提高激光诱导击穿光谱的在线测定煤质方法研究

Improvementof coal quality online detectionmethodof laser-induced breakdown spectroscopy

灰分和热值是影响煤炭工业生产与利用的重要因素,如何快速、精确地进行在线煤质测定也是开采和使用中亟需解决的关键问题。利用激光诱导击穿光谱技术(LIBS)结合偏最小二乘回归(PLSR)和极限学习机模型(K-ELM)用于在线煤炭灰分和热值分析,通过对样品中Si、Al、Fe、Ca、Mg、Na、K、Ti、N、H等元素特征谱线进行主导因素选择(DF),进一步提高预测模型的分析精度。结果发现,利用DF方法,PLSR和K-ELM模型的预测精度均有所提高;相比于常用的PLSR预测模型,DF-K-ELM模型的煤炭灰分和热值的决定系数分别提高了3.4%和5.6%,均方根误差分别降低了0.214和0.297。所研究的煤炭灰分和热值预测模型对于提高LIBS煤质在线分析具有重要的参考价值。

Ash content and calorific value are important factors affecting the production and utilization of coal industry.How to quickly and accurately determine the online coal quality is also a key problem to be solved in mining and use.Laser induced breakdown spectroscopy(LIBS) combined with Partial Least Squares Regression(PLSR) and Kernel Based Extreme Learning Machine(K ELM) was used for online coal ash and calorific value analysis.Characteristic spectral lines of H and other elements are selected as dominant factors(DF) to further improve the analysis accuracy of the prediction model.Results show that the prediction accuracy of PLSR and K ELM models is improved by DF method;Compared with the commonly used PLSR prediction model, the determination coefficients of ash content and calorific value of DF K ELM model are increased by 3.4% and 5.6% respectively, and the root mean square error is reduced by 0.214 and 0.297 respectively.Therefore, the coal ash and calorific value prediction model studied has important reference value for improving LIBS coal quality online analysis.