A Real-time Prediction System for Molecular-level Information of Heavy Oil Based on Machine Learning

作     者：Yuan Zhuang Wang Yuan Zhang Zhibo Yuan Yibo Yang Zhe Xu Wei Lin Yang Yan Hao Zhou Xin Zhao Hui Yang Chaohe Yuan Zhuang;Wang Yuan;Zhang Zhibo;Yuan Yibo;Yang Zhe;Xu Wei;Lin Yang;Yan Hao;Zhou Xin;Zhao Hui;Yang Chaohe

作者机构：State Key Laboratory of Chemical SafetySINOPEC Research Institute of Safety Engineering Co.LtdQingdaoShandong 266000People’s Republic of China State Key Laboratory of Heavy Oil ProcessingChina University of PetroleumQingdaoShandong 266580People’s Republic of China College of Chemistry and Chemical EngineeringOcean University of ChinaQingdaoShandong 266100People’s Republic of China 

出 版 物：《China Petroleum Processing & Petrochemical Technology》 (中国炼油与石油化工（英文版）)

年 卷 期：2024年第26卷第2期

页      面：121-134页

核心收录：

学科分类：0820[工学-石油与天然气工程] 12[管理学] 081702[工学-化学工艺] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 0817[工学-化学工程与技术] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：the National Natural Science Foundation of China(22108307) the Natural Science Foundation of Shandong Province(ZR2020KB006) the Outstanding Youth Fund of Shandong Provincial Natural Science Foundation(ZR2020YQ17) 

主　　题：heavy distillate oil molecular composition deep learning SHAP interpretation method 

摘      要：Acquiring accurate molecular-level information about petroleum is crucial for refining and chemical enterprises to implement the“selection of the optimal processing route*** the development of data prediction systems represented by machine learning,it has become possible for real-time prediction systems of petroleum fraction molecular information to replace analyses such as gas chromatography and mass ***,the biggest difficulty lies in acquiring the data required for training the neural *** address these issues,this work proposes an innovative method that utilizes the Aspen HYSYS and full two-dimensional gas chromatography-time-of-flight mass spectrometry to establish a comprehensive training ***,a deep neural network prediction model is developed for heavy distillate oil to predict its composition in terms of molecular *** training,the model accurately predicts the molecular composition of catalytically cracked raw oil in a *** validation and test sets exhibit R2 values of 0.99769 and 0.99807,respectively,and the average relative error of molecular composition prediction for raw materials of the catalytic cracking unit is less than 7%.Finally,the SHAP(SHapley Additive ExPlanation)interpretation method is used to disclose the relationship among different variables by performing global and local weight comparisons and correlation analyses.