The identification of counterfeit high-quality edible oil using EEM fluorescence spectroscopy based on GA-optimized...

High-quality edible oils, characterized by their high nutritional value, are priced significantly higher than other types. However, adulteration and falsification remain a prevalent and prominent issue during production and processing. To address the authenticity identification of high-quality edible oils, this study focuses on multiple counterfeit types of sesame oil, peanut oil, and olive oil, and proposes a feature-optimized recognition method based on excitation-emission matrix (EEM) fluorescence spectroscopy fused with two-dimensional linear discriminant analysis (2D-LDA) and genetic algorithm (GA). The method employs 2D-LDA for feature extraction, followed by GA implementation where each feature is treated as a gene: initial populations are randomly generated, and iterative optimization is performed via roulette wheel selection, crossover, and mutation operations, ultimately outputting the optimal feature subset and corresponding model performance metrics. The model achieves a classification accuracy of 0.9773 for 11 types of genuine and counterfeit edible oils; after data augmentation, it still maintains a high accuracy of 0.9941, confirming its robustness. This method not only enables efficient dimensionality reduction of EEM fluorescence spectra but also reduces computational complexity, holding significant implications for the deployment and application of the algorithm.