Near-infrared Spectroscopy Study Of Application Of The Rapid Determination Of The Gasoline Nature

With the new standard of clean gasoline being performed in china, further control on the content of harmful substances such as benzene, aromatic hydrocarbon and alkene has become more strictly. At the same time, gasoline, being one of the most utilized fuels, has to meet higher requirements on some properties such as antiknock. To obtain enough qualified clean gasoline there give rise to the needs for fast and accurate determination of gasoline product.Being regarded as a fast determination tool, near-infrared spectroscopy (NCR) technology has been applied broadly in the analytic field of many other petrochemical products.In this paper Flourier near-infrared spectroscopy is applied in the study on fast determination of group content and antiknock property of gasoline through establishing calibration models.First, the content of benzene, aromatic hydrocarbon and alkene, and the octane number (RON and MON) of gasoline are determined to obtain the reference data with national standard analytic methods.Secondly, the near infrared spectrums of gasoline samples are measured with FT near-infrared spectrometer. With the reference data determined by standard analytic methods, calibrations models were established respectively including benzene content, aromatic hydrocarbon content, alkene content, and octane number (RON and MON). With partial least square algorithm provided by OPUS software package, conditions in calibration model construction such as pretreatment procedures of spectrums, the choice of spectrum range, and the number of ranks of the main component etc. are defined through optimization. Fifty one models for the above five mentioned properties with different parameters are established respectively. Finally, the above optimized quantitative calibration models are verified through accuracy tests. Further repeatability tests on the decided calibration models of aromatic hydrocarbon content and alkene content are conducted.The results indicate that the calibration models can be used to fast determine the group content and relevant properties of gasoline samples with strong relevance between prediction data and reference data obtained by standard analytic methods and well satisfaction to the requirements of accuracy and repeatability in standard analytic methods.