摘要
为了建立一种简便有效的花生油掺伪的定性和定量鉴别方法,采集花生油中分别掺伪0~90%大豆油、棕榈油和棉籽油样品的可见—近红外光谱图,结合主成分分析、判别分析、改进偏最小二乘法,建立花生油掺伪的定性鉴别和定量预测模型。结果表明,在定性鉴别中,对花生油中分别掺入大豆油、棕榈油和棉籽油的整体正确判别率分别达到了100%、96.1%和85.3%。在定量分析中,对MPLS法建立的花生油二元掺伪定标模型进行验证,结果表明,掺入大豆油、棉籽油和棕榈油的预测相关系数R_p^2分别为0.998、0.997和0.995,相对标准差RSD分别为2.33%、3.04%和3.83%,相对分析误差RPD分别为3.542、2.642和2.581,说明这三种掺假花生油所建立的最优定标模型的预测精度高,其中花生油中掺入大豆油的预测精度最高,检测花生油中掺入棉籽油与棕榈油的最低掺假量为3%。为花生油二元掺伪模式提供了一种简便、快速、有效的分析方法。
Visible - Near infrared spectra of peanut oil adulterated respectively with soybean oil, palm oil and cottonseed oil in different proportion （V/V） from 0 to 90% were collected and analyzed to seek an effective and simple method for qualitative and quantitative detection of peanut oil adulteration. The result showed that in the qualitative identification the correct rate of the peanut oil mixed with soybean oil, palm oil and cottonseed oil was 100%, 96.1% and 85.3%, respectively. In the quantitative analysis, the peanut oil binary adulteration calibration model established using modified partial least square （MPLS） was validated. The result showed that the correlation coefficients of cross validation for three kinds of models were 0. 998, 0. 997 and 0. 995 respectively; the relative standard deviation were 2. 327%, 3. 040% and 3. 830%, respectively; the relative percent deviation were 3. 542, 2. 642 and 2.581, respectively. The results indicated that NIR technique can be used as an effective method for quality control and adulteration identification of peanut oil, the prediction accuracy of soybean oil adulteration was high- est, and the adulteration content of palm oil and cottonseed oil above 3% can be accurately predicted by these models. This result can supply a simple, rapid and effective method for identifying the adulterated binary system of peanut oil.
出处
《粮油食品科技》
2015年第6期84-88,共5页
Science and Technology of Cereals,Oils and Foods
基金
河南省教育厅重点项目资助(15A550009)
郑州市科技攻关项目资助(20140769)
关键词
可见-近红外光谱
花生油掺伪
定性鉴别
定量分析
visible -near infrared spectroscopy
adulterate peanut oil
qualitative identification
quantitative analysis
