Dynamic Modeling Of Gasoline Properties Based On Near Infrared Spectroscopy

Rapid detection of gasoline properties plays significant role in the process of gasoline production.Although some methods for the rapid determination have been proposed,the prediction accuracy of gasoline properties still needs to be improved due to the increasing rigorous pressure on environment and refinery intensive production.In addition,these modeling methods have higher requirement on the uniformity of sample distribution which is frequently unavailable in the production process.A dynamic modeling method based on near infrared spectroscopy is put forward in this paper to improve the real-time detection performance of gasoline properties.The research background is first presented in this paper,followed by dynamic modeling method of near infrared spectrum in chapter 2 in which the pretreatment method of near infrared spectrum is introduced.In addition,the rapid detection principle of gasoline properties based on near infrared spectroscopy is proposed,including principal component analysis(PCA)and partial least squares(PLS).Finally,the dynamic modeling principle of gasoline properties based on near infrared spectroscopy is introduced,including the acquisition of near infrared spectrum,the dynamic selection of the optimum modeling components,etc.The third chapter introduces the prediction method of gasoline properties based on Euclidean distance similarity discrimination.The method of selecting similar samples is first introduced,and then the detailed detection procedure of gasoline properties based on Euclidean distance similarity discrimination is proposed.Two industry cases concerning 92#gasoline research octane number and Reid vapor pressure are analyzed in the end of chapter 3.The detection method of gasoline properties based on weighted absorbance is proposed in chapter 4.After the rule of similar samples selection is first presented,the steps of detection method of gasoline properties based on weighted absorbance is introduced.A detailed analysis is given about 92#gasoline research octane number and density in one refinery enterprise at last.The fifth chapter introduces the detection method of gasoline properties based on principal component analysis similarity discrimination.First,the selection method of similar samples and the detailed procedure of the detection method are proposed.Next,two examples are analyzed with regard to 92#gasoline research octane number and 10%evaporation temperature.Finally the selection principle of sample set is studied through cases analysis in the end of chapter 5.Compared with normal detection method of gasoline properties,the dynamic modeling method of gasoline properties based on near infrared spectroscopy can dynamically select calibration samples and achieve superior forecasting precision.Moreover,model maintenance is simple in this method,which has important significance to enhance the quality and efficiency of gasoline production.