基于电子鼻技术的LLE＋SMA算法对木香的定性鉴别研究

目的 采用电子鼻技术建立气味指纹图谱，对不同产地木香进行定性鉴别。方法 收集8个不同产地木香样品48批，采用电子鼻获取各样品气味信息，采用基于Fisher鉴别准则的LDA算法和非线性降维LLE+SMA算法区别不同产地木香。结果 基于Fisher鉴别准则的LDA算法并不能对不同产地木香进行很好地区分，部分产地样品存在大量重叠现象，而采用LLE+SMA算法判别效果极为显著，可以完全把8个不同产地木香区分开。结论 电子鼻技术应用于不同产地木香的区分是可行的，为木香质量评价提供一种新的思路与方法，具有重要理论价值与实际意义。

Qualitative identification of Aucklandiae Radix with LLE + SMA classification models based on electronic nose

Objective To establish an odor fingerprint with electronic nose technology to qualitatively identify Aucklandiae Radix odor from different producing areas. Methods Eight batches of Aucklandiae Radix samples from eight different producing areas were collected. The odor information of each sample was obtained by electronic nose. The LDA algorithm based on Fisher''s identification criterion and the nonlinear dimensionality reduction LLE + SMA algorithm were used to distinguish the Aucklandiae Radix odor of different origins. Results It was found that the LDA algorithm based on Fisher''s discriminant criterion could not distinguish the Aucklandiae Radix scent of different producing areas. Some of the samples in the place of origin had a lot of overlap, and the LLE + SMA algorithm could distinguish the odor very well. It can completely distinguish eight batches of Aucklandiae Radix samples from eight different producing areas. Conclusion It is feasible to apply the electronic nose technology to the odor differentiation of Aucklandiae Radix from different producing areas, and provide new ideas and methods for the quality evaluation of Aucklandiae Radix.