| Cetane number is an important index reflecting burning performance of diesel. There are already many methods for ctetane number prediction in nowadays. NIR technology, as one of the methods for predicting fuel properties, has the advantage of quick, non-pollution and safety. This paper develops a mechanism model which is used to predict CN of diesel fuel based on NIR spectroscopy technology.The paper conducts a detailed analysis specific to CN property, research a series of C-H groups related to CN property during hydrocarbons components of diesel fuel, and also study the affect C-H groups caused to CN property. Meanwhile, this paper extracts the NIR absorb information of related groups by using the methods of second differential and two-dimension spectroscopy, bring up a mechanism model based on cetane number and NIR information groups related. This model is a nonlinear model and estimate parameters of the model by L-M algorithm. a spectroscopy and properties data of diesel fuel are choosed as sample data. Two models are built to predict cetane number, one is PLS model and the other is mechanism model. The mechanism model selects 14 points spectroscopy data as input. The simulation experiments shows that the mechanism model gains higher predictive accuracy, because parameters in mechanism model are all related with CN property, so redundant information is removed. What's more, mechanism model needn't to conduct data preprocessing which simplify the procedure of building a model.
|
PDF Full Text Request