摘要
现代过程工业正逐渐倚重于生产小批量、多品种、高附加值产品的间歇过程.基于多元统计模型的过程监测是保障生产安全和产品质量的重要工具.从间歇过程独特的数据特性出发,将现有的多元统计建模方法进行合理的分类,简要回顾了各类方法的起源、发展及延伸的历程.除了阐述每种方法的基本原理,还详细讨论了各种方法的适用背景,相互关联及优缺点等内容,并对这一领域中依然存在的问题以及研究前景给出中肯的评述.
The focus of modern process industry has shifted to the production of highervalue-added products through batch processes. Statistical process monitoring (SPM) has shown to be effective in ensuring process safety and product consistency for batch processes. This paper presents a review of multivariate statistical modeling and monitoring for batch process applications. Based on analysis of the nature of batch processes, this paper reviews each key method in terms of its motivation, development, and application prospective. The review ends with the authors' personal views of challenges and future directions for the area.
出处
《自动化学报》
EI
CSCD
北大核心
2006年第3期400-410,共11页
Acta Automatica Sinica
基金
国家自然科学基金(60374003)
973计划子课题(2002CB312200)资助
关键词
间歇过程
多元统计模型
过程监测
主成分分析
偏最小二乘
三线性分解模型
Batch processes, multivariate statistical model, process monitoring, principal component analysis (PCA), partial least squares (PLS), trilinear decomposition model