关键词: 菊花, 三维荧光光谱, 特征组分鉴别, 平行因子算法, 支持向量机, BP神经网络

Abstract: Chrysanthemum has ornamental value and a fragrant smell, and can be used for medicinal purposes and as tea. It contains unique characteristic components that give it its aroma and medicinal properties. To improve the detection efficiency of these components, a fast identification method based on three-dimensional fluorescence spectral coupled parallel factor algorithm (3DEEM-PARAFAC) was proposed. The spectral characteristics of four species of chrysanthemum were analyzed after obtaining their three-dimensional fluorescence spectral matrix (EEMs). To remove interference data such as Raman scattering and Rayleigh scattering, data preprocessing was performed, and outliers were removed. Feature extraction was then performed using PARAFAC, which identified two characteristic fluorescent components of chrysanthemum as amino acids and flavonoids based on variance interpretation rate and residual analysis. Finally, this study employs Support Vector Machines (SVM) and Back Propagation Neural Network (BPNN) to analyze the characteristic variables and establish a fast and lossless identification model. The training set results for SVM and BPNN were (100%, 95.93%), respectively, while the test set results were (94.50%, 89.61%). The study found that 3DEEM-PARAFAC combined with SVC enables both qualitative and quantitative analysis of chrysanthemum's characteristic components and rapid identification of the flower. This has significant implications for nondestructive analysis and detection of chrysanthemum.

Key words: chrysanthemum, three-dimensional fluorescence spectroscopy, characteristic component identification, parallel factor analysis, support vector machine, Back Propagation Neural Network