Quantitative analysis of reaction gases or exhaust using an online process mass spectrometer

Abstract

Online quantitative analysis of reaction gases or exhaust in industrial production is of great significance to improve the production capacity and process. A novel method is developed for the online quantitative analysis of reaction gases or exhaust using quantitative mathematical models combined with the linear regression algorithm of machine learning. After accurately estimating the component gases and their contents in the reaction gases or exhaust, a ratio matrix is constructed to separate the relevant overlapping peaks. The ratio andcalibrationstandardgasesaredetected,filtered,normalized,andlinearlyregressedwithanonlineprocess massspectrometertocorrecttheratiomatricesandobtaintherelativesensitivitymatrices.Aquantitative mathematicalmodelcanbeestablishedtoobtainthecontentofeachcomponentofthereactiongasesor exhaustinrealtime.Themaximumquantificationerrorandrelativestandarddeviationofthemethodare within0.3%and1%,afteronlinequantificationoftherepresentativeyeastfermentertailgas.