递推PLS方法及其在过程监控中的应用*

作者：苏金明1，李春富1，孙如田2，侯平智1,王建中1 日期：2010-06-21/span> 浏览：3523 查看PDF文档

递推PLS方法及其在过程监控中的应用*
苏金明1，李春富1，孙如田2，侯平智1,王建中1
（1.杭州电子科技大学 自动化学院，浙江 杭州 310018; 2.克拉玛依职业技术学院，新疆 克拉玛依 833600）

摘要：针对处理批量数据时传统的偏最小二乘（PLS）模型无法在线更新的问题，提出了结合遗忘因子法的递推PLS算法，进而更新Qa控制限，有效地克服了传统PLS中Qa无法反映系统时变性的缺点。并对异丁烯酸甲酯（MMA）聚合反应过程中的数据进行了仿真分析。研究结果表明，递推PLS方法大大降低了监控过程中的误报率和漏报率，提高了监控系统的性能。
关键词：递推偏最小二乘；过程监控；间歇过程
中图分类号：TP277文献标识码：A文章编号：1001-4551(2010)03-0093-04

Recursive PLS method and its application in process monitoring

SU Jinming1, LI Chunfu1, SUN Rutian2, HOU Pingzhi1, WANG Jianzhong1
（1. School of Automation, Hangzhou Dianzi University, Hangzhou 310018, China;
2. Karamay Vacational & Technical College, Karamay 833600, China）

Abstract: Aiming at the problem that recursive partial least square(PLS) model is not updated with process changes, which may lead to increased false alarm rate. Recursive PLS with forgetting factor method was proposed. When new data was available, the PLS model and the corresponding Qa control limit were updated to accommodate the process change. A MMA polymerization batch process was studied. The simulation result demonstrates that recursive PLS method can greatly reduce the false alarm rate and omission rate in the monitoring process, and improve its performance.
Key words: recursive partial least square(PLS); process monitoring; batch processes