Fig. 1. Schematic diagram of experimental setup

Fig. 2. Flow chart of copper alloy classification algorithm based on interpretable deep learning model

Fig. 3. Plasma emission spectra under SP-LIBS and target-enhanced DP-LIBS conditions

Fig. 4. Comparison of spectra before and after baseline correction. (a) Raw spectrum; (b) spectrum after baseline correction

Fig. 5. Model performance under different optimizers. (a) Training loss value; (b) training accuracy

Fig. 6. Feature points of samples with top 100 IG values and experimental spectra

Fig. 7. Confusion matrices under different wavelength feature selection methods. (a) Manual feature extraction; (b) PCA; (c) feature selection based on IG-value

Fig. 8. Comparison of classification model performance metrics under different experimental conditions

Fig. 9. Model evaluation results