摘要
为使中药材种类及产地鉴别结果更精准,研究了当数据量充足无缺失,且药材中红外或近红外光谱特征明显和当数据量较少,中红外和近红外光谱数据差异明显,数据类别标签较多,且存在数据缺失两种情况下中药材种类及产地鉴别方法.前者可通过主成分分析降维,再经过K-Means聚类鉴别其结果;或通过人工神经网络在分类后给出需要预测的中药材种类及产地鉴别结果;后者可对中药材的近红外和中红外光谱数据进行图示对比分析,同时结合K近邻算法数据分析,进行分类,通过相互验证方式鉴别其结果.
In order to make the identification results of the types and origins of traditional Chinese medicine more accurate,the identification methods of the types and origins of traditional Chinese medicine were studied when the amount of data was sufficient and there was no missing,and the characteristics of the mid-infrared or near-infrared spectra of the medicine were obvious,and when the amount of data was small,the differences between the mid-infrared and near-infrared spectra were obvious,and there were many data category labels,and there were data missing.The former can be reduced by principal component analysis,and then identified by K-Means clustering;Or the artificial neural networks can be used to give the identification results of the type and origin of Chinese medicinal materials to be predicted after classification;The latter can carry out graphic comparative analysis on the near-infrared and mid-infrared spectral data of Chinese medicinal materials,and combine K-Nearest Neighbor algorithn data analysis to classify them separately,and identify their results through mutual verification.
作者
张晓丽
ZHANG Xiaoli(Mathematics and Computer Science,Yuncheng Advanced Normal College,Yuncheng Shanxi 044000)
出处
《宁夏师范学院学报》
2023年第1期43-49,共7页
Journal of Ningxia Normal University
基金
山西省教育科学“十三五”课题(GH-19292).
关键词
主成分分析
K-MEANS聚类
KNN近邻算法
Principal component analysis
K-Means clustering
KNN nearest neighbor algorithm
