基于LCD和GMM的滚动轴承寿命预测方法
发布时间：2019-07-01 20:02
【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。
[Abstract]:The key of rolling bearing life prediction lies in vibration signal feature extraction and pattern recognition. A rolling bearing life prediction method based on local feature scale decomposition (LCD) and Gaussian mixture model (GMM) is proposed. Firstly, after LCD decomposition of the whole life data of rolling bearings, the eigenvalues of each component are extracted and the time factor is added to reconstruct the eigenvector, and then the eigenvector of the whole life data is clustered by GMM, and the whole life data is divided into several degraded states in time domain. Finally, the data under different degradation states are used as training samples to train the neural network, and the trained neural network is used to predict the life of rolling bearings. The analysis results of experimental data show that the combination of LCD,GMM and radial basis function neural network can effectively realize the life prediction of rolling bearings.
【作者单位】： 河南工业职业技术学院机械工程系;湖南大学机械与运载工程学院;
【基金】：国家自然科学基金项目(51075131)
【分类号】：TH133.33
，
本文编号：2508757
[Abstract]:The key of rolling bearing life prediction lies in vibration signal feature extraction and pattern recognition. A rolling bearing life prediction method based on local feature scale decomposition (LCD) and Gaussian mixture model (GMM) is proposed. Firstly, after LCD decomposition of the whole life data of rolling bearings, the eigenvalues of each component are extracted and the time factor is added to reconstruct the eigenvector, and then the eigenvector of the whole life data is clustered by GMM, and the whole life data is divided into several degraded states in time domain. Finally, the data under different degradation states are used as training samples to train the neural network, and the trained neural network is used to predict the life of rolling bearings. The analysis results of experimental data show that the combination of LCD,GMM and radial basis function neural network can effectively realize the life prediction of rolling bearings.
【作者单位】： 河南工业职业技术学院机械工程系;湖南大学机械与运载工程学院;
【基金】：国家自然科学基金项目(51075131)
【分类号】：TH133.33
，
本文编号：2508757
本文链接：https://www.wllwen.com/kejilunwen/jixiegongcheng/2508757.html
