A brief review of new data analysis methods of laser-induced breakdown spectroscopy: machine learning

Abstract

Laser-induced breakdown spectroscopy (LIBS) is a technology of content analysis and composition analysis based on the atomic excitation and emission spectrum of materials. It has been intense activity in the field because of its advantages such as fast detection speed, no environmental limitation and no sample pretreatment. The low accuracy of LIBS is a primary problem in current applications, and the better data analysis methods is the key to solve this problem. Recently, machine learning algorithms significantly improve the accuracy of LIBS compared with traditional analysis methods. Therefore, the researchers gradually begin to pay attention to the application of machine learning algorithms in the LIBS data analysis. It is a programming method to study how computers simulate the learning process of human beings to acquire new knowledge and skills and continuously improve their performance. It is widely used in data analysis, pattern recognition, artificial intelligence and other fields. Here, we introduce the basic principle of LIBS and machine learning algorithms, review the research situation and progress of the application of machine learning algorithms to LIBS, and put forward the problems and challenges of its application.

Keywords:

• LIBS
• laser-induced breakdown spectroscopy
• machine learning
• atomic emission

SUBJECT CLASSIFICATION CODES:

• 140.3025

Additional information

Funding

This work was supported by the National Key Research and Development Program of China under Grant “Research on The Quantitative Analysis of Impurity Transport Deposition and Wall Surface Composition Evolution” (No. 2017YFE0301306) and Research on ITER Boundary Impurity Transport and Wall Material corrosion (No. 2017YFE0301300).