Development of a SVM model for Prediction of Hydrocracking Product Yields


Cite item

Full Text

Open Access Open Access
Restricted Access Access granted
Restricted Access Subscription Access

Abstract

In this study Support Vector Machine (SVM) and Lump Kinetic Model was used to predict the hydrocracking products yield by using data which obtained from Tehran Refinery in Iran. The minimum calculated squared correlation coefficient and key parameters of SVM were used to evaluate the SVM performance. Output variables of SVM model were optimized for hydrocracking products. Kinetic parameter and mean squared error values for the 4-Lumped Kinetic models have been developed. Mean Squared Errors (MSE) of training set for Diesel, Naphtha and LPG were obtained 0.179, 0.116, and 0.174 and for testing set are 0.164, 0.148, and 0.132, respectively. It can be concluded that SVM can be used as a reliable and accurate estimation method with comparison between SVM data and Tehran refining unit. Such validated models may be regarded as valuable tools for process optimization, control, design, catalyst selection and provide a better insight into the process.

About the authors

K. Sharifi

Petroleum Refining Division, Research Institute of Petroleum Industry

Author for correspondence.
Email: sharifikh@ripi.ir
Iran, Islamic Republic of, Tehran, P. O. Box 1485733111

A. Safiri

Petroleum Refining Division, Research Institute of Petroleum Industry

Email: sharifikh@ripi.ir
Iran, Islamic Republic of, Tehran, P. O. Box 1485733111

M. Haghighi Asl

Petroleum Refining Division, Research Institute of Petroleum Industry; Department of Engineering, Senior Process Engineer, South Pars Gas Complex

Email: sharifikh@ripi.ir
Iran, Islamic Republic of, Tehran, P. O. Box 1485733111; Assaluyeh, P. O. Box 311/75391

H. Adib

School of Chemical Engineering, Amirkabir University of Technology (Tehran Polytechnic)

Email: sharifikh@ripi.ir
Iran, Islamic Republic of, Tehran, 15914

B. Nonahal

Petroleum Refining Division, Research Institute of Petroleum Industry

Email: sharifikh@ripi.ir
Iran, Islamic Republic of, Tehran, P. O. Box 1485733111

Supplementary files

Supplementary Files
Action
1. JATS XML

Copyright (c) 2019 Pleiades Publishing, Ltd.