基于d-q变换及WOA-LSTM的异步电机定子匝间短路故障诊断方法  

Asynchronous motor stator turn-to-turn short circuit faultdiagnosis based on d-q transform and WOA-LSTM

作　　者：王喜莲[1] 秦嘉翼 耿民 WANG Xilian;QIN Jiayi;GENG Min(School of Electrical Engineering,Beijing Jiaotong University,Beijing 100044,China;Motor Vehicle Maintenance Department,CRRC Tangshan Co.,Ltd.,Tangshan 063035,China)

机构地区：[1]北京交通大学电气工程学院,北京100044 [2]中车唐山机车车辆有限公司动车检修部,河北唐山063035

出　　处：《电机与控制学报》2024年第6期56-65,共10页Electric Machines and Control

摘　　要：为了实现对异步电机定子绕组匝间短路故障的可靠在线诊断,提出一种基于d-q变换及鲸鱼优化算法(WOA)优化的长短期记忆网络(LSTM)的故障诊断方法。通过理论推导可知,d-q变换可有效提取定子电流中的特征频谱数据。采用鲸鱼优化算法对长短期记忆网络中的3个关键参数进行优化,建立WOA-LSTM故障分类模型。为了验证基于d-q变换和WOA-LSTM故障诊断方法的有效性,分别以小波变换、快速傅里叶变换及d-q变换提取电流频谱数据作为输入数据集,以一台YE2-100L1-4型异步电机为实验对象进行实验验证。研究结果表明:相比于小波变换及快速傅里叶变换,采用d-q变换能更准确的提取出定子电流中的故障特征,更精确地反映电机故障状态,有助于提高故障分类准确率;相比于传统的LSTM算法,经WOA优化后的LSTM算法分类准确率可达98.3%,能可靠地实现不同程度匝间短路故障的诊断。In order to realize reliable online diagnosis of inter-turn short-circuit faults in asynchronous motor stator windings,a fault diagnosis method based on d-q transform and whale optimization algorithm(WOA)optimized long-short-term memory network(LSTM)was proposed.It is known through theoretical derivation that the d-q transform can effectively extract the characteristic spectral data in the stator current.The whale optimization algorithm was used to optimize the three key parameters in the long short-term memory network and the WOA-LSTM fault classification model was established.In order to verify the effectiveness of the fault diagnosis method based on d-q transform and WOA-LSTM,wavelet transform,fast Fourier transform and d-q transform were used to extract the current spectrum data as the input data set,and a YE2-100L1-4 asynchronous motor was used as the experimental object for experimental verification.The results show that compared with wavelet transform and fast Fourier transform,the d-q transform can more accurately extract the fault features in the stator current,more accurately reflect the fault state of the motor,and help to improve the fault classification accuracy.Compared with the traditional LSTM algorithm,the classification accuracy of LSTM algorithm optimized by WOA can reach 98.3%,which can reliably realize the diagnosis of inter-turn short-circuit faults of different degrees.

关 键 词：异步电机 故障诊断 定子绕组匝间短路 d-q变换理论 鲸鱼优化算法 长短期记忆神经网络 

分 类 号：TM343[电气工程—电机]