论文部分内容阅读
为了提高激光诱导击穿光谱技术的检测能力和准确性,分别采用单变量分析和多变量分析方法[多元线性回归(MLR)]对样品中Cr进行定量分析。分别利用Cr I:425.435 nm和Cr I:427.48 nm两条特征谱线进行单变量分析,并在5种不同激光能量下获得了Cr的检测限,结果表明谱线Cr I:425.435 nm的分析结果要优于Cr I:427.48 nm,获得Cr最佳的检测限为5.8 mg/g。利用多变量分析方法,研究了浓度预测值与浓度实际值之间的线性相关性,与单变量分析方法相比,线性相关性由0.98提高到了0.99以上;采用留一交叉验证方法,比较了两种方法的预测相对误差(REP),单变量分析方法的REP分别为6.73%和7.59%,而MLR分析方法的REP则为4.66%,结果表明激光诱导击穿光谱技术结合MLR能够提高样品浓度预测的准确性。
In order to improve the detection ability and accuracy of laser-induced breakdown spectroscopy, univariate analysis and multivariate analysis [MLR (Multiple Linear Regression)] were used to quantitatively analyze Cr in samples. The univariate analysis of Cr I: 425.435 nm and Cr I: 427.48 nm, respectively, and the detection limits of Cr at 5 different laser energies were performed. The results show that the spectral line Cr I: 425.435 nm Better than Cr I: 427.48 nm, the best detection limit of Cr is 5.8 mg / g. The multivariate analysis method was used to study the linear correlation between the concentration predicted value and the actual concentration value. Compared with the univariate analysis method, the linear correlation coefficient was increased from 0.98 to 0.99. By using a cross validation method, (REP), REP of univariate analysis were 6.73% and 7.59%, respectively, while REP of MLR method was 4.66%. The results showed that laser induced breakdown spectroscopy combined with MLR can improve the prediction of sample concentration Accuracy