机构地区: 湖南大学环境科学与工程学院
出 处: 《环境工程学报》 2015年第5期2425-2429,共5页
摘 要: 基于污泥固体停留时间(SRT)为20 d的污泥中温厌氧消化实验,建立一个3层BP神经网络,以前1~20 d的进泥挥发性悬浮固体(VSS)、当天消化罐p H值和碱度共22个参数为输入,预测污泥消化系统日产气量,结果表明,网络具有良好的学习能力、泛化能力和辨识能力,能够较为准确地预测出系统日产气量。此外,根据进泥VSS不同,利用网络预测能力,调节p H值和碱度到合适的值,系统日产气量有明显提高,进一步证明了网络具有良好的预测能力和实用性。 Based on the experiment of sludge mesophilic anaerobic digestion with 20 days of solid retention time( SRT),a 3-layer BP neural network was built. In the network,22 input parameters were volatile suspended solid( VSS) of inflow sludge during 20 days period before the day,p H and alkalinity of sludge that day and output was yield of biogas that day. The results of simulation showed good learning,generalization and recognition ability of the network. Exact prediction on daily yield of biogas was achieved. Besides,by utilizing the prediction ability to adjust p H and alkalinity to appropriate values under different VSS quantities,the daily yield of biogas significantly increased,which further demonstrates the network possesses preferable prediction ability and practicability.
领 域: [环境科学与工程]