Infrared-high humidity air as a novel blanching technique: Physicochemical characteristics and deep learning-based prediction of drying kinetics in carrots

ABSTRACT The effects of combined infrared and high relative humidity (RH) air blanching for different durations (0, 2, 4, 6, and 8 min) on carrot drying kinetics, quality characteristics, and underlying mechanisms were investigated. Compared with the unblanched control samples, blanching for 6 min reduced the drying time by 44.44%. The correlation mechanism revealed that an optimal blanching time softens tissue, increases cell pore size, and enhances water mobility. Carrots blanched for 6 min exhibited complete peroxidase enzyme inhibition, improved color retention, elevated total phenol and β-carotene content, and enhanced antioxidant capacity. Additionally, an optimized attention-based long short-term memory (LSTM) model outperformed traditional thin-layer drying models and LSTM-based approaches without attention mechanisms, achieving exceptional prediction accuracy for drying time, blanching time, drying temperature, RH, and moisture ratio, with R ² values of 0.99905, 0.99902, 0.99919, and 0.99906 for training, validation, testing, and overall datasets, respectively. These findings demonstrate a cohesive experimental-computational framework in which high-humidity infrared blanching enhances process performance while the attention-based LSTM provides a precise and scalable tool for modeling complex moisture-transfer dynamics.