基于LM-BP神经网络的穿心莲药材分类识别  

The Application of BP Neural Network Improved with LM Algorithm in the Reorganization and Classification of Andrographis paniculata Nees

作　　者：周舒冬[1] 李倚岳 李丽霞[1] 张敏[1] 

机构地区：[1]广东药学院,广东广州510310 [2]广东省万年青制药有限公司,广东汕头510021

出　　处：《时珍国医国药》2009年第3期553-555,共3页Lishizhen Medicine and Materia Medica Research

基　　金：广东省中医药局2006年科研立项课题(No.2060136)

摘　　要：目的建立高效准确的穿心莲样品识别模型,为进行质量控制提供参考。方法收集不同产地的12个穿心莲药材样品的指纹图谱,提取4个主成分利用LM-BP神经网络进行模式识别。结果建立了穿心莲药材指纹图谱的LM-BP神经网络模型,经过对不同产地穿心莲的识别,证明其有较好的识别功能。结论LM-BP算法在识别速度和精度上都比传统BP算法有了较大提高。Objective To build an effective and exact reorganization and classification model for the quality control of Andrographis paniculata Nees. Methods 12 samples of Andrographis paniculata Nees collected from different places were determined by HPLC and four principal components were selected into the LM - BP model. Results The fast and specific model was established. The classifier could identify different samples with a high accuracy. Conclusion The BP neural network improved by LM algorithm can greatly increase the speed and the accuracy of the Andrographis paniculata Nee classification as compared with the ordinary BP neural network.

关 键 词：LM—BP神经网络 穿心莲 指纹图谱 

分 类 号：R282.5[医药卫生—中药学]