Response Surface Methodology, Artificial Neural Network and Adaptive Neuro-Fuzzy Inference System for Oyster Mushroom Drying Kinetics

Kaushik Hazarika^*,  Abhijit Borah, Rejaul Hoque Bepary, Tabaruk Hussain and Saurajyoti Baishya

Department of Agricultural Engineering, Assam Agricultural University, Jorhat, Assam, India.

Corresponding Author Email: kaushik.hazarika11@gmail.com

Abstract:

The mushroom drying experiments were carried out in a tray dryer at temperatures ranging from 50 to 70 °C with drying times from 6 to 10 hours, and sample thicknesses from 5 to 20 mm. The data obtained from the drying experiments were used to fit drying models and to train ANN and ANFIS for moisture content of the sample. A response surface methodology analysis was carried out to integrate experimental data with regression-based models. The drying rate increased significantly (p<0.05) with increasing temperature and decreased with decreasing sample thickness, which indicated a typical falling-rate drying behaviour. RSM provided significant regression fits (R2=0.9067), whereas ANN (3-10-1 architecture) and ANFIS models showed superior predicted accuracy with overall R² values of 0.9966 and 0.9953, respectively. Among the three models, ANN exhibited the best performance with the lowest mean square error, followed closely by ANFIS, while RSM was less accurate in capturing the nonlinear drying behaviour. These findings enable accurate prediction of nonlinear mushroom drying kinetics, providing for future process monitoring and optimisation strategies.

Keywords:

ANN; ANFIS; Drying; Oyster Mushroom; RSM

[ HTML Full Text]