摘要
基于近红外光谱技术,建立了不同产地茯苓块快速无损鉴别方法。利用近红外光谱仪采集了8个不同产地茯苓块的光谱信息;通过单一及组合预处理方法消除光谱中的多种干扰;结合主成分分析方法、软独立模式分类法和Fisher线性判别分析方法分别构建了不同产地茯苓块的鉴别模型。结果表明:光谱中存在较为明显的背景以及噪声干扰;仅采用主成分分析结合光谱预处理的方法无法实现不同产地茯苓块的准确鉴别分析,鉴别率仅为14.1%;采用软独立模式分类法可显著提高不同产地茯苓块的鉴别率,采用原始光谱或去趋势预处理可获得最佳鉴别结果,鉴别率为54.2%;采用Fisher线性判别分析方法时,用原始光谱即可得到最佳鉴别结果,鉴别率为91.7%。以上结果表明,近红外光谱技术结合Fisher判别分析方法可实现不同产地茯苓块的准确鉴别分析。
A fast and nondestructive identification of Poria cocos blocks from different origins based on near infrared spectroscopy technology was proposed. The spectra of Poria cocos blocks from eight origins were collected by near infrared spectroscopy. Single and combined preprocessing methods were used to eliminate various interferences in the spectra. Combined with principal component analysis, soft independent pattern classification and Fisher linear discriminant analysis methods, the identification models were obtained. The results showed that there were obvious background and noise interferences in the spectra. Accurate identification cannot be achieved with principal component analysis combined with spectral pretreatment method, and the best identification accuracy was 14.1%. Soft independent pattern classification method could significantly improve the result. The best identification result could be obtained by using the original spectra or de-trended pretreatment, with the identification accuracy of 54.2%. With Fisher linear discriminant analysis method, the best result can be obtained by using the original spectra, and the best identification accuracy was 91.7%. The above results showed that near infrared spectroscopy combined with Fisher linear discriminant analysis method can achieve accurate identification analysis of Poria cocos block from different origins.
作者
李嘉仪
余梅
郑郁
李跑
LI Jiayi;YU Mei;ZHENG Yu;LI Pao(College of Food Science and Technology,Hunan Agricultural University,Changsha 410128;School of Medicine,Hunan Normal University,Changsha 410006;Hunan Agricultural Product Processing Institute,Hunan Academy of Agricultural Sciences,Changsha 410125)
出处
《分析试验室》
CAS
CSCD
北大核心
2021年第12期1381-1386,共6页
Chinese Journal of Analysis Laboratory
基金
国家自然科学基金(31601551,31671931)
湖南省自然科学基金青年科学基金(2019JJ50240)
湖南省教育厅科学研究项目优秀青年项目(18B118)
中国博士后科学基金面上项目(2019M650187)资助。
