College of Pharmaceutical Engineering of Traditional Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin 301617, PR China; Tianjin Key Laboratory of Intelligent and Green Pharmaceuticals for Traditional Chinese Medicine, Tianjin 301617, PR China.
College of Pharmaceutical Engineering of Traditional Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin 301617, PR China.
Food Chem. 2024 Dec 1;460(Pt 3):140737. doi: 10.1016/j.foodchem.2024.140737. Epub 2024 Aug 3.
In order to achieve rapid and effective identification of Hebei yam, a qualitative discrimination model was constructed based on near infrared (NIR), mid infrared (MIR), and microscopic Raman spectra in combination with individual spectra and multispectral data fusion strategies. The results showed that the gray wolf optimizer-support vector machine (GWO-SVM) model constructed by mid-level fusion using the three feature spectra performed the best in distinguishing the geographic origin of the yam, with a prediction accuracy of 100.00% in both the training set and the test set, and an F1 score of 1.00. The results indicated that due to spectral complementarity, NIR, MIR and Raman combined with feature-level fusion can be used as a powerful, non-destructive, fast and feasible tool for geographic origin classification and brand protection of Hebei yam. This work is expected to be a potential method for origin identification analysis and quality monitoring in the food and pharmaceutical industries.
为实现对河北山药的快速有效鉴别，本研究构建了基于近红外（NIR）、中红外（MIR）和微观拉曼光谱的定性判别模型，结合个体光谱和多光谱数据融合策略。结果表明，利用三种特征光谱进行中层次融合构建的灰狼优化支持向量机（GWO-SVM）模型在区分山药产地方面表现最佳，在训练集和测试集中的预测准确率均为 100.00%，F1 得分为 1.00。结果表明，由于光谱互补性，NIR、MIR 和 Raman 结合特征级融合可作为一种强大、无损、快速且可行的工具，用于河北山药的产地分类和品牌保护。本研究有望成为食品和制药行业产地鉴别分析和质量监测的潜在方法。
