目的 采用灰色关联度法和FCM（Fuzzy C-Means）算法相结合的模式识别模型进行木香质量多指标综合评价研究。方法 HPLC法测定木香烃内酯、去氢木香内酯的含量及水蒸气蒸馏法测定木香挥发油的含量作为木香评价考察指标。采用灰色关联度法进行中药质量排序，FCM算法进行中药质量分类。结果 不同产地木香平均关联度排序为云南>广东>广西>湖南>四川>安徽亳州>北京>河北，FCM算法将不同产地木香样品分为3类。由2种方法结合分析可知云南、广东、广西在质量高的类组中；湖南、四川在质量中等的类组中；北京、安徽亳州、河北在质量低的类组中。结论 构建了用于评价中药材质量的综合模式识别模型体系，首次在木香药材中采用模糊聚类算法，旨在形成一种木香质量评价研究方法，为现代模式识别、数据挖掘等新方法在中药质量评价方面的应用提供思路引领和经验借鉴。

Objective To comprehensively evaluate the Aucklandiae Radix quality by combining gray correlation method and FCM (Fuzzy C-Means) algorithm. Methods HPLC method was used to determine the content of costunolide and dehydrocostus, and the content of volatile oil was determined by steam distillation. The gray correlation method was used to sort the quality of Chinese herbal medicines, and the FCM algorithm was used to classify the quality of medicines. Results The average correlation degree of Aucklandiae Radix from different producing areas was:Yunnan > Guangdong > Guangxi > Hunan> Sichuan > Bozhou > Beijing > Hebei. FCM algorithm divided samples into three categories:Yunnan, Guangdong and Guangxi were in the high quality group; Hunan and Sichuan were in the medium quality group; Beijing, Bozhou, and Hebei were in the low quality group. Conclusion Constructing an integrated pattern recognition model system for evaluating the quality of Chinese medicinal materials, and the fuzzy clustering algorithm was adopted for the first time in Aucklandiae Radix. The purpose of this study is to form a kind of research method of Aucklandiae Radix quality evaluation and provide a way to guide and apply the new methods of modern pattern recognition and data mining in the application of traditional Chinese medicine quality evaluation.

国家自然科学基金资助项目（81473413）国家自然科学基金资助项目（81274060）