---

作 者 ：邵平， 王钧， 王星丽， 瞿亮， 孙培龙

期 刊 ：核农学报 2016年 30卷 1期 页码：103-109

关键词：掺假；灵芝提取物；近红外；偏最小二乘法；配对T检验；

Keywords:adulteration, Ganoderma lucidum extraction, near infrared spectroscopy, partial least squares, paired sample T test,

摘 要 ：为了探索快速测定灵芝提取物掺假(淀粉)含量的方法,采用近红外光谱扫描掺杂0%、5%、10%、20%、40%、60%、80%淀粉的灵芝提取物,对其光谱进行预处理和波段选择,并结合偏最小二乘法(PLS)建立灵芝提取物掺假定量快速无损检测方法。结果表明,使用多元散射校正(MSC)预处理方法,波数范围8 000~7 500、6 000~5 500和5 000~4 000cm^-1,主因子数为8时,建立的偏最小二乘法模型的校正决定系数(R_cal)、校正均方根差(RMSECV)、验证决定系数(R_val)和验证均方根差(RMSEP)分别为0.9962、0.0249、0.9960和0.0241。运用该模型对验证集样品进行预测并统计分析,可知预测值与实际掺假值之间无显著差异。本研究为灵芝功能食品市场检测提供了方法基础。

Abstract:To explore a rapid prototype to determine Ganoderma lucidum extraction (GLE) adulterated with starches, a correlative model consisting of near infrared spectrum and the contents of adulteration were established. The contents of adulteration in GLE samples were predicted using near infrared spectroscopy (NIR) combined with partial least squares (PLS). After scanning starch content 5%, 10%, 20%, 40%, 60% and 80% of the GLE samples and choice of the wave number and the pretreatment of the spectral, the model was established by PLS. The results of cross validation showed that with the wavelengths of 8 000 to 7 500 cm^-1, 6 000 to 5 500 cm^-1, 5 000 to 4 000 cm^-1, the pretreatment of MSC and 8 factors, the calibration coefficient of determination (R_cal), root mean standard error cross validation (RMSECV), validation coefficient of determination (R_val), and root mean standard error cross prediction (RMSEP) of the model were 0.9962, 0.0249, 0.9960, and 0.0241, respectively. The model was used to verify samples and the statistical results showed that there is no significant difference between the predictive and actual values. It provides the basis for Ganoderma lucidum extraction functional food.

---