Open AccessMathematical Modeling for the Industrial 2-Mercaptobenzothiazole Batch Production Process
Author(s) -
Enzhi Liang,
Zhanqian Song,
Bin Liu,
Bujin Qi,
Yanpei Nie,
Zhihong Yuan
Publication year - 2022
Publication title -
acs omega
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.779
H-Index - 40
ISSN - 2470-1343
DOI - 10.1021/acsomega.1c06646
Subject(s) - generalizability theory , process (computing) , computer science , set (abstract data type) , mathematical optimization , production (economics) , partial least squares regression , industrial production , mathematics , machine learning , statistics , keynesian economics , economics , macroeconomics , programming language , operating system
As an important chemical intermediate, 2-mercaptobenzothiazole (MBT) is widely used in various processes, especially in the rubber industry. However, there is no first-principles model that describes the synthetic process of MBT. This paper focuses on the formulation of a reliable mathematical model represented by a series of differential and algebraic equations for the industrial batch MBT production process. It is difficult to estimate all of the unknown parameters in the model because of the lack of sufficient industrial/experimental data. Thus, a reduced estimable parameter set is derived by performing estimability analysis on the original estimation problem. A multiple-starting-point strategy is then applied to numerically solve the non-convex parameter estimation problem with the weighted least-squares approach. Subsequently, a cross-validation technique is employed to evaluate the generalizability of the proposed model. Finally, it is confirmed that the proposed model produces encouraging prediction performance with regard to independent test data.