基于CAE与混合灰狼优化SVR的滚动轴承性能退化趋势预测  

Degradation Trend Prediction for Rolling Bearing Performance Based on CAE and Hybrid Grey Wolf Optimization SVR

作　　者：李卓漫 王海瑞[1] 于童 LI Zhuo-man;WANG Hai-rui;YU Tong(Faculty of Information Engineering and Automation,Kunming University of Science and Technology)

机构地区：[1]昆明理工大学信息工程与自动化学院

出　　处：《化工自动化及仪表》2021年第6期620-624,共5页Control and Instruments in Chemical Industry

基　　金：国家自然科学基金项目(61263023,61863016)。

摘　　要：为了提高滚动轴承性能退化趋势预测的准确度,提出基于卷积自编码器和混合灰狼优化支持向量回归机的性能退化趋势预测方法。首先提取轴承振动信号的时域频域特征建立高维特征向量;然后通过卷积自编码器对高维特征降维,构建退化特征指标,以反映轴承性能退化趋势;最后通过引入混合灰狼算法优化支持向量回归机中的参数,构建趋势预测模型。通过与其他模型的试验分析对比,证实所提模型可以更加准确高效地预测滚动轴承性能退化趋势,并且具有普遍适用性。For improving the prediction accuracy of the property degradation trend of rolling bearings,a model of property degradation trend prediction which based on convolutional auto-encoder and hybrid grey wolf optimization for the support vector regression machine was proposed.Firstly,having the time and frequency domain features of the bearing vibration signal extracted to establish a high-dimensional feature vector,and then having convolution auto-encoder used to reduce the feature dimension and to construct the degradation feature index so as to reflect the bearing property degradation trend;and finally,having the hybrid grey wolf optimization introduced to optimize the parameters of support vector regression machine and construct the trend prediction model.Comparing this model with other ones can predict the property degradation trend of rolling bearing more and it has universal applicability.

关 键 词：退化趋势预测 滚动轴承 卷积自编码器 混合灰狼优化算法 支持向量回归机 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置] TP399[自动化与计算机技术—控制科学与工程]