CAJ | 학술논문

本文研究电机转子断条故障的诊断机理,选取小波包分析作为信号处理的前置手段,得到输出神经网络的特征向量.神经网络通过学习训练得到诊断结果,将小波的局部特性和神经网络的自学习特性结合起来,使诊断系统具有自适应分辨性和良好的容错性.本研究已经应用于电机故障诊断教学实验,针对特征向量提取方法实现了故障的诊断和区分.
본문연구전궤전자단조고장적진단궤리,선취소파포분석작위신호처리적전치수단,득도수출신경망락적특정향량.신경망락통과학습훈련득도진단결과,장소파적국부특성화신경망락적자학습특성결합기래,사진단계통구유자괄응분변성화량호적용착성.본연구이경응용우전궤고장진단교학실험,침대특정향량제취방법실현료고장적진단화구분.
This paper is concerned with fault diagnosis mechanism of broken rotor bars, choosing Wavelet Packet Analysis as the pre-positive means to get the eigenvectors of output neutral network. Neutral network, through learning and training, got the diagnosis results. By combining neural-network’s self-study specific property with wavelets’ partial specific property, it possessed self-suitable distinguish and good tolerance specific property. This research has been applied in teaching experiments of motor fault diagnosis, and feature extraction made fault diagnosis realized and the fault can be distinguished as well.