近红外光谱技术识别车用汽油的应用研究

研究了不同牌号汽油的近红外光谱识别技术。研究结果表明，90号和93号汽油烃族组成的差异在近红外光谱得到充分体现，利用近红外光谱技术对二者进行识别是可行的；但识别比较困难，无法采用传统的马氏距离法进行准确识别。采用WPT-BAYES（小波包变换-贝叶斯）方法，可以大幅度提高识别精度，其正确识别率高达98％以上，满足使用要求。

Model Recognition of Gasoline by NIR Technology

This paper is focused on model recognition methods of gasoline by NIR technology.Results show that their components differences between 90 gasoline and 93 gasoline are revealed in their near infrared spectra,which demonstrates that NIR can be used to recognize them.It is difficult to classify these two kinds of gasoline;they cannot be well recognized by traditional mahalanobis distance methods.WPT-BAYES(combination of wavelet packet transform and Bayes) method can improve its recognition accuracy to 98%.