基于GC-MS的喷气燃料组成-性质定量关系研究

GC-MS-Based Study on Quantitative Composition-Property Relationships of Jet Fuels

利用气相色谱-质谱(GC-MS)联用技术对67个喷气燃料进行分子水平表征,并按9类烃族组成(正构烷烃、异构烷烃、单环烷烃、多环烷烃、烷基苯、茚满及四氢萘、萘类及茚类、苊及苊烯类、苯并噻吩及二苯并噻吩)和13类碳数(C_(7)~C _(19))分类,形成一系列组成矩阵。测定了7类理化性质,包括密度、热值、闪点、黏度、烟点、苯胺点和馏程,定义了各碳数下不同烃族的代表性化合物和平均性质(密度、热值、闪点和黏度)。采用多元线性回归(MLR)、偏最小二乘(PLS)和修正加权平均(MWA)3种算法建立了喷气燃料组成-性质定量模型。综合比较,组成与密度、热值、闪点、烟点和苯胺点的定量模型均基于PLS预测效果更优,对应的均方根误差(RMSE)分别为0.0008 g/cm^(3)、0.0302 MJ/kg、1.3418℃、6.01×10^(-5 )mm、0.7293℃;组成与黏度的定量模型基于MWA时预测效果更优,其RMSE为0.0333 mm^(2)/s。

Sixty-seven jet fuels were characterized at the molecular level by gas chromatograph-mass spectrometer(GC-MS).And they were classified into 9 hydrocarbon compounds(n-paraffins,iso-paraffins,monocycloparaffins,polycycloparaffins,alkylbenzenes,indanes and tetralines,naphthalenes and indenes,acenaphthenes and acenaphthylenes,thianaphthene and dibenzothiophene)and 13 carbon numbers(C^(7)—C^(19)),thus forming a series of composition matrices.Moreover,seven kinds of physicochemical properties including density,net heat,flash point,viscosity,smoke point,aniline point and boiling range were measured.The representative compounds and the corresponding properties(density,net heat,flash point and viscosity)of different hydrocarbon compounds under each carbon number were defined.Multiple linear regression(MLR),partial least square(PLS),and modified weighted average(MWA)were used to establish the quantitative model of composition and properties of jet fuels.By comprehensive comparison,the quantitative models based on PLS had achieved better performance in predicting density,net heat,flash point,smoke point and aniline point,corresponding to the RMSE of 0.0008 g/cm 3,0.0302 MJ/kg,1.3418℃,6.01×10^(-5)mm,and 0.7293℃;the quantitative models based on MWA had better performance in predicting composition and viscosity,corresponding to the RMSE of 0.0333 mm^(2)/s.