【摘要】 一种用于特征维数不等识别问题的自组织模糊神经网络被提出。网络专为特征维数不相等一类模式识别问题而构造,具有结构自组织,基于知识和学习能力等基本特征,采用两阶段工作模式——结构学习和参数学习。网络首先根据样本自组织学习网络结构,而后再进行集中批学习网络参数。作者用基于中药材色谱保留时间和保留峰面积的特征样本进行构造、训练学习后得到了一个中药材、中成药组成药材模糊识别网络。该网络既可用于中药材的识别,也可用于识别组成中成药各味中药材的复杂模式识别问题。经实验测试,达到了预期效果,为中成药药方解析和质量控制探索出一条新路。更多还原

【Abstract】 A self-organizing fuzzy neural network,which is used to solve the unequa l feature dimension recognition problems ,is proposed.The network is built spe cially for the pattern recognition when the feature dimensions are not e-qual. This network has such basic features as self-organizing structure and the abili ty based on knowledge and learning.It adapts two-phase modus operandi i.e.s tructure learning and parameter learning.First,it learns the network structure through self-organization by samples;and then learns the network parameters i n concentrated batch.The author use the feature samples based on traditional Ch inese medicine’s chromatographic reserved time and reserved peak aero build,tr ain and learn,and then a traditional Chinese medicine’ and traditional Chinese pharmaceutical medicine’ components fuzzy recognition network is obtained.Thi s network can be used not only to traditional Chinese medicine’ recognition,bu t also to recognize the components from traditional Chinese pharmaceutical medi cine.It has achieved better effect,and search out a new road for prescription analysis of traditional Chinese pharmaceutical medicine and quality control.更多还原