Advanced process control for salvianolic acid A conversion reaction based on data-driven and mechanism-driven model. (July 2022)

• Record Type: Journal Article
• Authors: Dong, Xiaoxiao; Yan, Xu; Qu, Haibin
• Is Part Of: Process biochemistry. Volume 118(2022)
• Page Start: 1
• Page End: 10
• Publication Date: 2022-07
• Subjects: SAA salvianolic acid A -- SAB salvianolic acid B -- RSM Radix Salviae Miltiorrhizae -- LA lithospermic acid -- DSS danshensu -- HPLC high-performance liquid chromatography -- NIRS near-infrared spectroscopy -- PA protocatechualdehyde -- RA rosmarinic acid -- 1stDer the first derivative -- S-G Savitzky-Golay -- SNV standard normal variate -- SPXY sample set partitioning based on joint X-Y distances -- LV latent variable -- RMSECV root mean square error of cross-validation -- PLSR partial least squares regression -- RMSEC root mean square error of calibration -- RMSEP root mean square error of prediction -- RC correlation coefficient of calibration -- RP correlation coefficient of prediction; Salvianolic acid A conversion reaction -- Near-infrared spectroscopy -- Kinetic model -- Process control; Biochemical engineering -- Periodicals; Biotechnology -- Periodicals; Biochemistry -- periodicals; Biotechnology -- periodicals; Chemical Engineering -- periodicals; Génie biochimique -- Périodiques; Biotechnologie -- Périodiques; Biochemical engineering; Biotechnology; Periodicals; 660.63
• DOI: 10.1016/j.procbio.2022.04.001
• Languages: English
• ISSNs: 1359-5113

Abstract

Abstract: On-line monitoring and process control are the most efficient techniques to ensure high quality, low cost, and effective resource utilization in modern process engineering. The high-temperature and high-pressure operating conditions make the process control challenging during conversion reaction. In this study, the contents of main substances produced during chemical conversion of salvianolic acid A were monitored and controlled by an advanced process control method. At first, a kinetic model was established according to the reaction mechanism. Then, a high-precision partial least squares calibration model was established based on near-infrared spectroscopy. Finally, the conversion reaction was controlled in real-time by combining the data-driven model with the mechanism model. The correlation coefficients obtained by the data-driven model were all > 0.9. The fitting correlation coefficients of reaction rate constants were all > 0.9. The concentration of the reaction product was controlled within the limit by adjusting the reaction temperature based on the concentration-temperature curve. Combining the data-driven model with the mechanism model, the reaction process was controlled in real-time. This study emphasized the advantages of advanced process control for high temperature and high-pressure reaction processes and exemplified it as an important step in implementing intelligent production. Graphical Abstract: ga1 Highlights: Rapid and accurate monitoring of SAA conversion reaction using NIR. Optimal reaction temperature was obtained. Prediction of transformation substances based on kinetic model. Combining the data-driven model with the mechanism model, the reaction process was controlled in real-time.