引用本文:刘金平,王杰,刘先锋,等.基于递归稀疏主成分分析的工业过程在线故障监测和诊断[J].控制与决策,2020,35(8):2006-2012
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览次   下载 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于递归稀疏主成分分析的工业过程在线故障监测和诊断
刘金平1, 王杰1, 刘先锋1, 唐朝晖3, 马天雨1,2, 肖文辉3
(1. 湖南师范大学智能计算与语言信息处理湖南省重点实验室,长沙410081;2. 湖南师范大学物理与电子学院,长沙410081;3. 中南大学自动化学院,长沙410083)
摘要:
提出一种基于递归稀疏主成分分析(recursive sparse principal component analysis,RSPCA)的工业过程故障监测与诊断方法,可用于时变工业过程的自适应故障监测与诊断.通过引入弹性回归网,将主成分问题转化为Lasso与Ridge结合的凸优化问题,采用秩-1矩阵修正对协方差矩阵进行递归分解,递归更新稀疏载荷矩阵和监测统计量的过程控制限,以实现连续工业过程长时间自适应故障监测,对检测出来的故障通过贡献图法实现对故障的诊断.在田纳西-伊斯曼(TE)过程进行实验验证,结果表明,与传统的故障监测方法相比,所提出的方法有效降低了故障漏检率和误报率,且时间复杂度低,确保了故障监测的灵敏度和实时性.
关键词:  递归稀疏主成分分析  工业过程故障监测  弹性回归网  田纳西-伊斯曼过程
DOI:10.13195/j.kzyjc.2018.1661
分类号:TP273
基金项目:国家自然科学基金项目(61971188,U1701261,61771492);湖南省自然科学基金项目(2018JJ3349);图像信息处理与智能控制教育部重点实验室(华中科技大学)开放基金项目(IPIC2017-03);湖南省研究生科研创新项目(CX2018B312).
Online fault monitoring and diagnosis using recursive sparse principal component analysis
LIU Jin-ping1,WANG Jie1,LIU Xian-feng1,TANG Zhao-hui3,MA Tian-yu1,2,XIAO Wen-hui3
(1. Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing,Hunan Normal University,Changsha 410081,China;2. School of Physics and Electronics,Hunan Normal University,Changsha 410081,China;3. School of Automation,Central South University,Changsha 410083,China)
Abstract:
A fault monitoring and diagnosis method for industrial processes based on recursive sparse principal component analysis (RSPCA) is proposed, which can be used for adaptive fault monitoring and diagnosis of time-varying industrial processes. By introducing the elastic regression net, the principal component analysis problem is changed to a convex optimization problem combining Lasso and Ridge regression. Successively, the covariance matrix is decomposed recursively by rank -1 matrix correction, which leads to the recursive updating of the sparse load matrix and the process control limit of monitoring statistics to realize the long-time adaptive fault monitoring of continuous industrial processes. With regard to the detected faults, the fault diagnosis is realized by using the contribution graph method. Experiments in Tennessee-Eastman (TE) process show that the proposed method effectively reduces the rate of failure miss detection and false alarm with low time complexity compared with traditional fault monitoring methods, which can ensures the sensitivity and real-time performance of the fault monitoring.
Key words:  recursive sparse principal component analysis  industrial process fault diagnosis  elastic regression network  Tennessee-Eastman process

用微信扫一扫

用微信扫一扫