Drying Kinetics and Modelling of Andrographis Paniculata (Burm. f.) Leaves in an Infrared Dryer Using ANN and ANFIS Algorithms

This study investigates the characteristics of A. paniculata leaves dried with an infrared dryer. Several models have been developed to forecast the moisture ratio (MR), including ANN, ANFIS, and empirical models. The inputs to the ANN and ANFIS models were drying time and air temperature (40, 50, and 60°C), and the outputs were the MR values. Two statistical indices, the R2 and RMSE values, were indicators of the precision of the model, and the highest value of R2 and the lowest value of RMSE were the indicators of the model's suitability. The findings indicate that the Deff values ranged from 4.47x10-11 to 7.26 x10-11 m2/s (40-60°C). The obtained Ea value was 28.36 kJ/mol. Compared to the other empirical models, the Midilli et al. model achieved better forecast MR values (R² = 0.9985, RMSE = 0.0119). The greatest ANN model for predicting MR values has a tan-sigmoid activation function and a 2-6-1 topology (R² = 0.9993, RMSE = 0.0031). With the greatest possible R² value (0.9994) and the smallest RMSE (0.0026), the ANFIS model (bell-shaped MFs) predicted values far more accurately than ANN and empirical models. The ANFIS model had the highest R2 values and the lowest RMSE, suggesting that the predicted and experimental MR values were more closely aligned. Due to their useful applications, ANFIS models can be integrated into industrial control systems to determine MR in real time and adjust the machine's operation autonomously. ANFIS can be integrated into software tools for dryer design, performance evaluation, and operational planning.