标题:An integrated method for condition monitoring of rotating machines
作者:Chen G.; Lu G.; Yan P.
作者机构:[Chen, G] Key Laboratory of High Efficiency and Clean Mechanical Manufacturing of MOE, National Demonstration Center for Experimental Mechanical Engin 更多
来源:Proceedings - 2018 Prognostics and System Health Management Conference, PHM-Chongqing 2018
出版年:2019
页码:144-149
DOI:10.1109/PHM-Chongqing.2018.00031
关键词:ARIMA model; Condition monitoring; One-class SVM; Rotating machine
摘要:The task of this paper is to detect the structural changes in machine status, such that those abnormal and unexpected running behaviors for the considered machine can be found and localized during its successive operations. To accomplish a task of this kind, we propose a new method by integration of the autoregressive integrated moving average model (ARIMA) and one-class SVM. The ARIMA model has been introduced and adopted for this problem in a previous work, where parameters in the model are somewhat difficult to estimate. In this paper, we propose to use the one-class SVM for the ARIMA model identification. This replacement can take advantage of machine learning method to learn a more reliable and more accurate model; meanwhile, it uses an unsupervised manner to detect the potential change which makes it different from other existing integrated methods that require massive samples to accomplish a prior model training. We conducted experiments based on an experimental setup. Results along with comparisons with representative methods demonstrated the effectiveness and propriety of the proposed method in real engineering applications. © 2018 IEEE.
收录类别:SCOPUS
资源类型:期刊论文
原文链接:https://www.scopus.com/inward/record.uri?eid=2-s2.0-85061788471&doi=10.1109%2fPHM-Chongqing.2018.00031&partnerID=40&md5=8d422578d0d30481b8402c1e9b8f2332
TOP