Prediction and changing patterns of physicochemical properties of rice flour during radio frequency heating

Poor structural formability of native rice flour limits its application in gluten-free food processing. Radio frequency (RF) heating, as an emerging dielectric heating technology, is a promising alternative to conventional thermal methods as it enables volumetric heating through molecular dipole rotation and ionic conduction, achieving faster heating rates and higher energy efficiency for improving rice flour properties. Conventional quality assessment is time-consuming and costly, motivating adoption of back-propagation artificial neural networks (BP-ANN) due to their demonstrated superiority in decoding complex nonlinear relationships between multi-parametric process inputs and physicochemical outputs. This study investigated the effects of RF technology under different treatment temperatures (40–70 °C) and rice slurry concentrations (5 %–40 %) on the gelatinization degree (GD), pasting, functional, and rheological properties of rice flour and successfully established a quality prediction model using BP-ANN. Results showed that RF treatment enhanced GD and functional properties (e.g., water absorption capacity increased by 3.14 times, oil absorption capacity by 82.7 %), but excessive heating (> 55 °C) compromised viscosity and gel network stability. Multivariate analysis revealed that the treatment temperature was the primary factor driving physicochemical changes in rice flour. The BP-ANN achieved high-precision predictions, outperforming other machine learning methods. These findings provide feasibility for the intelligent modification of rice flour during RF heating.