目的 以温郁金和桂郁金的红外光谱为聚类分析对象，研究红外光谱结合(Soft independent modeling of classanalogy，SIMCA)聚类分析法对温郁金和桂郁金进行识别与分类的可行性。方法 以4000~650 cm-1范围内的红外光谱为对象，通过基线补偿(Offset)和散射校正(MSC)等预处理后，采用SIMCA聚类分析法建立识别模型。结果 所建模型对温郁金和桂郁金的识别率分别达到 97 %和 98 %，拒绝率均为100 %。对所建模型测试结果正确。结论 该方法可实现对温郁金和桂郁金的快速鉴别。

Objective With the infrared spectrum of Curcuma wenyujin Y.H. Chen et C. Ling and Curcuma kwangsiensis S.G. Lee et C.F. Liang as the object，the feasibility of identification and classification of Curcuma wenyujin Y.H. Chen et C. Ling and Curcuma kwangsiensis S.G. Lee et C.F. Liang was studied by using infrared spetroscopy coupled with SIMCA(soft independent modeling of classanalogy) pattern recognition method. Methods In the spectra region between 4000～650 cm-1，two models of Curcuma wenyujin Y.H. Chen et C. Ling and Curcuma kwangsiensis S.G. Lee et C.F. Liang were built separately after preprocessing by baseline correction(offset) and multiplicative scatter correction (MSC) with SIMCA pattern recognition method. Results The result showed that the recognition rate was 97 % for Curcuma wenyujin Y.H.Chen et C.Ling and was 98 % for Curcuma kwangsiensis S.G.Lee et C.F.Liang，and the rejection rate of them both was 100 %. The method was tested in the blind samples，and satisfactory results were achieved. Conclusion Curcuma wenyujin Y.H.Chen et C. Ling and Curcuma kwangsiensis S.G. Lee et C.F. Liang could be identified rapidly by the combination of IR fingerprint spectrum and clustering analysis.

R284.1

卫生部科学研究基金(WKJ2010-2-019)；浙江省中医药科学研究基金计划(2012ZB032)；浙江省教育厅科研计划(Y201224449)。