期刊
RENEWABLE ENERGY
卷 74, 期 -, 页码 87-94出版社
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.renene.2014.07.054
关键词
Breadfruit; Bioethanol; Yeast; Response surface methodology; Artificial neural network; Optimization
This study investigated the use of Breadfruit Starch Hydrolysate (BFSH) as the sole carbon source for bioethanol production and the optimization of the fermentation parameters. The results showed that the yeast was able to utilize the BFSH with and without nutrient supplements, with highest bioethanol yield of 3.96 and 3.60% volume fraction, respectively after 24 h of fermentation. A statistically significant quadratic regression model (p < 0.05) was obtained for bioethanol yield prediction. Response Surface Methodology (RSM) optimal condition values established for the bioethanol yield were BFSH concentration of 134.81 g L-1, time of 21.33 h and pH of 5.01 with predicted bioethanol yield of 3.95% volume fraction. Using Artificial Neural Network (ANN), multilayer normal feedforward incremental back propagation with hyperbolic tangent function gave the best performance as a predictive model for bioethanol yield. ANN optimal condition values were BFSH concentration of 120 g L-1, time of 24 h and pH of 4.5 with predicted bioethanol yield of 4.21% volume fraction. The predicted bioethanol yield was validated experimentally as 4.10% volume fraction and 4.22% volume fraction for RSM and ANN, respectively. Coefficient of Determination (R-2) and Absolute Average Deviation (AAD) were determined as 1 and 0.09% for ANN and 0.9882 and 1.67% for RSM, respectively. Thus, confirming ANN was better than RSM in both data fittings and estimation capabilities. (C) 2014 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据