| 基于互信息的主成分分析结合支持向量回归的滚动轴承剩余寿命预测研究 |
| |
| 引用本文： | 欧白羽,杨勐,韩旭. 基于互信息的主成分分析结合支持向量回归的滚动轴承剩余寿命预测研究[J]. 北京化工大学学报(自然科学版), 2021, 48(6): 108-117. DOI: 10.13543/j.bhxbzr.2021.06.014 |
| |
| 作者姓名： | 欧白羽 杨勐 韩旭 |
| |
| 作者单位： | 1. 陆军航空兵学院 陆军航空兵研究所, 北京 101121;2. 陆军航空兵学院 航空机械工程系, 北京 101123;3. 93146部队, 北京 100000 |
| |
| 摘 要： | 滚动轴承作为旋转机械设备中的关键部件，影响着设备的可靠性运行。为了智能开展设备维护工作，提高设备的运转效率，提出一种基于互信息（mutual information，MI）的主成分分析（principal component analysis，PCA）（MI-PCA）结合支持向量回归（support vector regression，SVR）的滚动轴承剩余寿命预测方法。首先利用小波包降噪算法剔除原始振动信号中的异常数据点和噪声，并基于降噪数据提取其时域、频域和时频域特征；然后结合特征与剩余寿命的互信息值进行特征筛选，再通过PCA降维算法获得可表征轴承退化状态的敏感特征，用于SVR的输入；最后构建并训练SVR剩余寿命预测模型，并将其应用于滚动轴承全寿命试验数据。试验结果表明与基于MI和基于PCA的SVR回归预测模型（MI-SVR模型、PCA-SVR模型）相比，基于MI-PCA的SVR模型具有更高的预测精度（预测精度可达97%），能够实现滚动轴承剩余寿命的精准预测，为开展及时有效的设备维护工作提供了决策依据。
|
| 关 键 词： | 滚动轴承 剩余寿命预测 支持向量回归 主成分分析 互信息 |
| 收稿时间： | 2021-04-29 |
Remaining useful life prediction of rolling bearings based on mutual information based principal component analysis combined with support vector regression |
| |
| OU BaiYu,YANG Meng,HAN Xu. Remaining useful life prediction of rolling bearings based on mutual information based principal component analysis combined with support vector regression[J]. Journal of Beijing University of Chemical Technology, 2021, 48(6): 108-117. DOI: 10.13543/j.bhxbzr.2021.06.014 |
| |
| Authors: | OU BaiYu YANG Meng HAN Xu |
| |
| Affiliation: | 1. Army Aviation Research Institute of Army Aviation Institute, Beijing 101121;2. Department of Aviation Machinery Engineering of Army Aviation Institute, Beijing 101123;3. 93146 Troops, Beijing 100000, China |
| |
| Abstract: | As a key component in rotating machinery, rolling bearings affect its reliable operation. In order to carry out equipment maintenance intelligently and improve the operation efficiency of the equipment, a remaining useful life prediction method for rolling bearings using mutual information (MI) based principal component analysis (PCA) (MI-PCA) combined with support vector regression (SVR) is proposed. First, the wavelet packet denoising algorithm is used to eliminate the abnormal data points and noise in the original vibration signal, and the time-domain, frequency-domain and time-frequency-domain features are then extracted based on the denoising data. The feature selection is subsequently conducted according to the mutual information between the features and the remaining useful life, and the sensitive features representing the bearing degradation state are obtained using the PCA algorithm, which is used as the input of the SVR. Finally, the remaining useful life prediction model of the whole life test data of rolling bearings based on the SVR is established and trained. Compared with the SVR regression prediction models based on MI and PCA (MI-SVR model, PCA-SVR model), the test results show that the SVR model based on MI-PCA has higher accuracy in predicting the remaining useful life of rolling bearings. The prediction accuracy can reach 97%, which provides a basis for carrying out timely and effective equipment maintenance work. |
| |
| Keywords: | rolling bearings remaining useful life prediction support vector regression principal component analysis mutual information |
|
| 点击此处可从《北京化工大学学报(自然科学版)》浏览原始摘要信息 |
|
点击此处可从《北京化工大学学报(自然科学版)》下载全文 |
