Adaptive Neuro - Fuzzy Inference System ( ANFIS ) application in modeling the oil extraction from peanut with microwave pretreatment

In this research the modeling of extracting oil from peanut using the adaptive Neuro-Fuzzy Inference System (ANFIS) was studied. For this reason the microwave time and the rotary speed of screw press was considered as the input and the yield of oil extraction, density, color, acidity, and oil oxidative stability was considered as the output. Three Gaussian, triangular and trapezoidal membership functions were considered with 2-2 and 3-3 membership function. The results indicated that the trapezoidal function with 3-3 membership function is chosen for 3 output variables of oil extraction efficiency as the optimum model. Also for the output variable of density, color, and oxidative function the Gaussian function with 3-3 membership function is chosen as the optimum model and at last for the oil acidity the triangular model with 3-3 membership function is chosen. On one hand it was recognized that with increase in the microwave time the amount of oil extraction efficiency and density was increased but the amount of oil acidity and color was decreased at first and was increased later. Also, increase in the rotary speed of screw press led to decrease the oil extraction efficiency and density but the amount of oil color and oxidative stability was increased. In the end it can be said that amount of high correlation coefficients between the laboratory results and outputs of model represent the acceptable accuracy and usability of these models in control the peanut oil extraction process with microwave pre-treatment.