Automatic Discrimination Method for Detection of Mineral Oil Based on Multiple Second-order Difference Quotient Filtering
10.19756/j.issn.0253-3820.241321
- VernacularTitle:基于多重二阶差商滤波的矿物油检测自动判别方法
- Author:
Juan REN
1
;
Bing-Ning LI
;
Ling-Ling LIU
;
Ting CHEN
;
Qing-Jun LIU
;
Yan-Wen WU
Author Information
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
- Keywords:
Mineral oil detection;
Liquid chromatography-gas chromatography;
Swelling peak contour;
Second-order difference quotient;
Filtering
- From:
Chinese Journal of Analytical Chemistry
2025;53(1):104-114
- CountryChina
- Language:Chinese
-
Abstract:
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.The results showed that there were no significant differences in the detected mineral oil contents between these two methods.
