基于改进的偏最小二乘法的LIBS钢液成分定量分析

摘要

In order to study the effect of laser induced breakdown spectroscopy ( LIBS ) matrix effect on measurement accuracy, the improved partial least squares ( PLS) was used to forecast the data and reduce the influence of matrix effect on the to-be-detected parameters. The comparative study with univariate calibration and partial least squares model calibration shows that the fitting degree of calibration curves of Mn and Si improve from 0. 991 and 0. 993 to 0. 996 and 0. 997, and the relative error of prediction for validation samples were decreased from 6. 231%, 6. 912% to 5. 510%, 6. 125%. The above results show that the improved PSL can reduce matrix effects and improve the calibration accuracy. The analysis of performance has been improved significantly.%为了研究激光诱导击穿光谱技术基体效应对其测量精确度的影响,采用改进的偏最小二乘法进行数据预测,以有效降低基体效应对待测元素的影响。通过与单变量定标和偏最小二乘模型定标进行对比研究可知,待测元素Mn和Si的定标曲线的拟合度分别从0.991和0.993提高到了0.996和0.997,对于验证样品的预测相对误差也分别从6.231%,6.912%下降到了5.510%和6.125%。结果表明,采用改进的偏最小二乘法可以有效减小基体效应,定标精度有较明显提高,对分析性能有了较大的提高。