Quantitative analysis and identification of magnesium alloys using fs-LA-SIBS combined with machine learning methods

Abstract

This work employs the femtosecond laser-ablation spark-induced breakdown spectroscopy (fs-LA-SIBS) technique for the quantitative analysis of magnesium alloy samples. It integrates four machine learning models: Random Forest (RF), Support Vector Machine (SVM), Partial Least Squares (PLS), and k-Nearest Neighbors (KNN) to evaluate their classification performance in identifying magnesium alloys. In regression tasks, the models aim to predict the content of four elements: manganese (Mn), aluminum (Al), zinc (Zn), and nickel (Ni) in the samples. For classification tasks, the models are trained to recognize different types of magnesium alloy samples. Performance evaluation is based on sensitivity, specificity, and accuracy. The results indicate that the RFR model performs optimally for regression tasks, while the Random Forest Classification (RFC) model outperforms other models in classification tasks. This work confirms the feasibility of quantitative analysis and identification of magnesium alloys using the fs-LA-SIBS technique combined with machine learning methods. It establishes a technical foundation for real-time monitoring of alloys in subsequent laser-induced breakdown spectroscopy (LIBS) instruments.