Integrating artificial intelligence and statistical analysis into a prediction model for drying performance and reduced energy consumption in industrial infrared conveyor-belt drying systems

Recent studies in drying technology increasingly integrate artificial intelligence with thermodynamic analysis to enhance energy efficiency and predictive accuracy. This study investigates the drying behavior of onion slices in a semi-industrial infrared–convective conveyor-belt drying system under varying air velocities (0.3–0.7 m/s), air temperatures (45–65 °C), and infrared (IR) power levels (200–600 W). Drying time decreased significantly from approximately 500 min at 200 W and 45 °C to less than 250 min at 600 W and 65 °C. The effective moisture diffusivity (Deff) increased from 0.25 × 10 -10 to 0.45 × 10 -10 m 2 /s with increasing temperature and IR intensity, while activation energy ranged between 7.85 and 11.87 kJ/mol. Total energy consumption varied between 0.60 and 4.94 kWh depending on operating conditions, and the lowest specific energy consumption (SEC) of 10.72 kWh/kg was achieved at 600 W, 65 °C, and 0.3 m/s. Thermal efficiency improved from 7.73% to 16.92% with increasing temperature and radiation intensity. The integration of thermodynamic modelling with Artificial Neural Networks (ANN), Self-Organizing Maps (SOM), and Principal Component Analysis (PCA) to model, visualize, and interpret drying performance in an industrial convective conveyor system based on infrafred contribute to reduced energy consumption and enhanced process efficiency. • AI is applied to a typical continuous drying process for energy-saving drives • The energy aspects of an infrared dryer for onion slices have been investigated • Machine learning, principal component analysis (PCA), and a self-organizing map (SOM) were utilized to predict and model. • The controlled, showing a 7.9 % drying time decrease and a 45 % energy-saving • This work could help designers improve the efficiency and energy conservation of onion drying.