Application of design of experiments and artificial neural network in optimization of ultrasonic energy-assisted transesterification of Sardinella longiceps fish oil to biodiesel

Abstract

The competent and efficient utilization of feedstocks is highly essential in the transesterification of biodiesel from sardine fish oil. The identification of optimal reaction parameters is of high importance to maximize the yield of biodiesel produced from sardine fish oil at low cost. Application of ultrasonic energy-assisted biodiesel production from sardine fish oil catalyzed by KOH catalyst has been studied under different conditions. Response surface methodology (RSM) based on central composite rotatable design (CCRD) was employed to optimize the three important process parameters: methanol/oil molar ratio (X1), KOH catalyst concentration (X2), and reaction time (X3) for transesterification of sardine fish oil using ultrasonic energy. Artificial neural network (ANN) models with two feed-forward back-propagation neural network architecture, multilayer perceptron networks and radial basis function networks have been developed to obtain a good correlation between the input variables responsible for the input reaction parameters and the output parameter yield of fatty acid methyl ester (FAME) from sardine fish oil to biodiesel. The developed ANN models were trained and tested with the experimental data obtained from the RSM–CCRD method. The developed ANN models’ performances were compared with experimental data and were statistically compared by the coefficient of determination (R2), root-mean-square error, and mean absolute error. From the statistical analysis, it was found that the estimated yield of FAME from both RSM and ANN models was able to predict the FAME yield, and the results showed that the ANN model is much more accurate in the prediction of FAME yield as compared to the RSM model.