利用三维同步荧光光谱法获取不同氧化状态下的葵花籽油荧光光谱数据, 同时采集葵花籽油品质指标。 运用平行因子法对三维同步荧光光谱矩阵进行降维处理, 通过iPLS(interval partial least squares), BiPLS(backward interval partial least squares)和SiPLS(synergy interval partial least squares)模式识别方法进行数学建模。 结果表明: 波长差Δλ=50 nm时, 样品同步荧光光谱具有显著差异, 筛选用于数学建模初始数值。 不同模式识别方法建模结果显示, iPLS, BiPLS和SiPLS法所得校正集模型和预测集模型的相关系数分别为0.908 3, 0.961 2, 0.954 5和0.872 3, 0.925 2, 0.852 5, 交互验证均方根误差分别为0.050 3, 0.033 1, 0.035 9和0.073 3, 0.054 1, 0.065 5, 比较发现采用BiPLS法建模效果最好。 该研究将为葵花籽油品质快速辨别提供理论基础和技术支持, 为其他食用油脂的快速检测提供方法指导。

Fluorescence spectra of sunflower oil under different oxidation conditions were obtained by three-dimensional synchronous fluorescence spectrometry, and the quality indexes were collected at the same time. The parallel-factor method was used to reduce the dimension of the three-dimensional synchronous fluorescence spectrum, and the mathematical models were established by iPLS, BiPLS and SiPLS pattern recognition methods. The results show: the two-dimensional synchronous fluorescence spectra of the samples have significant differences when the wavelength difference Δλ=50 nm, which are used for mathematical modeling of initial values. The results of partial least squares modeling show that the correlation coefficients of calibration set and prediction set of iPLS, BiPLS, SiPLS are 0.908 3, 0.961 2, 0.954 5 and 0.872 3, 0.925 2, 0.852 5. It’s found that BiPLS method is test. The study provides the theoretical basis and technical support for rapid identification of sunflower oil quality, and offers a theoretical basis for other related oil rapid detection.