Optimization of atmospheric pollutant detection and identification based on LIBS technology

Abstract

The problem of air pollution has been increasingly serious around the world, highlighting the importance of air pollution prevention and control. Therefore, there is an urgent need for effective air pollution control methods. In this study, a new approach was introduced, combining laser-induced breakdown spectroscopy (LIBS) and the self-designed standard deviation extraction method. Experiments were conducted from three perspectives: the identification of volatile organic compound (VOC) isomers, the classification of atmospheric particulate matter, and the measurement of carbon concentration. LIBS was used to detect three different air pollutants in real time, providing information on the elemental composition of the samples. After applying the standard deviation extraction method, the two isomers of fluorobromobenzene were successfully distinguished. By simulating carbon emission sources using dry ice, the inclusion of the standard deviation extraction method effectively improved the recognition accuracy for small distances, compared to using principal component analysis (PCA) alone. Furthermore, the efficient and accurate performance of the standard deviation extraction method also provides guidance for detection in more complex real-life scenarios. Combining LIBS, the standard deviation extraction method, and machine learning, the study explored four different forms of fly incense, resulting in a significant improvement in classification accuracy. The results indicate that the method based on LIBS and the standard deviation extraction method successfully achieved the detection and identification of substances with similar characteristics in air pollutants, providing important data support for pollution source identification and pollution trend prediction and greatly improving the accuracy of recognition.