|
| FTIR-based method for indicator monitoring during the ring-opening of epoxidized vegetable oils |
| |
| DOI： |
| KeyWord:vegetable oils-based polyols hydroxyl value epoxy value polymeric hydroxyl value chemometrics |
| FundProject:河南工业大学粮油食品学院省部级科研平台开放课题项目（GO202306）；学生学术及科技创新培育项目（GJXY202406） |
| Author Name | Affiliation | | LI Wenlong1, YE Zihan2, WANG Yan1, SUN Yuewen1, BI Yanlan1,
PENG Dan1, LI Jun1 | 1.Food Engineering Technology Research Center/Key Laboratory of Henan Province, Henan University of
Technology, Zhengzhou 450001, China
2.College of International Education, Henan
University of Technology, Zhengzhou 450001, China |
|
| Hits: 24 |
| Download times: 14 |
| Abstract: |
| To meet the online rapid monitoring of properties of vegetable oils-based polyols during industrial production, epoxidized soybean oils were used as the raw material to prepare 57 vegetable oils-based polyols (hydroxyl value of 94.1-229.4 mgKOH/g, epoxy value of 0-3.68%, polymeric hydroxyl value of 85.1-207.5 mgKOH/g) with different hydroxyl contents. Fourier transform infrared spectroscopy (FTIR) was taken to scan those vegetable oils-based polyols with 43 of them as the calibration set and the rest as the validation set so as to establish the quantitative models of hydroxyl, epoxy, and polymeric hydroxyl values. The effects of Normalization, Second-order derivatives, Standard Normal Variable Transformation, Multiple Scattering Correction, and Orthogonal Scattering Correction (OSC) on the quantitative models of the three indicators were examined. The optimal modeling wavenumber ranges were identified and the effects of three modeling methods Principal Component Regression (PCR), Partial Least Squares (PLS), and Support Vector Machine Regression（SVR） were compared on modeling performance. The results showed that the root mean square errors of calibration of PLS models obtained using the OSC-preprocessed data were reduced by 12.2%, 12.2%, and 13.0% for hydroxyl, epoxy, and polymeric hydroxyl values, respectively, compared to those using the original data. The optimal wavenumber ranges were 500-4 000 cm-1, 800-900 cm-1, and 1 000-1 200 cm-1 for hydroxyl value, epoxy value, and polymeric hydroxyl value models, respectively. The best modeling method for hydroxyl value and polymeric hydroxyl value models was PLS, and for epoxy value model was PLS and PCR. The coefficients of determination (R2) of the optimal models for the three indicators all reached above 0.98. Thus, it is feasible that FTIR technique is used to establish quantitative models for rapid monitoring of hydroxyl, epoxy, and polymeric hydroxyl values of vegetable oils-based polyols. |
| View full text View/Add Comment Download reader |
| Close |
|
|
|
