Quinolone antibiotics have good antibacterial properties and are commonly used antibiotics in the dairy industry. Currently, the problem of excessive antibiotics in dairy products is very serious. As an ultra-sensitive detection technology, Surface-Enhanced Raman Scattering (SERS) was applied to the detection of quinolone antibiotics in this work. In order to classify and quantify three antibiotics (Ciprofloxacin, Norfloxacin, Levofloxacin) with highly similar molecular structures, a combination of magnetic COF-based SERS substrate and machine learning algorithms (PCA-k-NN, PCA-SVM, PCA-Decision Tree) was used. The classification accuracy of the spectral dataset could reach 100% and the results of LOD calculation were: CIP: 5.61 × 10⁻⁹M, LEV: 1.44 × 10⁻⁸M, NFX: 1.56 × 10⁻⁸M. This provides a new method for the detection of antibiotics in dairy products.

喹诺酮类抗生素具有良好的抗菌性能，是乳品行业常用的抗生素。目前，乳制品中抗生素超标问题十分严重。本工作将表面增强拉曼散射(SERS)作为一种超灵敏的检测技术应用于喹诺酮类抗生素的检测。为了对分子结构高度相似的三种抗生素（环丙沙星、诺氟沙星、左氧氟沙星）进行分类和量化，结合基于磁性COF的SERS底物和机器学习算法（PCA-k-NN、PCA-SVM、PCA-Decision Tree）被使用了。光谱数据集分类准确率可达100%，LOD计算结果为：CIP：5.61×10 ^-9 M，LEV：1.44×10 ^-8 M，NFX：1.56×10⁻⁸ M。这为检测乳制品中的抗生素提供了一种新方法。