A nonlinear quality-related fault detection approach based on modified kernel partial least squares

详细信息    查看全文

• 作者：Jianfang Jiao^a ; ^{jiaojianfangheu@163.com} ; Ning Zhao^b ; ^{joyning@126.com} ; Guang Wang^a ; ^{guang.wang@hit.edu.cn} ; Shen Yin^b ; ^{shen.yin2011@googlemail.com}
• 关键词：Data-driven ; Quality-related ; Fault detection ; Kernel partial least squares ; Singular Value Decomposition ; Nonlinear monitoring
• 刊名：ISA Transactions
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：66
• 期：Complete
• 页码：275-283
• 全文大小：1551 K
• 卷排序：66

文摘

The proposed method maps original process variables space into feature space to deal with nonlinearities.. A KPLS model is used to build the linear relationship between kernel and output matrices. The kernel matrix is decomposed into two orthogonal parts by singular value decomposition. The statistics for each parts are determined appropriately for the purpose of quality-related fault detection. The proposed method has a more simple diagnosis logic and more stable performance.