灰色系统理论在中药色谱指纹图谱模式识别中的应用研究

建立了基于灰色系统理论的中药色谱指纹图谱模式识别模型。运用基于范数与信息熵赋权法的灰色关联分析,求出各待比较图谱特征变量数据序列与理想图谱特征变量数据序列之间的灰色关联度,并依据模糊匹配中的“最大匹配度”原则进行判断,从而达到品种识别和质量评价的目的。该模型在56批次不同品种化橘红药材样品的高效液相色谱分析中取得较满意的结果: 熵权与范数的灰色关联分析均能准确识别出毛橘红与光橘红两个品种,克服了传统相似度或灰色关联在化橘红色谱指纹图谱分析中的误判问题;对药材中化学成分的种类与含量十分接近的毛橘红不同栽培品种的识别率超过92.85%。此外,该模型计算量较少,整个模式识别过程通过计算机编程实现,操作简单。实验结果显示,灰色系统理论在中药色谱指纹图谱模式识别中有良好的适应性。

Research on the application of grey system theory in the pattern recognition for chromatographic fingerprints of traditional Chinese medicine

A model based on grey system theory was proposed for pattern recognition in chromatographic fingerprints (CF) of traditional Chinese medicine (TCM). The grey relational grade among the data series of each testing CF and the ideal CF was obtained by entropy and norm respectively, then the principle of “maximal matching degree” was introduced to make judgments, so as to achieve the purpose of variety identification and quality evaluation. A satisfactory result in the high performance liquid chromatographic (HPLC) analysis of 56 batches of different varieties of Exocarpium Citrus Grandis was achieved with this model. The errors in the chromatographic fingerprint analysis caused by traditional similarity method or grey correlation method were overcome, as the samples of Citrus grandis ‘Tomentosa’ and Citrus grandis (L.) Osbeck were correctly distinguished in the experiment. Furthermore in the study on the variety identification of Citrus grandis ‘Tomentosa’, the recognition rates were up to 92.85%, although the types and the contents of the chemical compositions of the samples were very close. At the same time, the model had the merits of low computation complexity and easy operation by computer programming. The research indicated that the grey system theory has good applicability to pattern recognition in the chromatographic fingerprints of TCM.