摘要
采用正交试验设计,响应面设计以及基于BP神经网络的遗传算法模型,对餐厨垃圾厌氧消化产沼气的操作参数进行优化,并比较分析这3种模型的优化效果。结果表明,采用正交试验设计模型,当接种比、含固率、初始pH分别为3、8%和9时,沼气产量最大,实际产量为1 015 mL/gTS,并且这3个因素对产气量的影响顺序为接种比>初始pH>含固率;采用响应面设计模型,当接种比、含固率、初始pH分别为2.42、8.62%和8.49时,最大沼气产量为1 049.85 mL/gTS,实际产量为1 029.5 mL/gTS,接种比对产气量的影响极其显著;采用基于BP神经网络的遗传算法模型,当接种比、含固率、初始p H分别为2.66、8.06%和8.87时最大产气量的预测值和实测值分别为1 085.8 mL/gTS和1 067.25 mL/gTS,实测值分别比正交设计模型和响应面设计模型的实测值提高5.15%和3.67%,表明在餐厨垃圾厌氧消化产沼气参数优化实验中,采用基于BP神经网络的遗传算法具有更高的准确度。
The orthogonal experimental method (OEM), response surface methodology (RSM) and an genetic algorithm based on back propagation neural network (BPNN-GA) were carried out to investigate the effects of operation parameters on the cumulative biogas generation from food waste by anaerobic digestion. The modeling and optimizing abilities of these three models were compared, The results showed that according to the OEM test, the maximum biogas yield could be obtained when the I/S, TS and pH were 3, 8% and 9, respectively. The biogas yield from the experiment reached 1 015 mL/gTS, and US had the most effect on biogas production, followed by the initial pH and TS. According to the RSM test, the optimal operational parameters (I/S, TS, initial pH) for biogas production were 2.42, 8.62% and 8.49, and the predicted and actual experimentally biogas yield were 1 049.85 mL/gTS and 1 029.5 mL/gTS, respectively. Among these three factors, the effect of US on biogas generation was extremely significant. According to the BPNN-GA method, the maximum predicted biogas yield of 1 085.8 mL/gTS could be achieved when the I/S, TS and initial pH were 2.66, 8.06% and 8.87, respectively, while its actual biogas yield from the experiment was 1 067.25 mL/gTS. The experimentally biogas by BPNN- GA method was 5.15% and 3.67% higher than that of the OEM and RSM, respectively, indicating that the genetic algorithm based on back propagation neural network had the highest optimizing ability during the optimization digestion from food waste by anaerobic digestion.
出处
《环境科学与技术》
CAS
CSCD
北大核心
2017年第8期164-170,176,共8页
Environmental Science & Technology
基金
国家自然科学基金(51508230
51678279)
国家科技支撑项目(2013BAB11B02)
关键词
餐厨垃圾
厌氧消化
响应面设计
BP神经网络
遗传算法
food waste
anaerobic digestion
response surface methodology
BP neural network
genetic algorithm