Ping Shao

Analysis of conditions for microwave-assisted extraction of total water-soluble flavonoids from Perilla Frutescens leaves

Perilla Frutescens is a traditional Chinese medicinal herb. Microwave-assisted extraction (MAE) technique was developed for the fast extraction of flavonoids from Perilla Frutescens leaves. Several influential parameters of the MAE procedure (microwave power, extraction cycles, solvent to material ratio, irradiation time and pH value) were investigated for the optimization of the extraction using single factor and Box-Behnken experimental design. Response surface methodology analysis showed good correspondence between experimental and predicted values. It was found that the most effective parameter was solvent to material ratio, which was in good agreement with the experimental value. The adjusted coefficient of determination (Radj2) for the model was 93.45%. Probability value (Pu2009<u20090.001) demonstrated a very high significance for the regression model. The maximum yield of flavonoids with MAE was obtained by dual extraction with solvent to material ratio of 1: 16.5, irradiation time of 23xa0min, and pH value of 8.4 at microwave power of 600xa0W. The optimal yield with MAE 6.07xa0mg/g were slightly lower than that of Soxhlet extracted and higher than that of heat reflux extraction with water. The extracts exhibited high scavenging effects on DPPH radical. With increasing concentration between 0.10 and 0.80xa0mg/mL, the scavenging rate achieved 62.3%.

ARTIFICIAL NEURAL NETWORK ANALYSIS OF IMMOBILIZED LIPASE CATALYZED SYNTHESIS OF BIODIESEL FROM RAPESEED SOAPSTOCK

Refined vegetable oils are the predominant feedstocks for the production of biodiesel. However, their relatively high costs render the resulting fuels unable to compete with petroleum-derived fuel. Artificial neural network (ANN) analysis of immobilized Candida rugosa lipase (CRL) on chitosan catalyzed preparation of biodiesel from rapeseed soapstock with methanol was carried out. Methanol substrate molar ratio, enzyme amount, water content and reaction temperature were four important parameters employed. Back-Propagation algorithm with momentous factor was adopted to train the neural network. The momentous factor and learning rate were selected as 0.95 and 0.8. ANN analysis showed good correspondence between experimental and predicted values. The coefficient of determination (R2) between experimental and predicted values was 99.20%. Biodiesel conversion of 75.4% was obtained when optimum conditions of immobilized lipase catalysed for biodiesel production were methanol substrate molar ratio of 4.4:1, enzyme amount of 11.6%, water content of 4% and reaction temperature of 45°. Methyl ester content was above 95% after short path distillation process. Biodiesel conversion was increased markedly by neural network analysis.