基于LCD和GMM的滚动轴承寿命预测方法  

The life forecasting method for rolling bearing based on LCD and GMM

作　　者：刘吉彪[1] 程军圣[2] 刘燕飞[2] 

机构地区：[1]河南工业职业技术学院机械工程系,南阳473000 [2]湖南大学机械与运载工程学院,长沙410082

出　　处：《现代制造工程》2016年第7期120-124,共5页Modern Manufacturing Engineering

基　　金：国家自然科学基金项目(51075131)

摘　　要：滚动轴承寿命预测的关键在于振动信号特征提取和模式识别,提出基于局部特征尺度分解(LCD)和高斯混合模型(GMM)的滚动轴承寿命预测方法。首先在对滚动轴承全寿命数据进行LCD分解后,提取各个分量的特征值并加入时间因子重构特征向量;然后利用GMM对全寿命数据的特征向量进行聚类,将全寿命数据在时域上分成若干个退化状态;最后将不同退化状态下的数据作为训练样本对神经网络进行训练,并采用训练好的神经网络对滚动轴承寿命进行预测。实验数据的分析结果表明,将LCD、GMM和径向基神经网络相结合可以有效地实现滚动轴承寿命预测。The key of the life forecasting for roller bearing is vibration signal feature extraction and pattern recognition. The life forecasting for roller bearing based on Local Characteristic-scale Decomposition （ LCD ） and Gaussain Mixture Model （ GMM ） is proposed. Firstly, after the whole life data of rolling bearing is decomposed by LCD method, the feature values of the components are extracted and the feature vector is reconstructed by using the time factor;then the feature vectors are clustered in time domain by using GMM and the whole life data is divided into some degenerate states in time domain;finally, the different degenerate state data is served as training samples for neural network training and the life of roller bearing is forecasted by using the trained neural network. The experiment results show that the combination of LCD, GMM and neural network can be effectively applied to the life forecasting for rolling bearing.

关 键 词：局部特征尺度分解 高斯混合模型 神经网络 时间因子 滚动轴承 寿命预测 

分 类 号：TN911.7[电子电信—通信与信息系统] TH165[电子电信—信息与通信工程]