激光诱导击穿光谱的飞灰碳含量定量分析方法

燃煤飞灰碳含量是影响锅炉工作效率的重要特性指标之一，文中开展激光诱导击穿光谱技术（LIBS）实现飞灰未燃碳的定量分析方法研究，为LIBS应用于飞灰含碳量的快速/在线检测奠定基础。根据所探测的LIBS特征光谱，将线性和非线性化学计量学方法，包括多元线性回归（MLR）和偏最小二乘回归（PLSR）线性分析分析方法，以及非线性的极限学习机（ELM）和支持向量机回归（SVR）模型应用于飞灰未燃碳的预测分析中，结合交叉验证法对模型进行验证。对比线性和非线性模型的结果可以看出，非线性模型的预测结果明显优于线性模型，其中采用基于K-CV参数优化的非线性SVR模型具有比较理想的分析结果，有助于提高飞灰碳含量分析的精确度和准确度，采用三折叠交叉验证法对模型进行验证，得到模型的决定系数R2均为0.99，相对偏差的平均值ARD分别为1.54%、3.45%、3.51%，相对标准误差RSD的平均值分别为7.53%、2.89%、7.18%。

Quantitative analysis method of unburned carbon content of fly ash by laser-induced breakdown spectroscopy

The unburned carbon content of fly ash is an important index for the working efficiency of the coal-fired boiler. In this work, laser-induced breakdown spectroscopy(LIBS) was applied to realize the quantitative analysis of unburned carbon in fly ash. Based on the detection of LIBS characteristic spectrum, the common chemometrics methods include linear model, such as multiple linear regression(MLR), partial least-squares regression(PLSR) and nonlinear model, such as extreme learning machine(ELM) model, support vector machine regression(SVR) model were proposed to the prediction analysis of unburned carbon in fly ash, and the cross-validation method was used to verify the model. The results show that the prediction results from nonlinear models are better than that of linear models, among which the SVR model based on K-CV parameter optimization is helpful to improve the prediction accuracy and accuracy of the content of unburned carbon in the fly ash. Based on the three-fold cross validation method, the R2 of the model is 0.99, ARD is 1.54%, 3.45% and 3.51%, and the average value of RSD is 7.53%, 2.89%, 7.18%, respectively.