基于多重二阶差商滤波的矿物油检测自动判别方法  

Automatic Discrimination Method for Detection of Mineral Oil Based on Multiple Second-order Difference Quotient Filtering

作　　者：任娟 李冰宁 刘玲玲 陈婷 刘清珺[1] 武彦文 REN Juan;LI Bing-Ning;LIU Ling-Ling;CHEN Ting;LIU Qing-Jun;WU Yan-Wen(Institute of Analysis and Testing,Beijing Academy of Science and Technology(Beijing Center for Physical and Chemical Analysis),Beijing 100089,China)

机构地区：[1]北京市科学技术研究院分析测试研究所(北京市理化分析测试中心),北京100089

出　　处：《分析化学》2025年第1期104-114,共11页Chinese Journal of Analytical Chemistry

基　　金：北京市科学技术研究院创新工程项目(No.24CA001-04)资助。

摘　　要：由饱和烃(MOSH)和芳香烃(MOAH)组成的矿物油污染物在食用油和相关加工食品中普遍存在。目前,矿物油分析主要采用液相-气相色谱联用法(LC-GC-FID)净化除去干扰物并将MOSH和MOAH分离,通过接口技术分别将MOSH或MOAH转入GC的不同通道进行定量分析。MOSH和MOAH谱图均呈不规则的驼峰形状,其上方的尖峰为天然烃类干扰物,通常不会影响驼峰轮廓的判别。然而,当干扰物净化不完全时,会在驼峰上方形成一个或数个豁口,干扰驼峰轮廓的准确判断与勾勒,导致矿物油分析结果的重现性较差。本研究模仿人工绘制驼峰形状或轮廓的思路,提出了一种自动判定矿物油驼峰轮廓(即下包络线)的算法,采用多重二阶差商滤波的方法实现了对驼峰上方豁口的鉴定与去除。首先搜索并确定矿物油驼峰的最低值,即谷值点序列,然后对谷值点序列进行二阶差商滤波。与驼峰相比,尖峰的二阶差商为一个较大的负值,通过滤除谷值点序列中二阶差商(或多重二阶差商)较大的负值点,即可去除驼峰上方的尖峰。分别采用自动算法和人工方法检测了42个不同种类的食用油和奶粉等样品,结果表明,两种方法检测出的矿物油含量结果并无显著性差异。Mineral oil contaminants composed of saturated hydrocarbons(MOSH)and aromatic hydrocarbons(MOAH)are commonly found in edible oils and related processed foods.Currently,the analysis of mineral oils primarily employs the liquid chromatography-gas chromatography-flame ionization detector(LC-GC-FID)method.Liquid chromatography is used to purify and separate MOSH and MOAH from interfering substances,and the interface technology transfers MOSH or MOAH into different GC channels for quantitative analysis.The MOSH and MOAH chromatograms typically exhibit an irregular hump shape,with sharp peaks above the hump representing natural hydrocarbon interferences,which usually do not affect the identification of the hump profile.However,when the purification of interferences is incomplete,they can form one or more gaps above the hump,interfering with the accurate judgment and delineation of the hump profile,and leading to poor reproducibility of analysis results of mineral oil.In this study,an algorithm that mimicked the manual drawing of the hump shape or contour was proposed for automatically determining the mineral oil hump contour(i.e.,the lower envelope line).The algorithm used a multiple second-order difference quotient filtering method to identify and remove the gaps above the hump.The method involved first searching and determining the lowest value of the mineral oil hump,which was the valley point sequence,and then applying second-order difference quotient filtering to the valley point sequence.Compared to the hump,the second-order difference quotient of sharp peaks was a significantly larger negative value.By filtering out the points in the valley point sequence with larger negative second-order difference quotients(or multiple second-order difference quotients),the sharp peaks above the hump were removed.To verify the accuracy of the algorithm,42 different types of samples,including edible oils and milk powders were analyzed,using both the automatic algorithm and manual methods.

关 键 词：矿物油测定 液相色谱-气相色谱联用法 鼓包峰轮廓 二阶差商 滤波 

分 类 号：TS207.3[轻工技术与工程—食品科学] O657[轻工技术与工程—食品科学与工程]