Improved Prediction of Oxide Content in Cement Raw Meal by Near-Infrared Spectroscopy Using Sequential Preprocessing through Orthogonalization (SPORT)

作者：Yong Zhen Zhang, Yina Wang, Zhi Zhao, Lei Zhang, Hang Xiao · 发表于：Analytical Letters · 年份：2023 · DOI：10.1080/00032719.2023.2266070 · 被引用次数：6 · 研究领域：Spectroscopy and Chemometric Analyses、Spectroscopy Techniques in Biomedical and Chemical Research、Water Quality Monitoring and Analysis

Near-infrared (NIR) spectroscopy is a nondestructive technique extensively employed in various fields. Despite its advantages, near-infrared spectroscopy still faces significant challenges due to the intricate physical and chemical phenomena that arise from the interaction between light and matter. This interaction typically results in light absorption and scattering, leading to the NIR signal containing comprehensive information about these phenomena’s interactions. Accurate determination of CaCO3, SiO2, Fe2O3 and Al2O3 in cement raw meal requires minimizing scattering effects from the spectrum, but selecting an appropriate pretreatment technique is often challenging. In this paper, we enhance the predictive ability of NIRS for determining the four oxides in cement raw meal by implementing sequential preprocessing through orthogonalization (SPORT). The SPORT method uses sequential orthogonal partial least squares (SO-PLS) to integrate data blocks obtained from different preprocessing techniques. We compare our method with conventional pretreatment methods for determining the content of four oxides in raw materials of cement using near-infrared spectroscopy. The results suggest that the SPORT method exhibits commendable calibration performance and distinctive characteristics. Moreover, SPORT demonstrates significant preprocessing selectivity, making it effective in addressing the challenges associated with complex interactions in near-infrared spectral analysis. In conclusion...