Classification of uneven steel samples by laser induced breakdown spectroscopy based on a Bessel beam

Steel is one of the most widely used industrial materials in the world, and the classification of steel grades is of great concern in the steel industry. However, traditional analysis methods are time-consuming and complex, which are unsuitable for efficient detection. Gaussian beam-laser induced breakdown spectroscopy (LIBS) is widely used in steel analysis because of its speed and accuracy. However, the sample surface needs to be flat to keep the stability of LIBS, which limits the usage of LIBS. The Bessel beam, a non-diffraction beam with a long focal depth, has unique advantages in solving the variation of defocus caused by an uneven surface of the sample. To be suitable for industrial field inspection, a classification model for alloy steel samples based on a Bessel beam-LIBS system is established. In this study, the LIBS technology combined with a Bessel beam was used for the rapid detection and classification of steel samples with uneven surfaces. 510 LIBS spectra of 17 kinds of alloy steel based on a Gaussian beam and Bessel beam were collected separately, and based on the spectra, the recognition accuracies of alloy steel samples were compared between two systems. The spectra of uneven steel samples based on the Bessel beam-LIBS system can be definitely discriminated using support vector machines (SVMs), with an average accuracy of over 93.17%, which is better than that of the Gaussian beam. This research reflects the potential of Bessel beams in detecting samples with uneven surfaces and provides an effective method for industrial inspection.

Automated summary

A classification model for alloy steel samples based on a Bessel beam-LIBS system is established, allowing for rapid detection and classification of steel samples with uneven surfaces. By comparing the LIBS spectra of 510 alloy steel samples, it is found that the Bessel beam-based system can better discriminate uneven steel samples, with an average accuracy of over 93.17%, surpassing that of the Gaussian beam. This research highlights the potential of Bessel beams in detecting samples with uneven surfaces and provides an effective method for industrial inspection.