基于BP神经网络的水环境承载力预警研究——以北运河为例

为了预测水环境承载力未来可能出现的超载状态并提出警告，达到水环境风险管控的目的，本文构建了基于BP神经网络的水环境承载力预警方法体系，并在北运河流域开展了实证研究.所构建预警模型包括COD、氨氮、总磷承载力预警子模型和水资源承载力预警子模型，且模型拟合效果较好（平均绝对百分比误差在20%左右）.研究结果表明：朝阳区、海淀区等8个行政区落在了红灯重警区域，水环境承载力超载状况最为严重；位于北运河中下游的石景山区、广阳区、北辰区和香河县处于橙灯中警区域；昌平区、顺义区等5个行政区处于黄色轻警区域；怀柔区和延庆区处于绿色无警区域，状况较好.最后，为了排除警情，基于双向调控的原则，本文从水环境全过程控制角度，分别对各区域提出了调控区域人口规模及经济发展方式、治理面源污染等具体排警措施.

Research on the early-warning method of water environment carrying capacity based on BP neural network:A case study of Beiyunhe River Basin

Aim to predict the possible overloading status of water environment carrying capacity (WECC) in the future and give warnings to control water environment risk, we constructed an early warning system of WECC by adopting Back Propagation Neural Network (BPNN),and Beiyunhe river basin was selected for a case study. The early-warning model involves four sub-models of carrying capacity (COD, NH₄, and TP and water resources), and it can provide reliable fitting result (the mean absolute percentage error is about 20%). The results show that 8 administrative districts including Chaoyang District and Haidian District are in the red-light warning area where the WECC is severely overloaded; Shijingshan District, Guangyang District, Beichen District and Xianghe County which located in the middle and downstream are in the orange-light waring area; 5 administrative districts including Changping District and Shunyi District are in the yellow-light warning area; Huairou District and Yanqing District are in the green-light warning area, and the carrying status is good. Finally, to eliminate the overloading situation, from the perspective of integrated control of water environment, we proposed specific management measures for each region on the basis of the principle of bidirectional regulation, such as regulation of population and economic development, non-point source pollution control, etc.