Comparison of Response Surface Methodology and Artificial Neural Network for the Solvent Extraction of Fatty Acid Methyl Ester from Fish Waste
Abstract
The evaluation of fatty acid methyl ester (FAME) extraction from fish waste were done by conducting experimental works using response surface methodology (RSM) and artificial neural network (ANN). The experiments were started with a preliminary experiment using one-factor-at-a-time method to evaluate the effect of temperature and mixing time on the production of FAME. Solvent extraction method was used to elucidated the best operating conditions with various temperatures (40 to 80 C) and mixing time (2 to 6 hours) using ethanol as a solvent. The FAME profile was then analyzed using Gas Chromatography Mass Spectrometry (GCMS) after each extraction. The result showed that the mean square error (MSE) for the ANN was lower (0.026 for oil yield and 0.019 for oleic acids) compared to RSM (0.23 for oil yield and 47.16 for oleic acids). Besides, the optimization using genetic algorithm (GA) demonstrated a higher oil yield (10.65 %) and oleic acid (30.01 mg/g) than using central composite design (CCD) with 10.48 % of oil yield and 18.19 mg/g of oleic acid. Based on the MSE analysis, it revealed that ANN model produced better prediction efficiency than the RSM model. Moreover, the results showed that the effects of each factor using GA to produce oil yield and oleic acid from fish waste were accepted to be used for FAME production.