Online detection of laser paint removal based on laser-induced breakdown spectroscopy and the K-nearest neighbor method

Abstract

Laser paint removal is a new highly efficient and environmentally friendly cleaning technology. Compared with traditional paint removal methods, laser paint removal is less labor intensive and can reduce environmental pollution. During laser paint removal, real-time monitoring is necessary to ensure efficient cleaning and process automation. Current methods for real-time monitoring of laser paint removal only determine whether the sample surface has been cleaned but provide no information on the status of any residual paint. In this article, spectral data of the sample surface have been obtained using laser-induced breakdown spectroscopy. It is shown that Zn and Fe spectral lines can be used in real time to characterize the effectiveness of paint removal and that the intensities of characteristic spectral lines are positively correlated with the single-pulse energy of the excitation light. The K-nearest neighbor algorithm was used to evaluate and automatically classify the extent of cleaning of sample surfaces in real time. When K = 3, the classification accuracy of distinguishing different levels of cleaning was 100%. The results of this study provide technical support for automatic and intelligent laser paint removal.