激光诱导击穿光谱技术与偏最小二乘回归法在煤炭灰分检测中的应用

Application of partial least-squares regression method and LIBS in rapid detecting of ash content of coal

为了解决激光诱导击穿光谱技术(LIBS)的自吸收效应所导致的测量精度降低问题,将LIBS与偏最小二乘回归法(PLS)结合后建立分析模型以有效应用于煤炭灰分的检测。在传统的PLS模型基础上考虑谱线自吸收的影响,根据自吸收后的谱线强度与浓度之间的非线性关系,增加基于谱线强度的平方项并提出谱线自吸收效应的非线性PLS模型。经过自吸收修正后的PLS模型,预测样品的平均误差从1.696%降至1.504%,最大误差从5.780%降至3.507%,模型的预测性能和泛化能力得到了显著提升。

In order to solve the problem of precision reduction caused by self-absorption effect of LIBS,combining the LIBS with partial least squares regression(PLS)method,the analysis model established could be used in coal ash test effectively.Based on the conventional PLS model,the self-absorption of spectrum was considered,according to the non-linear relationship between the intensity and concentration of self-absorbed spectral lines,the square term based on spectral line intensity was added,the non-linear PLS model of self-absorption effect of spectrum was proposed.After self-absorption correction,the average error of predicted samples decreased to 1.504%from 1.696%,the maximum error of predicted samples decreased to 3.507%from 5.780%,the prediction performance and generalization ability of modal have been significantly improved.