Comparative analysis of thin-layer drying models in predicting papaya moisture content using optimization algorithms

Purpose This study aims to comparatively analyze 15 widely used thin-layer drying models for predicting the moisture ratio (MR) of papaya slices and to evaluate the performance of four optimization algorithms – HFO-1, IGWO, GRG and LM – in estimating model parameters. Design/methodology/approach Papaya slices with thicknesses of 6, 8 and 10 mm were dried at air temperatures of 70, 80 and 90 °C in a convective tray dryer. Parameter estimation for each model was performed using four optimization algorithms. Model accuracy was evaluated using RMSE and MAE. Comparative performance ratios and model–algorithm matches were examined to determine the most accurate and stable methods. Findings Overall, the HFO-1 method is superior to LM, achieving lower RMSE values in 83.70% of all evaluated cases, with 8.15% showing equal performance. Compared with GRG, RMSE values are equal in 69.63% of the cases, while HFO-1 performs better in 20.74%. In relation to IGWO, the RMSE values are equal in 52.60% of the cases, while HFO-1 is better in 45.18% of them. Across all drying temperatures and slice thicknesses, the Modified Henderson and Pabis-I model demonstrated the best predictive performance. The findings showed that HFO-1 consistently achieved strong error minimization and stable convergence, particularly under nonlinear and multimodal model structures. Research limitations/implications The study focuses on convective drying of papaya slices under controlled laboratory conditions. Results may vary for other drying techniques or food materials. Practical implications Selecting the most effective algorithm–model combination can support better drying process design, improve energy efficiency and enhance moisture prediction accuracy in industrial food drying applications. Originality/value This study is the first to apply both IGWO and HFO-1 algorithms to optimize parameters of thin-layer drying models. Evaluating 15 different models simultaneously makes this work one of the most comprehensive analyses in drying-model optimization. HFO-1's strong accuracy and convergence capacity highlight its suitability for broader applications in food engineering and drying kinetics.