论文部分内容阅读
采用LIBS技术与火焰原子吸收法(AAS),获取23个浓度梯度的含Pb元素脐橙样品的LIBS光谱及Pb元素真实浓度信息,再对LIBS谱线信息进行数据预处理,建立PLS定量分析模型。当采用9点平滑结合SNV作为预处理方法时,PLS模型最佳,其校正集相关系数(R_t)、交叉验证均方根误(RMSECV)、预测集相关系数(R_p)、预测均方根误差(RMSEP)分别为0.9633,1.56,0.9542和2.58,脐橙中Pb元素预测结果的平均相对误差为6.9%。与小组前期对脐橙中Pb元素单变量和多元定标法相比,LIBS结合PLS建模时提高对脐橙微量重金属检测的准确性。
LIBS spectra and Pb element true concentration information of 23 concentrations gradient Pb-bearing navel orange samples were obtained by LAS method and flame atomic absorption spectrometry (AAS). The LIBS spectrum information was preprocessed to establish PLS quantitative analysis model. The PLS model was the best when the 9-point smoothing combined with SNV was used as the preprocessing method. The correlation coefficient (R_t), cross-validation root mean square error (RMSECV), prediction set correlation coefficient (R_p), prediction root mean square error (RMSEP) were 0.9633, 1.56, 0.9542 and 2.58, respectively. The average relative error of Pb predictions in navel orange was 6.9%. Compared with the previous univariate and multivariate calibration of Pb in navel orange, LIBS combined with PLS modeling can improve the accuracy of detecting trace heavy metals in navel orange.