基于FA-XGBoost算法的拉曼光谱预测沙棘籽油含量  

Prediction ofSeabuckthorn Seed Oil Content by Raman Spectroscopy based on FA-XGBoost Algorithm

作　　者：王照[1] 刘斌[2] 吴军 WANG Zhao;LIU Bin;WU jun(Henan vocational college of nursing,Anyang 455000,China;Key Laboratory of Smart Agriculture Systems,Ministry of Education,China Agricultural University,Beijing 100083,China;Chinese Academy of Agricultural Mechanization Sciences Group Co,Ltd.Beijing 100083,China)

机构地区：[1]河南护理职业学院,河南安阳455000 [2]中国农业大学智慧农业系统集成研究教育部重点实验室,北京100083 [3]中国农业机械化科学研究院集团有限公司,北京100083

出　　处：《光散射学报》2025年第1期77-85,共9页The Journal of Light Scattering

基　　金：国家自然科学基金(32072357);豫教〔2023〕03010河南省职业教育教学改革研究与实践项目。

摘　　要：沙棘籽油具有很高的营养价值和药用价值,高额利润驱使不法商家以次充好、以假乱真,目前缺少沙棘籽油掺假鉴别的方法学研究。因此,本文提出一种基于FA-XGBoost算法赋能拉曼光谱技术实现沙棘籽油掺假检测的技术方法。采用葵花籽油按不同体积分数掺假沙棘籽油,使用近红外显微拉曼光谱仪测量所有样本的拉曼光谱,定性分析了沙棘籽油与葵花籽油的拉曼光谱特征峰。通过FA-XGBoost算法构建了针对葵花籽油掺假沙棘籽油的回归模型,使用测试集评价了FA-XGBoost模型的预测效果,其中决定系数为0.9798,均方误差为0.0031。总之,本文提出了一种面向沙棘籽油掺假的定量检测方法,通过FA-XGBoost算法与拉曼光谱的结合实现了葵花籽油掺假沙棘籽油的浓度预测,该方法对规范国内沙棘籽油消费市场具有潜在的应用价值和现实意义。Seabuckthorn seed oil has high nutritional and medicinal value,and high profits drive illegal businesses to pass shoddy goods off as good,confusing the false with the true.More methodological research is needed to identify seabuckthorn seed oil adulteration.Therefore,this paper proposes a method to detect the adulteration of seabuckthorn seed oil based on the FA-XGBoost algorithm,which enables Raman spectroscopy.Seabuckthorn seed oil was adulterated with sunflower seed oil in different volume fractions.The Raman spectra of all samples were measured by a near-infrared micro Raman spectrometer.The Raman spectra of seabuckthorn seed oil and sunflower seed oil were qualitatively analyzed.A regression model for adulterating sunflower seed oil with seabuckthorn seed oil was constructed using the FA-XGBoost algorithm.The predictive performance of the FA-XGBoost model was evaluated using a test set with a determination coefficient of 0.9959 and a mean squared error of 0.0031.This paper proposes a quantitative detection method for seabuckthorn seed oil adulteration.The concentration prediction of sunflower seed oil adulterated with seabuckthorn seed oil is realized through the combination of the FA-XGBoost algorithm and Raman spectroscopy.This method has potential application value and practical significance for regulating the domestic seabuckthorn seed oil consumption market.

关 键 词：拉曼光谱 因子分析 XGBoost 沙棘籽油 定量分析 

分 类 号：O433[机械工程—光学工程] O657.3[理学—光学]