摘要
提出了一种基于主元分析和聚类分析的机械传动系统磨损定位与状态识别方法,通过对数据空间的非线性映射和模糊集合理论实现了主元分析的非线性和鲁棒性改进,实现了同时考虑数据非线性与离群值的磨损定位与状态识别。此外,提出了一种基于主元聚类的数据非线性判别方法。最后,通过综合传动装置的全寿命周期油液光谱数据验证了本文方法优于不考虑数据非线性和离群值的方法,有效降低了数据非线性与离群值对磨损定位与状态识别的影响,提高了磨损定位与状态识别的准确性。
A new wear localization and state identification method of mechanical transmission system is proposed based on improved Principal Component Analysis(PCA)and cluster analysis,which considers the data nonlinearity and outliers.The nonlinear and robust improvement of the PCA is realized by nonlinear mapping and fuzzy set theory.Moreover,a nonlinear data discrimination method is proposed based on principal component clustering.Finally,a case study for power shift steering transmission is conducted.The results demonstrate that the proposed method is superior to the method without considering the nonlinear data and outliers,and it can effectively improve the accuracy of wear localization and state identification.
作者
闫书法
马彪
郑长松
陈建文
李慧珠
YAN Shu-fa;MA Biao;ZHENG Chang-song;CHEN Jian-wen;LI Hui-zhu(School of Mechanical Engineering,Beijing Institute of Technology,Beijing 100081,China;Norinco Group Jianglu Machinery and Electronics Group Com pany,Xiangtan 411100,China)
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2019年第2期359-365,共7页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金项目(51475044)
关键词
车辆工程
磨损定位
状态识别
主元分析
聚类分析
机械传动系统
vehicle engineering
wear location
state identification
principal components analysis
cluster analysis
mechanical transmission system
