石油炼制与化工 ›› 2022, Vol. 53 ›› Issue (1): 86-92.

• 分析与评定 • 上一篇    下一篇

基于近红外光谱快速预测石脑油单体烃分子组成

刘秋芳,褚小立,陈瀑,李敬岩   

  1. 中国石化石油化工科学研究院
  • 收稿日期:2021-04-14 修回日期:2021-09-04 出版日期:2022-01-12 发布日期:2021-12-24
  • 通讯作者: 褚小立 E-mail:chuxl.ripp@sinopec.com
  • 基金资助:
    国家重点研发计划资助

RAPID PREDICTION OF HYDROCARBON MOLECULAR COMPOSITION OF NAPHTHA BASED ON NEAR INFRARED SPECTROSCOPY

  • Received:2021-04-14 Revised:2021-09-04 Online:2022-01-12 Published:2021-12-24
  • Contact: Chu Xiaoli E-mail:chuxl.ripp@sinopec.com

摘要: 提供了一种快速测定石脑油分子水平组成的方法,以石脑油的族组成(PINA)和单体烃组成的实验室气相色谱分析数据样本为基础,建立了石脑油的单体烃分布比例库;采用样本增强(DA)法生成大量虚拟样本,以石脑油近红外光谱(NIR)作为输入变量,结合偏最小二乘法(PLS)建立了石脑油PINA组成预测模型,利用K-近邻回归法(KNR)建立了石脑油单体烃分布比例预测模型。预测结果表明,在数据样本范围内,利用所建模型可以快速测定石脑油的单体烃分子组成。

关键词: 石脑油, 详细组成, 虚拟样本, 近红外光谱

Abstract: A rapid method for the determination of molecular level composition of naphtha was provided. Based on the data samples of group composition (PINA) and single hydrocarbon composition of naphtha by gas chromatography in laboratory, the distribution ratio library of single hydrocarbon in naphtha was established. A large number of virtual samples were generated by sample enhanced (DA) method. A prediction model of naphtha PINA composition was established by using near infrared spectroscopy (NIR) as input variables and partial least squares (PLS) method. A prediction model of naphtha mono-hydrocarbon distribution ratio was established by K-nearest neighbor regression (KNR) method. The prediction results show that within the range of experimental data samples, the model can be used to quickly determine the monomer hydrocarbon molecular composition of naphtha.

Key words: naphtha, detailed composition, virtual sample, near infrared spectroscopy