Simsek Halis
a Department of Agricultural & Biosystem Engineering , North Dakota State University , Fargo , ND , USA.
Environ Technol. 2016 Nov;37(22):2879-89. doi: 10.1080/09593330.2016.1167964. Epub 2016 Apr 24.
Wastewater-derived dissolved organic nitrogen (DON) typically constitutes the majority of total dissolved nitrogen (TDN) discharged to surface waters from advanced wastewater treatment plants (WWTPs). When considering the stringent regulations on nitrogen discharge limits in sensitive receiving waters, DON becomes problematic and needs to be reduced. Biodegradable DON (BDON) is a portion of DON that is biologically degradable by bacteria when the optimum environmental conditions are met. BDON in a two-stage trickling filter WWTP was estimated using artificial intelligence techniques, such as adaptive neuro-fuzzy inference systems, multilayer perceptron, radial basis neural networks (RBNN), and generalized regression neural networks. Nitrite, nitrate, ammonium, TDN, and DON data were used as input neurons. Wastewater samples were collected from four different locations in the plant. Model performances were evaluated using root mean square error, mean absolute error, mean bias error, and coefficient of determination statistics. Modeling results showed that the R(2) values were higher than 0.85 in all four models for all wastewater samples, except only R(2) in the final effluent sample for RBNN modeling was low (0.52). Overall, it was found that all four computing techniques could be employed successfully to predict BDON.
来自废水的溶解有机氮（DON）通常占高级污水处理厂（WWTPs）排放到地表水的总溶解氮（TDN）的大部分。在考虑对敏感受纳水体中氮排放限制的严格规定时，DON就会成为问题，需要加以减少。可生物降解的DON（BDON）是DON的一部分，在满足最佳环境条件时可被细菌生物降解。使用人工智能技术，如自适应神经模糊推理系统、多层感知器、径向基神经网络（RBNN）和广义回归神经网络，对两级滴滤式污水处理厂中的BDON进行了估算。亚硝酸盐、硝酸盐、铵、TDN和DON数据用作输入神经元。从该厂四个不同位置采集了废水样本。使用均方根误差、平均绝对误差、平均偏差误差和决定系数统计量对模型性能进行了评估。建模结果表明，除RBNN建模的最终出水样本中的R(2)值较低（0.52）外，所有四个模型中所有废水样本的R(2)值均高于0.85。总体而言，发现所有四种计算技术都可成功用于预测BDON。
