基于LF-NMR及判别分析的葵花籽油过氧化值定性分析

Qualitative Investigation of Sunflower Oil Peroxide Value Based on LF-NMR and Discriminant Analysis

为实现利用低场核磁共振(LF-NMR)技术快速检测葵花籽油过氧化值是否超标,收集了乐惠(LH)、金龙鱼(JLY)、福临门(FLM)、日清(RQ)、海狮(HS)、葵王(KW)、多力(DL)和融氏(RS)8品牌,共计162个不同氧化程度的葵花籽油LF-NMR弛豫特性(T21、S21、T22、S22、T23、S23、T2W),其中过氧化值超标样品数为74,并结合判别分析建立了葵花籽油过氧化值是否超标的定性判别模型。结果表明,葵花籽油LF-NMR弛豫信号随氧化时间变化存在一定的规律性,借助于判别分析进一步提取有效信息,发现模型总体判别准确率可达85.8%,其中RQ和RS葵花籽油判别准确率均达到90%以上,除HS仅为77.27%外,其余品牌判别准确率均在80%以上,判别效果良好。说明可以利用葵花籽油的LF-NMR弛豫特性对其过氧化值进行快速有效的定性分析,省时省力,具有重要实践意义。

A rapid, qualitative method was developed to determine whether the peroxide value of sunflower oil exceeded the limit using low-field nuclear magnetic resonance(LF-NMR). One hundred and sixty-two samples oxidized to different degrees, including 74 oil samples exceeding the limit, were collected from 8 brands(LH, JLY, FLM, RQ, HS, KW, DL, and RS) and were studied. A direct discrimination method was used to determine the peroxide values using LF-NMR relaxation characteristics(T21, S21, T22, S22, T23, S23, and T2W) as independent variables. The results indicated that consistency could be found between the relaxation characteristics and oxidation time. The overall accuracy of the discrimination model was 85.8%, among which those of RQ and RS sunflower oil were above 90%, while the rest were above 80%, except for that of HS(77.27%)。These results indicate that there was a good discriminatory effect. Therefore, it is feasible to rapidly assess peroxide values using LF-NMR relaxation characteristics of sunflower oil, which is a time and energy-saving method and has important practical implications.