目的 基于快速气相电子鼻对3种五味子Schisandrae Chinensis Fructus（SCF）饮片进行气味特征分析，并建立人工神经网络（artificial neural network，ANN）快速判别模型，拟解决生产流通及临床使用过程中的品种混淆问题。方法 通过建立五味子饮片Heracles NEO快速气相电子鼻分析方法，对生五味子、醋五味子（vinegar-processed Schisandrae Chinensis Fructus，VSCF）及蜜五味子（honey-processed Schisandrae Chinensis Fructus，HSCF）气味指纹图谱进行了分析，通过计算各成分峰双柱Kovats相对保留指数，结合Arochembase数据库指认了各样品气味成分及潜在气味差异标志物，最终建立人工神经网络快速判别模型。结果 建立了五味子饮片快速气相电子鼻分析条件并通过了方法学考察，通过对3种五味子饮片气味指纹图谱分析，指认了36个气味特征成分，结合热图分析发现，乙酸和2-甲基-2-丙烯酸己酯可能是醋五味子气味差异标志物；3-甲基十六烷和3-乙基十六烷可能是蜜五味子的潜在气味差异标志物。基于快速气相电子鼻检测结果的人工神经网络判别模型训练正确率和测试正确率均为100.0%，可成功实现对3种五味子饮片的快速鉴别。结论 Heracles NEO快速气相电子鼻结合人工神经网络可较好分析五味子饮片气味特征并进行快速、准确判别分析，可应用于在线检测等质量监控环节，也为其他饮片快速检测方法的建立提供参考。

Objective The odor characteristics of three Wuweizi (Schisandrae Chinensis Fructus, SCF) pieces were analyzed based on Heracles NEO flash GC e-nose, and the rapid discrimination model of artificial neural network (ANN) was established to solve the variety confusion in the process of production, circulation and clinical use. Methods The Heracles NEO flash GC e-nose analysis method was established to analyze the odor fingerprints of SCF, vinegar-processed Schisandrae Chinensis Fructus (VSCF) and honey-processed Schisandrae Chinensis Fructus (HSCF). The odor components and potential odor difference markers were identified by calculating the two-column Kovats relative retention index of the peak of each component with ArochemBase database. Finally, the rapid discrimination model of ANN was established. Results The flash GC e-nose analysis conditions of SCF pieces were established and the method was investigated. By analyzing the odor fingerprint of three kinds of SCF pieces, 36 odor characteristic components were identified. It was found that acetic acid and hexyl 2-methyl-2-propenoate might be the odor difference markers of VSCF. 3-Methylhexadecane and 3-ethyl hexadecane may be potential markers of odor difference in HSCF. The training accuracy and test accuracy of the ANN model were both 100.0%. Conclusion Heracles NEO flash GC e-nose combined with ANN can better analyze the odor characteristics of SCF pieces, conduct fast and accurate discriminant analysis. It can be applied to quality control links such as online detection, and also provide reference for the establishment of other rapid detection technologies for traditional Chinese medicine.

R283.6

国家重点研发计划“中药饮片质量识别关键技术研究”（2018YFC1707000）；国家自然科学基金项目“基于肠道菌群介导的五味子醋制增强收敛止泻效应机制研究”（81873008）；江苏省研究生培养创新工程“研究生科研与实践创新计划”（KYCX21_1752）