摘要
针对超低速滚动轴承故障诊断困难问题,提出一种自适应噪声的完备集合经验模态分解(CEEMDAN)与深度信念网络(DBN)相结合的超低速滚动轴承故障声发射(AE)诊断方法。通过EEMD和CEEMDAN方法分别对轴承AE信号进行分解,结果表明,CEEMDAN具有较好的分解完备性和抗模态混叠性;将EEMD能量熵和CEEMDAN能量熵分别作为模式识别分类器的特征向量进行故障诊断,后者的识别准确率较高;通过与SVM、BP神经网络方法对比,DBN方法的模式识别效果更好,且表现出较好的稳定性。因此,文章所提方法能够有效的应用于超低速滚动轴承的故障诊断。
Aiming at the difficulty of fault diagnosis of ultra low speed rolling bearings,an acoustic emission(AE)diagnosis method based on CEEMDAN and Deep Belief Network(DBN)is presented.The method of EEMD and CEEMDAN is used to decompose the AE signal of bearing respectively,and the results show that CEEMDAN has better decomposition completeness and anti-aliasing property.Using EEMD Energy Entropy and CEEMDAN Energy Entropy as feature vectors of pattern recognition classifier respectively,the recognition accuracy of the latter is higher.Compared with the SVM,BP neural network method,the DBN method has better pattern recognition performance and stability.Therefore,the method proposed in this paper can be effectively applied to the fault diagnosis of ultra low speed rolling bearings.
作者
张鹏林
徐桃萍
马小东
杨天雨
ZHANG Peng-lin;XU Tao-ping;MA Xiao-dong;YANG Tian-yu(State Key Laboratory of Advanced Processing and Recycling of NonferrousMetals,Lanzhou University of Technology,Lanzhou 730050,China)
出处
《组合机床与自动化加工技术》
北大核心
2019年第9期77-80,84,共5页
Modular Machine Tool & Automatic Manufacturing Technique
