基于THz-TDS技术定量分析煤炭中的灰分和挥发分

Quantitative analysis of ash and volatile in coal based on THz-TDS

摘要:

为提高煤炭中灰分、挥发分定量分析的效率，减少污染，提出一种基于太赫兹时域光谱(THz-TDS)技术的煤炭灰分、挥发分定量分析方法。研究煤炭太赫兹吸收特性与灰分的关系，结果表明，煤炭的太赫兹有效频谱范围为0.2~2.2 THz，在0.2~2.2 THz没有明显的吸收峰，灰分含量是影响太赫兹波段吸收特性的主要因素；对比预处理方法、支持向量机(SVM)模型和偏最小二乘(PLS)模型对预测效果的影响，采用样品数的3/4建立模型，其余1/4作为外部验证集进行验证，结果表明，Savitzky-Golay平滑和移动窗口平均(MAF)平滑对外部验证集的预测效果没有提升；SVM模型对灰分、挥发分的预测效果优于PLS模型。灰分、挥发分的SVM模型的外部验证集的相关系数(Rp)分别为0.933、0.724；预测均方根误差(RMSEP)分别为3.223、5.772。太赫兹技术结合SVM算法对煤炭灰分、挥发分进行预测，可将分析时间缩短至10 min内，提升分析效率。

Abstract:

In order to improve the efficiency of quantitative analysis of ash and volatile content in coal and reduce pollution, a method for quantitative analysis of coal ash and volatile content based on THz Time Domain Spectroscopy(THz-TDS) is proposed. In this paper, the relationship between terahertz absorption characteristics and ash content in coal is studied. The results show that the effective spectrum range of coal is 0.2~2.2 THz, and there is no obvious absorption peak in 0.2~2.2 THz; the ash content is the main factor affecting the absorption characteristics of the terahertz band. The effects of preprocessing methods, Support Vector Machine(SVM) model and Partial Least Squares(PLS) model on the prediction effect are compared. The results show that Savitzky-Golay smoothing and Moving Average Filter(MAF) smoothing do not improve the prediction effect of external validation sets, and the prediction effect of SVM model on ash and volatile content is better than that of PLS model. The correlation coefficients(R_p) and Root Mean Square Error of Prediction(RMSEP) of the external validation set of ash and volatile of SVM models are 0.933, 3.223, 0.724 and 5.772, respectively. The terahertz technology combined with SVM model can predict ash and volatile content in coal, and the analysis time can be shortened to be less than 10 minutes, which has improved the analysis efficiency.