Mechanism Model Development For Diesel Oil Properties Based On Near-Infrared Spectroscopy

As a fast, safe and environment-friendly analysis technology, near infrared spectrum analysis technology has been widely applied in petroleum and chemical industry with sound prospective. Since the former laboratory analysis and measurement method of diesel oil product is time-consuming , costly,and needs more input of labor and fund, it can not meet the demand of real-time control in modern industry. In this article, near-infrared spectrum analysis technology is utilized to predict nature of diesel oil. On the basis of analyzing constituent of diesel oil, models are created by selecting component groups and structural parameters of diesel oil products.Near-infrared spectrum analysis technology is uniquely superior in measuring structure groups of petroleum and petrochemical products. Based on absorption of C-H groups of near-infrared spectrum coverage, two variables called attribute structure branched chain and alkenes are constructed and introduced into models together with absorbance data of structural groups corresponded to characteristics. Besides, linear and nonlinear models are created to predict natures of diesel oil, and L-M algorithm is used to predict parameters of nonlinear models. The models have clear characteristics of component groups and structural in diesel oil, and comparing to the former models of Chemical Measurement, more information is contained in the models, so it can provide real-time and optimal information for production process. The models this paper built is compared with the former MLR and PLS models in terms of predicting effect. Since natures of diesel oil including viscosity, density, flash point, etc. are all final manifestation of co-functions by various components of diesel oil, thus are closely related with components structure of oil product itself,the results of the simulation experiment show that the model manifesting structure characteristic variables has higher predicting precision and compatibility compared with other models.