目的 基于性状电子检测技术，结合多元分析算法，建立不同产地薏苡仁的快速鉴别方法。方法 使用CM-5分光测色仪测定薏苡仁的色度值，建立决策树（decision tree，DT）模型、k最邻近算法（k-nearest neighbor，KNN）模型和贝叶斯（Bayes）判别模型。其次，根据超快速气相电子鼻检测气味成分，建立不同产地薏苡仁的判别因子分析（discriminant factor analysis，DFA）模型及热图，探究不同产地薏苡仁的气味信息差异。同时，采用偏最小二乘法判别分析（partial least squares discrimination analysis，PLS-DA）模型探究不同产地的差异标志物。最后，将薏苡仁表面颜色与内在气味成分进行相关性分析。结果 2-丙烯酸、2-甲基丁醛、甲苯、丙醛可以作为区分不同产地薏苡仁的主要气味标志物，丁二酮、2,3-乙酰基丙酮、3-己醇等多种气味成分与L^*、a^*、b^*值存在显著相关性。结论 电子眼联合超快速气相电子鼻技术能够快速、准确鉴别不同产地的薏苡仁，该方法对于多产地中药的鉴别和质量控制具有借鉴意义。

Objective To establish a rapid identification method of coicis semen from different habitats based on electronic trait detection technology and multiple analysis algorithm. Methods Firstly, the chromaticity value of coicis semen was determined by CM-5 spectrophotometer, and the Decision Tree (DT) model, k-nearest neighbor (KNN) model and Bayes discriminant model were established. Secondly, the discriminant factor analysis (DFA) model and heat map of coicis semen of different habitats were established according to the detection of odor components by ultra-fast electronic nose analysis to explore the difference of odor information of coicis semen in different habitats. At the same time, the partial least squares discriminant analysis (PLS-DA) model was used to explore difference markers from different habitats. Finally, the correlation between the surface color of coicis semen and the internal odor components was analyzed. Results 2-Propenoate, 2-methylbutanal, toluene, propanal can be used as the main odor markers to distinguish coicis semen from different habitats, butan-2-one, 2,3-pentanedione, 3-hexanol and other odor components were significantly correlated with L^*, a^*, and b^* values. Conclusion Electronic eye combined with ultra-fast gas phase electronic nose technique can identify coicis semen from different habitats quickly and accurately. This method can be used as reference for the identification and quality control of multi-habitats Chinese medicines.

R286.2

国家重点研发计划（2018YFC1707000）