运用人工神经网络和非线性映射技术评价中药复方制剂戊己丸的质量

Study on Quality Evaluation of Traditional Chinese Patent Medicine Wuji Wan by Adaptive Resonance Network and Non-Linear Mapping

运用二值自适应共振理论（ＡＲＴ１）网络和非线性映射（ＮＬＭ）技术对中药戊己丸进行了省缺药材的定性识别研究。实验数据为药材浸出物的红外光谱，特征提取采用Ｓｈａｎｎｏｎ信息量法。结果表明，ＡＲＴ１网络能对多种情况进行识别，正确识别率为８５％，明显优于ＮＬＭ技术。

This paper is concerned with qualitative recognition of omitted ingredien ts in traditional Chinese patent medicine Wuji Wan by adaptive resonance network(ART1)and non-linear mapping(NLM).Shannon information content theory was used for the feature selection of IR spectra of extracts.The result demonstrated that ART1 is capable of classification and recognition of omitted ingredients in Wuji Wan.