Modeling and optimization of extractionof oil from Sesamum indicum seeds. A case of response surface methodology vs. artificial neural network

In this work, response surface methodology (RSM) and artificial neural network (ANN) was used to optimize of oil from Sesamum indicum seeds. ANN predicted optimal condition for extraction was Sesamum indicum powder weight (SIPW) = 54.71 g, extraction time (ET) = 44.88 min and solvent volume (SV) = 165.8 mL. The predicted Sesamum indicum oil yield (SIOY) was validated as 85.70% (w/w) while RSM predicted optimal condition was Sesamum indicum powder weight (SIPW) = 60.00 g, Extraction time (ET) = 44.48 min and solvent volume (SV) = 150 mL. The predicted SIOY under this condition was validated as 83.20% (w/w). The result obtained showed that ANN was superior and more effective optimization tool than RSM owing to its value of RMSE, AAD, R2 , R2 Adj. Meanwhile, the qualities of Sesamum indicum oil yield (SIOY) as compared to the earlier researched works indicated that the oil produced is of good qualities and needs no further purification. Fatty acids profile reflected that the oil is highly unsaturated. The study concluded that the oil is not only edible, but also could have an industrial application. Keywords: Optimization, Response surface methodology, Artificial neural network, Fatty acid profile, Sesamum indicum oil