基于PLS方法的催化重整装置氧含量软测量模型_英文_.docx

? 1622?系统仿真学报Vol. 15 No.11JOURNAL OF SYSTEM SIMULATIONNov. 2003Soft-Sensing Model of Oxygen Concentration inCatalyticReformer Based on PLSAlgorithmSUN Da-zhong, YE Fang-jun, GUXing-sheng(ResearchInstituteofAutomation,EastChinaUniversityofScienceTechnology, Shanghai200237,China)Abstract:Moreandmoreresearchhasbeenconductedinthefieldofsoft-sensingmodeling.Softsensorsestimatehard-to-measureprocessvariablesfromothereasy-to-measurevariables.Inthispaper,thePLSmethodhasbeenusedtobuildasoft-sensingmodelforoxygenconcentrationinacatalyticreformer.Theresults show that the models are highly accurate.Keywords: Partial Least Square (PLS); soft sensor ; soft-sensing model; oxygen concentration基于PLS方法的催化重整装置氧含量软测量模型孙大中，叶方军，顾幸生（华东理工大学自动化研究所，上海200237）摘要：当前，软测量建模受到了越来越多的重视，软仪表能够通过易测变量来估计出难测变量。本文应用PLS方法建立了催化重整装置氧含量的软测量模型，结果表明该模型具有很高的精度。关键词：部分最小二乘；软仪表；软测量模型；氧含量文章编号：1004-731X(2003)11-1622-03中图分类号：TP391.9文献标识码：AIntroductionIn the chemical process industries，it is often difficulttoestimate some important process variables due to thelimitationof process technology or measurement techniques. In thiscase,soft-sensing techniques can be used to obtain a modelbetweeneasily obtained measurements (secondary variables) andthoseimportant process variables (primary variables).Generallyspeaking, there are three kinds of approaches toestablishsoft-sensing models in processes: mechanisticmodeling,statistical regression modeling and artificialintelligentmodeling. As many chemical processes involve a largenumberof species or unknown reaction mechanisms, building areliablemechanistic model is time-consuming and sometimesisimpossible. Therefore, establishing a reliable model basedonsome statistical techniques or artificial intelligenttechniqueshas always been verypopular.Intheprocessofcontinuouscatalyticreforming(CCR),oxygenconcentrationisanimportantvariable.Inthisarticle,the objective is to build a soft-sensing model foroxygenconcentration so that it can be estimated usingothereasy-to-measure process