Laser-based quality assessment of stored apple: Integrating LIBS, LIPA, and DFT Analysis

This study presents a novel in-situ method that combines laser-induced breakdown spectroscopy (LIBS) and laser-induced plasma acoustic (LIPA) signals to classify the quality of apple flesh at various temperatures (2 °C, 15 °C, 25 °C). The combined LIBS-LIPA model achieved high classification performance, with accuracies ranging from 94% to 98% across KNN, Gradient Boosting, Random Forest, XGBoost, and SVM, significantly outperforming LIBS-only classification, which ranged between 48% and 85%. Shapley additive explanation (SHAP) is used to evaluate feature contributions and influence feature co, thereby increasing model transparency. The analysis reveals that acoustic signals play a significant role in the classification task, with key features such as LIPA-LD2 accounting for 15.66% and effectively capturing the structural changes. Density functional theory (DFT) calculations support the experimental results. They play a dominant role in acoustic features, and the combined model achieves essentially perfect classification accuracy across different temperatures.