摘要
为了满足食用菌提取物实际生产监管需要,本研究采用近红外漫反射光谱技术对来自不同地区的灵芝和云芝提取物样品进行定性识别研究。在800~2 750nm波段范围,采集灵芝和云芝提取物的漫反射光谱,应用主成分聚类分析和偏最小二乘判别法分别建立识别模型,用146个样品进行建模和48个外部样品集进行验证。结果表明：采用主成分聚类判别分析法,灵芝和云芝提取物的判别界线清晰,正确率达到88.54%;采用偏最小二乘判别法,建立的鉴别分类模型能较好地对灵芝和云芝提取物进行鉴别,校正集和预测集样品的识别正确率均为100%。因此,近红外结合主成分聚类分析和偏最小二乘判别法识别灵芝和云芝提取物是可行的,同时研究结果为灵芝和云芝提取物的快速识别提供了理论依据和使用方法。
The feasibility of using near infrared（ NIR） spectroscopy and multivariate analysis as tools to identify the extracts of Ganoderma lucidum and Versicolor was studied for the purpose of supervising the extracts. In the wavelength range of 800 to 12 500 nm,a set of pure and real extracts of Ganoderma lucidum and Versicolor collected from different companies were used as the representative training set of authentic objects. Indentification models based on the NIR spectra were developed using principal component analysis（ PCA） and partial least squares（ PLS）. 146 samples were used to build the model,48 external samples were used to test the model. By the PCA model,the results showed that excellent classification could be obtained after optimizing spectral pre-treatment with the correct rate of 87. 5%. By the PLC model,the correct rate was100%. The results indicated that NIR combined with classification techniques could be a suitable technology for the classification of the extracts of Ganoderma lucidum and Versicolor,at the same time,the investigation provides the theoretical support and practical method for rapid identification of the extracts of Ganoderma lucidum and Versicolor.
出处
《核农学报》
CAS
CSCD
北大核心
2015年第3期499-505,共7页
Journal of Nuclear Agricultural Sciences
基金
科技部国际科技合作项目(2014DFA31530)
