星云湖富营养化进程的神经网络模拟及污染控制对策研究

以星云湖为研究对象，通过多年水生态监测数据筛选出富营养化的关键因子，利用BP神经网络模拟叶绿素a与各因子之间的关系，定量分析了叶绿素a的压力响应情况，结果表明:CODMn、TP、TN是富营养化进程中3个关键因子；以0.02mg/L为富营养化湖泊中叶绿素a的控制目标，需分别削减61%的CODMn或77%的TP或20%的TN. 模拟结果显示，星云湖的藻类生长以氮为限制因子. 基于神经网络模拟分析星云湖的富营养化进程，为星云湖水污染控制提供重要的决策依据.

Neural network modeling of the eutrophication and strategy of pollution control in Lake Xingyun

Lake Xingyun was selected as a study object. The key factors of the eutrophication were screened out using PCA, and back-propagate neural network was used to simulate the relation between chlorophyll a and key factors, and the pressure-response effect between chlorophyll a and key factors was quantitatively analyzed. The conclusions are: CODMn, TP and TN were the key factors of the eutrophication. Set 0.02 mg/L as the control target of chlorophyll a, then 61% of CODMn or 77% of TP or 20% of TN should be reduced. This result indicated that N was the limiting factor of the phytoplankton in Lake Xingyun. This simulation of eutrophication provided the basic data for the remediation of Lake Xingyun.