Optimization of Bauhinia monandra seed oil extraction via artificial neural network and response surface methodology: A potential biofuel candidate

摘要

The influence of sample weight, time, and solvent type and their reciprocal interactions on Bauhinia monandra seed oil (BMSO) yield using artificial neural network (ANN) and response surface methodology (RSM) was investigated. Also, the BMSO obtained was characterized to determine its aptness for oleochemical industry. Numerically predicted optimum values for the extraction process using RSM model were found to be the same for the developed ANN model. The optimum values were sample weight of 60 g, time of 100 mm and petroleum ether with a corresponding BMSO yield of 14.8 wt%. Performance evaluation of the models by multiple coefficient of correlation (R), coefficient of determination (R-2) and absolute average deviation (AAD) showed that the ANN model was marginally better (R=0.9995, R-2=0.9991, AAD=0.27%) than the RSM model (R=0.9993, R-2=0.9986, AAD=0.49%) in predicting BMSO yield. Physicochemical properties of the BMSO such as acid value (7.56 mg KOH/g), indicated that it is non-edible and the fatty acids profile showed that the oil was highly unsaturated (87.9%), which makes it a potential candidate for biodiesel production. (C) 2015 Elsevier B.V. All rights reserved.