基于图卷积神经网络汽油单体烃辛烷值的预测

石油炼制与化工 ›› 2021, Vol. 52 ›› Issue (7): 82-87.

基于图卷积神经网络汽油单体烃辛烷值的预测

崔晨¹,何杉¹,吕文进¹,张霖宙²,周祥¹

1. 中国石化石油化工科学研究院
2. 中国石油大学（北京）重质油国家重点实验室

收稿日期:2020-12-07 修回日期:2021-03-11 出版日期:2021-07-12 发布日期:2021-06-30
通讯作者: 崔晨 E-mail:cuichen.ripp@sinopec.com
基金资助:
国家重点研发计划

GRAPH CONVOLUTION NEURAL NETWORK FOR PREDICTING THE OCTANE NUMBER OF PURE HYDROCARBONS IN GASOLINE

Received:2020-12-07 Revised:2021-03-11 Online:2021-07-12 Published:2021-06-30

摘要/Abstract

摘要： 基于图卷积神经网络的神经指纹方法，引入了池化操作，建立了改进的神经指纹方法；进而采用改进的神经指纹法建立了汽油单体烃辛烷值的预测模型，作为对分子级汽油辛烷值调合模型的支撑。通过用单体烃沸点和临界温度数据集对预测模型进行验证，发现池化操作的引入对神经指纹法的预测能力有明显提升，改进神经指纹法模型可自动选取对辛烷值有利和不利的结构特征，双键结构对单体烃马达法辛烷值的影响比芳环结构的影响更大。该预测模型对研究法辛烷值和马达法辛烷值的预测达到了同等水平，取得了良好的预测效果。

关键词: 图卷积神经网络, 神经指纹, 辛烷值, 汽油

Abstract: Based on the graph convolution neural network, an improved neural fingerprint method was established by introducing pooling operation. A model for predicting octane number of pure hydrocarbons in gasoline was established by this method as an important part of blending model of gasoline on molecular level. Through the verification on the boiling point and critical temperature data set, the prediction ability of neural fingerprint method was improved obviously by introducing pooling operation. The favorable and unfavorable features could be selected automatically by the improved neural fingerprint method. It was also found that the effect of double bond on motor octane number was greater than that of aromatic ring. The improved neural fingerprint method also had good results in the prediction of both research octane number and motor octane number of pure hydrocarbons in gasoline.

Key words: graph neural network, neural fingerprint, octane number, gasoline

崔晨何杉吕文进张霖宙周祥. 基于图卷积神经网络汽油单体烃辛烷值的预测[J]. 石油炼制与化工, 2021, 52(7): 82-87.

参考文献

[1] Healy W. C., Jr C. W. M., Peterson R. T. A new approach to blending octane[J]. Proc Am Inst, 1959, 39(3):132-136
[2] Ghosh P., Hickey K. J., Jaffe S. B. Development of a Detailed Gasoline Composition-Based Octane Model[J]. Industrial & Engineering Chemistry Research, 2006, 45(1):337-345
[3] Teng S. T., Williams A. D., Urdal K. Detailed Hydrocarbon Analysis of Gasoline by GC-MS (SI-PIONA)[J]. Journal of high resolution chromatography, 1994, (17)
[4] White C. M., Hackett J., Anderson R. R., et al. Linear temperature programmed retention indices of gasoline range hydrocarbons and chlorinated hydrocarbons on cross‐linked polydimethylsiloxane[J]. Journal of High Resolution Chromatography, 1992, 15(2):105-120
[5] Cui C., Billa T., Zhang L., et al. Molecular Representation of the Petroleum Gasoline Fraction[J]. Energy & Fuels, 2018, 32(2):1525-1533
[6] Knocking Characteristics of Pure Hydrocarbons. Special Technical Publication No. 225[S]. West Conshohocken, PA: American Society for Testing Materials, 1958
[7] API Technical Data Book on Petroleum Refining[M]. Washington, DC: American Petroleum Institute, 1982
[8] LOVELL W. G. Knocking Characteristics of Hydrocarbons[J]. Industrial and Engineering Chemistry, 1948, 40(12):2388-2438
[9] Katritzky A. R., Lobanov V. S., Karelson M. QSPR: the correlation and quantitative prediction of chemical and physical properties from structure[J]. Chemical Society Reviews, 1995, 24
[10] Joback K. G., Reid R. C. Estimation of Pure-component Properties From Group-contributions[J]. Chemical Engineering Communications, 1987, 57(1-6):233-243
[11] Albahri T. A. Structural Group Contribution Method for Predicting the Octane Number of Pure Hydrocarbon Liquids[J]. Industrial & Engineering Chemistry Research, 2003, 42(3):657-662
[12] Kubic W. L., Jenkins R. W., Moore C. M., et al. Artificial Neural Network Based Group Contribution Method for Estimating Cetane and Octane Numbers of Hydrocarbons and Oxygenated Organic Compounds[J]. Industrial & Engineering Chemistry Research, 2017, 56(42):12236-12245
[13] Wiener H. Structural determination of paraffin boiling points[J]. Journal of the American Chemical Society, 1947, 69(1):17-20
[14] Gani R., Harper P. M., Hostrup M. Automatic Creation of Missing Groups through Connectivity Index for Pure-Component Property Prediction[J]. Industrial & Engineering Chemistry Research, 2005, 44(18):7262-7269
[15] Kier L. B., Hall L. H. Molecular connectivity in structure-activity analysis[M]. Research Studies, 1986
[16] Hukkerikar A. S., Meier R. J., Sin G., et al. A method to estimate the enthalpy of formation of organic compounds with chemical accuracy[J]. Fluid Phase Equilibria, 2013, 348:23-32
[17] Hukkerikar A. S., Sarup B., Ten Kate A., et al. Group-contribution+ (GC+) based estimation of properties of pure components: Improved property estimation and uncertainty analysis[J]. Fluid Phase Equilibria, 2012, 321:25-43
[18] Smolenskii E. A., Ryzhov A. N., Bavykin V. M., et al. Octane numbers (ONs) of hydrocarbons: a QSPR study using optimal topological indices for the topological equivalents of the ONs[J]. Russian Chemical Bulletin, 2007, 56(9):1681-1693
[19] Ryzhov A. N., Strizhakova Y. A., Smolenskii E. A., et al. Modeling the octane numbers of alkenes by the inverse function method[J]. Petroleum Chemistry, 2011, 51(5):354-362
[20] Duvenaud D., Maclaurin D., Aguilera-Iparraguirre J., et al. Convolutional Networks on Graphs for Learning Molecular Fingerprints[J]. 2015,
[21] Xu Y., Pei J., Lai L. Deep Learning Based Regression and Multiclass Models for Acute Oral Toxicity Prediction with Automatic Chemical Feature Extraction[J]. Journal of Chemical Information and Modeling, 2017, 57(11):2672-2685
[22] Coley C. W., Barzilay R., Green W. H., et al. Convolutional Embedding of Attributed Molecular Graphs for Physical Property Prediction[J]. Journal of Chemical Information and Modeling, 2017, 57(8):1757-1772