Challenges and Emerging Solutions in Coal Analysis Technology Based on Laser-Induced Breakdown Spectroscopy

Abstract

Laser-Induced Breakdown Spectroscopy (LIBS) is a novel type of atomic emission spectroscopy. It has the advantages of rapid detection and no sample pretreatment, and can simultaneously detect the characteristics of multiple components. Therefore, it has great application potential in the analysis of the coal industry. However, due to the intricate compositional characteristics of coal, significant spectral discrepancies exist among different measurement results, while notable spectral similarities are observed across distinct coal types. These phenomena pose challenges to the accuracy and reliability of coal analysis outcomes. The paper summarizes the key factors that influence the performance of LIBS in coal analysis, such as the redundancy of spectral features, matrix effects, self-absorption effects, or environmental factors in online detection. To address these issues, this paper introduces methods for improving classification accuracy and quantitative analysis precision, such as feature selection algorithms, machine learning models for classification, multifactor quantitative methods, spectral preprocessing techniques, and transfer learning approaches. It also presents novel techniques for reducing spectral fluctuations during online detection and enhancing the generalization capability of online models. Overall, the advancement of LIBS-based coal analysis technology relies on the innovative integration of spectroscopy, chemometrics, system integration, and other relevant aspects to meet the demands of high-precision and high-reliability industrial online analysis.