食用植物油中重金属元素的激光诱导击穿光谱（LIBS）检测研究

发布时间：2018-10-30 13:05

【摘要】：近几年,农产品安全问题越来越备受人们的关注,重金属污染是其中之一,重金属进入人体后不易排出,长期积累会严重影响人们的身体健康。食用植物油是人们的日常用品,且每日消耗量巨大,因此,加强食用油中重金属的监测和检测具有非常重大的意义。激光诱导击穿光谱(LIBS)技术因具有快速、实时、在线、多种元素同时检测等优点,本文利用LIBS技术研究植物油中的重金属铅和铬的光谱特性,并结合不同的分析方法对其进行定量分析,主要研究内容及结果如下:应用单、双脉冲LIBS技术对大豆油中的Pb元素进行检测,其中铅含量范围在17.63~770.73 ppm。研究中采用木片作为Pb的富集基体,经LIBS实验后,比较了单、双脉冲LIBS技术的性能。结果表明,双脉冲LIBS技术的综合性能(谱线强度、稳定性、灵敏性、检测限)优于单脉冲;然后,使用双脉冲采集的光谱数据对大豆油中的Pb元素含量进行定量分析。结果表明,采用多元线性回归模型得到的建模样本平均相对误差为7.25%,三个验证样品的预测相对误差分别为5.05%、5.75%和0.13%。应用共轴双脉冲LIBS技术对三种植物油中的重金属Cr进行定量分析,其中Cr含量范围在11.98~139.83 ppm。首先,采用CN分子线、Ca原子线及(CN+Ca)谱线强度之和为内标线建立Cr含量的定标模型。结果表明,采用双谱线内标法效果最佳,大豆油、花生油和玉米油建立的定标曲线拟合度R2分别为0.995、0.992和0.996,两个验证样品的平均预测相对误差分别为7.27%、7.62%和6.46%;其次,分别对两种不同品牌的混合大豆油、花生油、玉米油进行建模预测,利用三条Cr的原子谱线和两条基体元素谱线(CN分子线和Ca原子线)分别建立Cr的三变量和五变量的最小二乘支持向量机(LS-SVM)定标模型,并对其进行预测。结果表明,引入基体元素的五变量LS-SVM模型性能更佳,验证样品的平均预测相对误差分别为7.46%、8.96%和8.95%。研究表明,LIBS技术检测植物油中的Pb和Cr元素具有一定的可行性,LIBS技术结合多元线性回归法、双谱线内标法和LS-SVM法能有效降低实验参数和基体效应对分析元素的干扰,从而提高定标曲线的拟合度以及减小定量分析误差。
[Abstract]:In recent years, people pay more and more attention to the safety of agricultural products. Heavy metal pollution is one of them, heavy metals are difficult to be discharged after entering human body, long-term accumulation will seriously affect the health of people. Edible vegetable oil is the daily necessities of people, and the daily consumption is huge. Therefore, it is of great significance to strengthen the monitoring and detection of heavy metals in edible oils. Laser induced breakdown spectroscopy (LIBS) has the advantages of fast, real-time, on-line and simultaneous detection of many elements. In this paper, the spectral characteristics of heavy metals lead and chromium in vegetable oil are studied by LIBS. The main contents and results are as follows: single and double pulse LIBS were used to detect Pb elements in soybean oil, and lead content was in the range of 17.63 ~ 770.73 ppm.. In this study, wood chips were used as the enriched matrix of Pb. After LIBS experiments, the properties of single and double pulse LIBS were compared. The results showed that the comprehensive performance (spectral intensity, stability, sensitivity, detection limit) of double pulse LIBS was better than that of single pulse, and then the content of Pb elements in soybean oil was quantitatively analyzed by using the spectral data collected by double pulse. The results show that the average relative error of the modeling sample is 7.25 and the prediction relative error of the three validated samples is 5.05 5.75% and 0.13% respectively. The quantitative analysis of heavy metal Cr in three kinds of vegetable oils by coaxial bipulse LIBS technique was carried out. The range of Cr content was 11.98 ~ 139.83 ppm.. Firstly, the calibration model of Cr content is established by using the sum of the intensity of CN molecular line, Ca atomic line and (CN Ca) line. The results showed that the calibration curve fitted to soybean oil, peanut oil and corn oil was 0.992 and 0.996 for soybean oil, peanut oil and corn oil, respectively, and the average prediction error was 7.27, respectively. 7.62% and 6.46%; Secondly, two different brands of mixed soybean oil, peanut oil and corn oil were modeled and predicted. Three atomic lines of Cr and two lines of matrix elements (CN molecular line and Ca atomic line) were used to establish the three-variable and five-variable least squares support vector machine (LS-SVM) calibration models of Cr, respectively, and to predict them. The results show that the five-variable LS-SVM model with matrix elements has better performance and the average relative error of prediction is 7.46% 8.96% and 8.95% respectively. The results show that LIBS technique is feasible for the detection of Pb and Cr elements in vegetable oil. LIBS technique combined with multivariate linear regression method, double spectral line internal standard method and LS-SVM method can effectively reduce the interference of experimental parameters and matrix effect on the analysis elements. Thus, the fitting degree of calibration curve is improved and the error of quantitative analysis is reduced.
【学位授予单位】：江西农业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TS227;O657.31

【参考文献】