大豆油中铬元素含量的激光诱导击穿光谱检测

发布时间：2018-02-16 11:19

本文关键词： 光谱学激光诱导击穿光谱大豆油铬含量最小二乘支持向量机　出处：《激光与光电子学进展》2016年04期 　论文类型：期刊论文

【摘要】：采用激光诱导击穿光谱(LIBS)技术对大豆油中的铬(Cr)含量进行检测研究。以一系列Cr含量不同的大豆油为样本,采用Ava Spec双通道高精度光谱仪在206.28~481.77 nm波段范围内采集LIBS光谱。根据样本的LIBS谱线图,确定Cr元素的主要特征谱线,并对Cr元素主要特征谱线应用线性回归或最小二乘支持向量机(LS-SVM)方法建立其单变量、二变量及多变量校正模型。利用建立的校正模型对样本Cr含量进行预测。研究结果表明,二变量及多变量校正模型的性能优于单变量校正模型,LS-SVM建立的多变量校正模型性能最优。对于单变量及二变量校正模型,预测样本的平均相对误差(RE)分别为14.16%和11.58%;而对于线性回归及LS-SVM建立的多变量校正模型,预测样本的平均RE分别为10.95%和4.97%。由此可见,LIBS技术检测大豆油中的重金属Cr含量具有一定的可行性,LS-SVM方法可以有效提高校正模型的预测精度。
[Abstract]:The content of Cr in soybean oil was determined by laser induced breakdown spectroscopy (LIBS). A series of soybean oils with different Cr content were used as samples. The Ava Spec double channel high precision spectrometer is used to collect the LIBS spectrum in the range of 206.28 ~ 481.77 nm. The main characteristic lines of Cr are determined according to the LIBS spectrum of the sample. The linear regression or least squares support vector machine (LS-SVM) method is used to establish the single variable, two variable and multivariable calibration models for the main characteristic lines of Cr elements. The calibration model is used to predict the Cr content of the samples. The performance of the two-variable and multivariable correction models is better than that of the single-variable correction model / LS-SVM. The average relative error of the predicted sample was 14.16% and 11.58, respectively, while for the linear regression and the multivariable correction model established by LS-SVM, the mean relative error of the predicted sample was 14.16% and 11.58, respectively. The average RE values of the predicted samples are 10.95% and 4.97, respectively. It can be seen that the LIBS technique is feasible for the detection of heavy metal Cr in soybean oil. LS-SVM method can effectively improve the prediction accuracy of the calibration model.
【作者单位】：江西农业大学生物光电技术及应用重点实验室;江西出入境检验检疫局综合技术中心;
【基金】：国家自然科学基金青年基金(31401278) 江西省自然科学基金(20132BAB214010)
【分类号】：TS225.13;TN249

【相似文献】