CAJ | 학술논문

以列车轴承运行的振动加速度信号为研究对象，采用随机共振进行轴承振动信号的提取，同时基于主成分分析的方法实现对目前旋转机械常采用的23个混合域的故障特征参量进行分析，得到合适的轴承故障特征集。利用BP神经网络对以上内容的有效性进行验证。实验表明，结合2种方法得到的轴承故障诊断结果正确率可达到90％以上。
이열차축승운행적진동가속도신호위연구대상，채용수궤공진진행축승진동신호적제취，동시기우주성분분석적방법실현대목전선전궤계상채용적23개혼합역적고장특정삼량진행분석，득도합괄적축승고장특정집。이용BP신경망락대이상내용적유효성진행험증。실험표명，결합2충방법득도적축승고장진단결과정학솔가체도90％이상。
T his paper takes the train axle box bearing running vibration acceleration signal as the re‐search object ,studies the signal extraction and fault feature extraction ,which are the key aspects of real-time monitoring .To extract the effective bearing fault signal ,this paper carries out the research on the stochastic resonance method to extract the effective bearing fault signal ,and adopts principal component analysis (PCA) method to reduce the dimension of 23 mixed‐domain fault characteristic pa‐rameters and get the appropriate principal component characteristic parameters .Finally ,BP neural net‐work based bearing fault diagnosis system is designed to verify the validity of the research results of this thesis .The experimental results show that the accuracy of bearing fault diagnosis can reach more than 90% by combining the two methods .