基于激光诱导击穿光谱(LIBS)的煤灰熔点快速检测

炉内结渣是影响火电机组和气化工艺可靠运行的关键因素之一，准确预测灰熔点可以提前调整炉膛出口温度以避免结渣。本论文采用激光诱导击穿光谱（LIBS）采集煤灰样中金属元素的光谱，分别建立煤灰中的金属元素的谱线强度与煤灰熔点的随机森林模型、支持向量机回归模型和线性回归模型，直接预测煤灰熔点温度。采用基于马氏距离(MD)的异常数据剔除算法和基于稀疏矩阵的基线估计与降噪算法（BEADS），对粉煤灰样的全光谱数据进行了预处理。随机森林模型对粉煤灰熔点的预测平均相对误差（MRE）为54.74%，支持向量机回归模型的预测平均相对误差为60.08%，而线性回归模型的预测平均相对误差达到了9.78%。研究结果表明，线性回归模型对煤灰熔点的预测结果更准确。

Rapid detection of ash melting point based on laser induced breakdown spectroscopy

Slagging in furnace is one of the key factors affecting the reliable operation of thermal power units and gasification process. The accurate prediction of ash melting point can adjust the furnace outlet temperature in advance to avoid slagging. In this study, the metal atomic spectra of coal ash samples were collected by Laser-induced breakdown spectroscopy (LIBS) to obtain the characteristic spectral lines of metals in coal ash, a random forest model (RF), a support vector machine regression model (SVM) and a linear regression model (LR) were established to directly predict the ash fusion temperature of coal ash samples. The raw spectral data of fly ash were preprocessed based on outlier data elimination algorithm based on Mahalanobis Distance (MD) and the baseline estimation and noise reduction algorithm based on coefficient matrix (BEADS). The predicted average relative error (MRE) of random forest model is 54.74%, MRE of support vector machine regression model is 60.08%, and MRE of linear regression model is 9.78%. Results show that the linear regression model is more accurate in predicting the melting point of coal ash.