低溶解氧下氨氧化过程神经网络预测控制模型  

Neural network prediction and control model for ammonia oxidizing process under low DO concentration

作　　者：冯红利 刘秀红[2] 杨庆[1] 黄斯婷[1] 崔斌[1] 周桐[1] 杨玉兵[1] 周薛扬 FENG Hong-li;LIU Xiu-hong;YANG Qing;HUANG Si-ting;CUI Bin;ZHOU Tong;YANG Yu-bing;ZHOU Xue-yang(Key Laboratory of Beijing Water Quality Science and Water Environment Recovery Engineering, Beijing University of Technology, Beijing 100022;School of Environment & Natural Resources, Renmin University of China, Beijing 100872)

机构地区：[1]北京工业大学北京市水质科学与水环境恢复工程重点实验室,北京100124 [2]中国人民大学环境学院,北京100872

出　　处：《中国环境科学》2017年第1期139-145,共7页China Environmental Science

基　　金：国家自然科学基金(51508561);北京市委组织部青年拔尖团队;北京市优秀人才培养资助计划

摘　　要：在低溶解氧(DO)状态下,以城市生活污水为研究对象,将神经网络预测的方法应用到污水处理过程中,建立了基于神经网络的氨氧化过程预测控制模型,预测并控制污水处理氨氧化过程.该模型分为两部分,一是根据在线pH值变化预测氨氧化结束时间,其相关系数R值为0.9985;二是根据在线pH值实时预测氨氮浓度,R值为0.9083.试验结果表明该模型预测精度高、可控性好,具有较好的适应性和稳定性,对实现并稳定短程硝化以及促进主流工艺(厌氧氨氧化)有重要的指导和借鉴意义.Under low dissolved oxygen(DO)concentration,the neural network prediction method was applied in SBR fortreating domestic wastewater.The neural network control model was built to predict and control the ammonia oxidizingprocess.The model was divided into two parts.In the first part with the correlation coefficient(R value)of0.9985,the endof ammonia oxidization was predicted based on the on-line pH variations.In the second part with R value of0.9083,theammonia concentration was real-time predicted based on the on-line pH variations.The results showed that the modelwith high prediction accuracy,good controllability,better adaptability and stability,can not only benefit for achieving andstabilizing short-cut,but also promote the application of anaerobic ammonium oxidation for treating domestic wastewater.

关 键 词：低溶解氧 氨氧化过程 神经网络 模型 PH 

分 类 号：X703[环境科学与工程—环境工程]