基于特征融合技术的滚动轴承退化预测方法研究  

Research on Rolling Bearing Degradation Prediction Method Based on Feature Fusion Technology

作　　者：于忠斌[1] 张林[2,3] 李硕 湛力[2] 刘杰[2] 周苏婷[2] 唐樟春 YU Zhong-bin;ZHANG Lin;LI Shuo;ZHAN Li;LIU Jie;ZHOU Su-ting;TANG Zhang-chun(Naval Armaments Division,Beijing 100000,China;Science and Technology on Reactor System Design Technology Laboratory,Nuclear Power Instituteof China,Chengdu 610213,Sichuan,China;School of Mechanical and Electrical Engineering,University of Electronic Science and Technology of China,Chengdu 611731,Sichuan,China)

机构地区：[1]海军装备部,北京100000 [2]中国核动力研究设计院核反应堆动力设备与可靠性研究室,四川成都610213 [3]电子科技大学机械与电气工程学院,四川成都611731

出　　处：《阀门》2021年第6期323-328,共6页VALVE MAGAZINE

基　　金：国家自然科学基金资助项目(批准号:51875087);广东省基础与应用基础研究基金资助项目(2019A1515011708)。

摘　　要：提出了一种结合多维特征融合及最小二乘支持向量回归机的滚动轴承退化趋势预测方法。首先提取了滚动轴承全寿命周期振动数据的各域特征,并利用主成分分析法将各域多维特征融合为表征轴承退化性能趋势的退化性能指标。将最小二乘支持向量回归机作为预测模型,同时利用粒子群算法优化模型参数。其次,利用误差累积和方法对预测模型的累积预测误差进行有效地跟踪和控制,进一步提高预测模型的准确性和稳定性。最后,利用实测的滚动轴承全寿命实验数据对所提方法开展验证,并将所得结果与几种现有方法结果进行对比,结果表明该方法能获得较好的滚动轴承退化趋势预测结果。Based on multi-feature fusion and least square support vector regression,this paper proposes a novel prediction method for rolling bearing degradation trend.Firstly,the features of various domains of rolling bearing vibration data were extracted,and the multi-dimensional features of each domain were fused to represent the degradation performance trend of bearing by principal component analysis.The least square support vector regression was used as the prediction model,and the particle swarm optimization algorithm was used to optimize the model parameters.Secondly,the cumulative error of the prediction model is effectively tracked and controlled by the error accumulation method,which further improves the accuracy and stability of the prediction model.Finally,the proposed method is verified by using the measured life test data of rolling bearings,and the results are compared with those of several existing methods.The results show that this method can obtain better prediction results of rolling bearing degradation trend.

关 键 词：滚动轴承 退化趋势预测 多维特征融合 主成分分析 最小二乘支持向量回归机 误差累积和 

分 类 号：TH134[机械工程—机械制造及自动化]