多元统计过程监控与安全生产  

Multivariate statistical process monitoring and safe operating

作　　者：徐用懋[1] 赵仕健[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《计算机与应用化学》2007年第4期475-478,共4页Computers and Applied Chemistry

摘　　要：统计过程控制是一种改善产品质量及保证安全生产的有力工具。针对现有多元统计监控技术大多假定所考察的生产过程本身仅存在一个标准运行条件,导致实际应用时往往引发大量的连续报警的问题,本文基于主角度建立了任意两个主元模型相似性的度量,提出了一种基于多主元模型的过程监控方法。通过该方法能有效地检测、诊断工业过程中的异常,以避免事故的发生,将带来巨大的经济效益。最后,讨论了相应的软件实现平台EZMon及其应用。Muhivairate statistical process monitoring （MSPM） is one powerful tool for product quality improving and plant operation safety and is of significant valuable provided it can detect and diagnose the process malfunction. Considering that most existing MSPM techniques assume that there is one nominal operating condition only and trigger numerous continuous warnings when applied in real ap- plications, one process monitoring methodology based on the proposed similarity index of any two principal components models through principal angles is proposed. The corresponding software platform, EZMon, and its applications is also addressed.

关 键 词：多元统计过程监控 多操作模式 多主元模型 EZMon 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]