为研究激光诱导击穿光谱检测水果中重金属元素的应用，将激光诱导击穿光谱技术和化学计量学相结合分析脐橙中铜元素的含量。通过偏最小二乘法（PLS）、区间偏最小二乘法（iPLS）、联合区间偏最小二乘法（si－PLS）优化建模区域，建立了经过标准正态变换（ SNV）校正后光谱的铜含量分析模型。实验结果表明，后两种改进的偏最小二乘法建立的预测效果模型明显优于全波长（320～340 nm） PLS模型，并且当采用siPLS将光谱划分为25个子区间划分，选择其中5、14、16、22四个子区间时建立的siPLS模型效果最佳，其校正集相关系数r和交互验证误差（RMSEC）分别为0．9883和5．61μg／g，预测集相关系数r和预测均方根误差（RMSEP）分别为0．9792和8．62μg／g。研究为进一步实现水果中痕量重金属元素的快速定量分析提供了方法和数据参考。%In order to study laser induced breakdown spectroscopy ( LIBS ) detection of heavy metals in fruit,LIBS combined with chemometrics methods was applied to predict the copper content in navel orange. These partial least squares ( PLS ) , interval partial least squares ( iPLS ) and synergy interval partial least squares( siPLS) were used to find the most informative ranges,and build models with predictive quality based on standard normal variate(SNV) correction spectra.The results showed that the models built by the iPLS and siPLS methods had higher predictive ability than that of PLS models which was developed on the whole wave-length rang 320-340nm.The optimal model was obtained by siPLS that separated the whole spectra into 25 in-tervals and combined four intervals:[ 5, 14, 16, 22 ] .The model was achieved with correlation coefficient 0.988 3 and root mean square error of cross(RMSEC) 5.61μg/g in calibration set and correlation coefficient 0.979 2 and root mean squared error of prediction RMSEP 8.62μg/g in prediction set.The siPLS model pro-vides method and data reference for quantitative analysis of trace heavy metal elements in fruit by LIBS.
展开▼
