field current is obtained. Finally, the predicted field currents are compared with the corresponding measured field currents, and a synchronous generator rotor winding inter-turn short-circuit fault is diagnosed when relative error exceeds a specific threshold. The dynamic simulation results of micro-synchronous generator show that, the method is better. It is easily applied to other synchronous generators, and is an effective rotor winding inter-turn short-circuit fault diagnosis method for synchronous generators.

关 键 词：同步发电机 转子绕组 匝间短路 故障诊断 贝叶斯正则化反向传播神经网络 

分 类 号：TM713[电气工程—电力系统及自动化]