Near infrared spectroscopy in combination with chemometrics as a process analytical technology (PAT) tool for on-line quantitative monitoring of alcohol precipitation

The application of near infrared (NIR) spectroscopy for on-line quantitative monitoring of alcohol precipitation of the Danhong injection was investigated. For the NIR measurements, two fiber optic probes designed to transmit NIR radiation through a 2 mm path length flow cell were applied to collect spectra in real-time. Particle swarm optimization- (PSO-) based least square support vector machines (LS-SVM) and partial least squares (PLS) models were developed for quantitative analysis of the critical intermediate quality attributes: the soluble solid content (SSC) and concentrations of danshensu (DSS), protocatechuic aldehyde (PA), hydroxysafflor yellow A (HSYA) and salvianolic acid B (SAB). The optimal models were then used for on-line quantitative monitoring of alcohol precipitation. The results showed that the PSO-based LS-SVM with a radial basis function (RBF) kernel was slightly better than the conventional PLS method, even though both methods exhibited satisfactory fitting results and predictive abilities. In this study, successful models were built and applied on-line; these models proffer real-time data and instant feedback about alcohol precipitation.

Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slideHighlights► NIR spectroscopy was applied to on-line monitoring of alcohol precipitation. ► PSO was used to optimize the hyper-parameters of LS-SVM model. ► LS-SVM and PLS models were developed for five critical quality attributes. ► PSO based LS-SVM provided slightly better fitting results and prediction accuracy.