文献类型：期刊文献

中文题名：多模态化工过程的全局监控策略

英文题名：Global monitoring strategy for multimode chemical processes

作者：解翔[1];侍洪波[1]

机构：[1]华东理工大学化工过程先进控制和优化技术教育部重点实验室,上海200237

年份：2012

卷号：63

期号：7

起止页码：2156

中文期刊名：化工学报

外文期刊名：CIESC Journal

收录：CSTPCD;;Scopus;北大核心:【北大核心2011】；CSCD:【CSCD2011_2012】；

基金：国家自然科学基金项目(61074079);上海市重点学科项目(B504)~~

语种：中文

中文关键词：多模态过程;在线监控;高斯混合模型;邻近保留投影

外文关键词：multimode processes;online monitoring;Gaussian mixture model;locality preserving projection

摘要：引言基于数据驱动的过程监控方法从20世纪80年代建立以来得到了蓬勃的发展,理论体系逐渐完善,功能模块不断丰富。特别是最近几年,来自人工智能,机器学习及信号处理领域的各种方法的引入为该领域注入了新的活力。
Multivariate statistical process monitoring（SPM）has gained tremendous attention in both academic and industrial circles over the past two decades.Most of the existing statistical monitoring algorithms are established on the assumption that the monitored industrial processes work under single operating mode.However,in order to meet the demands of markets,multimode has become a significant feature of modern chemical industry.To meet the monitoring demands of multimode processes,a global monitoring strategy based on locality preserving projection（LPP）and Gaussian mixture model（GMM）is proposed.By integrating probability indices of each operating modes,the online monitoring is guaranteed to be continuous and uninterrupted.The efficiency and effectiveness of the novel monitoring strategy is verified though the Tennessee Eastman（TE）simulation platform.