ANFIS, ANN, and RSM modeling of moisture content reduction of cocoyam slices

Abstract

The capability of response surface methodology (RSM), artificial neural network (ANN), and adaptive neuro-fuzzy inference systems (ANFIS) in modeling and predicting moisture content reduction in the drying of cocoyam (Colocasia taro) slices was the purpose of this study. Design of experiment was utilized. Modified Fick's second law of diffusion was used to determine effective moisture diffusivity (Deff). Results indicated that drying time, air velocity, and temperature significantly affected the drying process. Deff values obtained ranged from 2.97 × 10−10 to 7.30 × 10−10m/s. Page model with R2 = 0.994, SSE = 0.0053, RMSE = 0.0611, best described the kinetics of the drying process. ANN, RSM, and ANFIS all showed significant modeling and predicting ability with R2 of 0.9583, 0.9519, and 0.9971, respectively. Genetic algorithm (GA) optimization showed minimum moisture content of 14.362%, 11.919%, and 11.293% for RSM-GA, ANN-GA, and ANFIS-GA, respectively. Additional statistical analysis lent credence to ANFIS as the best in modeling and predicting the moisture content reduction of cocoyam slices.