Application of an adaptive neuro_fuzzy inference system (ANFIS) in the modeling of rapeseeds' oil extraction

In the present study, the temperature and moisture content of the output seeds of the cooking pot were considered as inputs or independent variables and the insoluble fine partial content of the extracted oil, moisture content of the extracted oil and obtained meals, as well as the oil content of the achieved meals and acidity value of the extracted oil were considered as responses and were designed. Three different activation functions, including Gaussian membership and triangular as well as trapezoidal were applied and studied. The trapezoidal function with a 3–3 membership function for the three output variables including insoluble fine partials of oil, oil acidity and moisture content of the meals as well as the triangle membership function with 2–2 and 3–3 functions, respectively, for moisture content of the extracted oil and the oil content of the obtained meals were evaluated and detected as optimized models in the current study. The above mentioned models demonstrated higher correlation coefficients (R2) between the experimental and predicted values and the lowest root mean squared errors, confirming the adaptability of the applied models in the present study. Practical applications Today, because of the high demand for crops for extensive application in the human diet, increases in the efficiency of the processing are attracting much more attention. In this regard, discovering and detection of the optimized conditions for processing with the minimum wastes looks very important and economic. Therefore, the uses of predictive methods in different food processes have been considered appropriate tools for improving of the efficiency of the processes as well as the enhancement of the quality of the produced products. In this respect, the ANFIS design as a novel predictive analytic tool, along with Response Surface Methodology (RSM) and Artificial Neural Network (ANN), are applied extensively. Thus regarding the above mentioned content, ANFIS design was applied to predict and optimize some of the selected physico‐chemical properties of the extracted oil through the extraction process. Therefore prediction of the optimized conditions of oil extraction could improve the quality of the extracted oil and performance of the extraction process with the minimum wastes during short and logical extraction time.