Department of Agro-Industry, Faculty of Agriculture Natural Resources and Environment, Naresuan University, Phitsanulok, Thailand.
Department of Agricultural Science, Faculty of Agriculture Natural Resources and Environment, Naresuan University, Phitsanulok, Thailand.
J Sci Food Agric. 2024 Sep;104(12):7249-7257. doi: 10.1002/jsfa.13546. Epub 2024 Apr 27.
Industrial starch hydrolysis allows the production of syrups with varying functionality depending on their Brix value and dextrose equivalent (DE). As the current methods for evaluating these products are labor-intensive and time-consuming, the objective of this study was to investigate the potential of near-infrared (NIR) spectroscopy for classifying the different tapioca starch hydrolysis products.
NIR spectra of samples of seven products (n = 410) were recorded in transflectance mode in the 12 000-4000 cm range. Next, orthogonal partial least squares (OPLS) regression models were built to predict the Brix and DE values of the different samples. To classify the different starch hydrolysis products, support vector machines (SVM) were trained using either the raw spectra or latent variables (LVs) obtained from the OPLS models. The best classification accuracy was obtained by the SVM classifier based on the LVs from the OPLS model for DE prediction, resulting in 95% correct classification over all classes.
These results show the potential of NIR spectroscopy for classifying tapioca starch hydrolysis products with respect to their functional properties related to the Brix and DE values. © 2024 Society of Chemical Industry.
工业淀粉水解可根据其 Brix 值和葡萄糖当量 (DE) 生产出具有不同功能的糖浆。由于目前评估这些产品的方法既耗时又费力，因此本研究旨在探讨近红外 (NIR) 光谱法在分类不同木薯淀粉水解产物方面的潜力。
在透射模式下记录了七种产品（n = 410）的 NIR 光谱，范围在 12,000-4000 cm 之间。接下来，建立了正交偏最小二乘（OPLS）回归模型来预测不同样品的 Brix 和 DE 值。为了对不同的淀粉水解产物进行分类，使用 OPLS 模型得到的原始光谱或潜在变量 (LV) 训练支持向量机 (SVM)。基于 OPLS 模型用于 DE 预测的 LV 的 SVM 分类器获得了最佳的分类准确性，所有类别均有 95%的正确分类。
这些结果表明，NIR 光谱法具有分类木薯淀粉水解产物的潜力，可根据与 Brix 和 DE 值相关的功能特性进行分类。 © 2024 化学工业协会。
