均值离差平方和在近红外光谱识别灵芝和云芝提取物的应用

摘要

To develop an automatic and quick method to discriminate between the extractions of Ganoderma lu⁃cidum and versicolor by near infrared reflectance (NIR) spectroscopy, the optimum pretreatment method was de⁃termined by the correct rate and the average of sum of different square value. And the NIR spectral data of the extractions of Ganoderma lucidum and versicolor samples were analyzed with the Distance Match model. The re⁃sults showed that in the full wavelengths, with the pretreatment of FD, the total correct rate was 90.79% in the calibration set, and that in the validation set, the total correct rate was 100%,which has a good application. In conclusion, according to the pretreatment of FD, it is feasible to apply NIR to discriminate the extractions of Ganoderma lucidum and versicolor.%为实现灵芝提取物和云芝提取物的自动化快速鉴别，利用近红外光谱仪对灵芝提取物和云芝提取物进行近红外光谱分析，根据正确率和均值离差平方和（Average of Sum of Differ⁃ence Square，ASDS）确定最佳预处理方法，建立距离匹配（Distance Match，DM）判别分析模型.结果表明：在全波长范围内，采用FD和MSC+FD预处理，在建模集中，对样品的识别率达到90.79%，模型预测效果好；在外部验证中，验证的识别率达到100%，具有很强的应用性.上述结果表明：利用近红外光谱和均值离差平方和，得出经过一阶导数处理的光谱，在DM模型对灵芝和云芝提取物分类中是可行的.