Gas Chromatography is applied to determine the composition of gasoline which is divided into 42 groups. Partial least squares quantitative methods are used to establish the regression models between experimental research octane number(RON),90% distillation, saturated vapor pressure and those groups of variables,which are optimized by eliminating abnormal samples,preprocessing data,adjusting the participation model along with the model forecast precision. It proves to meet the needs of refining and gasoline blending analysis.%采用气相色谱法测定汽油组成含量,并将汽油组成分为42组,使用偏最小二乘定量方法分别建立实测研究法辛烷值(RON)、馏程90%、饱和蒸汽压与42组变量之间的回归模型,并采用剔除异常样本、使用数据预处理、调整参与拟合变量这些方法优化模型参数,使模型的预测精度得到提高。该方法预测结果准确,满足工厂炼油加工和汽油调合分析的需要。
展开▼