The semi-quantitative analysis of hole defects in metal additive manufacturing components using LIBS

The defects of additive manufacturing (AM) components will lead to insufficient mechanical properties of the manufactured parts, which limit their application in advanced industries. Therefore, it is very important to identify and control the defects generated in the AM process. In this work, the semi-quantitative analysis of hole defects in metal AM components was carried out using the method of segmental feature selection (SFS) combined with the spectral line intensity-ratio (LIR). Five depths of hole defects were detected using laser-induced breakdown spectroscopy (LIBS). And three classifiers, support vector machine (SVM), naive Bayes (NB) and random forest (RF), were used to assist in verifying the effectiveness of the SFS combined with the LIR. Compared with the traditional manual selection method based on the wavelength, the method of using SFS combined with the LIR improved the evaluation index and robustness of the classifier. The accuracies of SVM, NB and RF were improved by 3.7%, 11.4% and 10.7%, respectively. And the average scores of five-fold cross validation were increased by 3%, 5.1% and 6%, respectively. This study showed that the method of SFS combined with LIR can improve the semi-quantitative analysis results of LIBS metal AM component hole defects, and provide an effective way to solve the spectral line drift. The method of segmented feature selection combined with spectral line intensity ratio was employed to improve the semi-quantitative analysis results of LIBS metal AM component hole defects.