摘要
结合谐波小波包和相关向量机设计了滚动轴承故障诊断方法,以实现轴承正常状态、内圈故障、滚动体故障及外圈故障状态的诊断。首先,利用谐波小波包对轴承的振动信号进行多层分解,根据各频段的小波分解系数计算各个频带能量,归一化之得到特征向量;其次,对传统的OAO-RVM模型进行简化,改进为新的OAORVM多模式分类模型;最后,利用滚动轴承试验台的振动数据对设计方法进行了验证。结果表明,设计的诊断方法在识别的准确率及算法计算效率方面均比传统的支持向量机诊断方法好。
A fault diagnosis method for rolling bearings is designed based on harmonic wavelet package and Relevance Vector Machine (RVM), the diagnosis are realized for normal state, inner ring fault, rolling element fault and outer ring fault of bearings. Firstly, the multiple - level decomposition for vibration signals of bearings is carried out by using harmonic wavelet package. The energy in various frequency bands is computed according to wavelet decomposition coef- ficients for every frequency band, and the feature vector is obtained after normalization. Secondly, the traditional OAO - RVM model is simplified and improved to new OAO - RVM multi - mode classification model. Finally, the design method is verified by using vibration data from test rig for rolling bearings. The results show that the identification accu- racy and algorithm computational efficiency of designed diagnosis method are better than those of traditional support vec- tor machine diagnosis method.
出处
《轴承》
北大核心
2015年第8期51-55,共5页
Bearing
基金
航空科学基金项目(2012ZD54013)
关键词
滚动轴承
故障诊断
谐波小波包
RVM
rolling bearing
fault diagnosis
harmonic wavelet package
relevance vector machine