应用人工神经网络评价湖泊的富营养化

应用人工神经网络方法，以化学需氧量、总氮、总磷和透明度作为评价参数，经反复尝试，构建了具有４层结构用于评价湖泊富营养化的误差逆传播网络．其输入层有４个神经元，２个隐含层也各有４个神经元，输出层有１个神经元．以太湖富营养化评价标准作为样本模式提供给网络，按照误差逆传播网络的学习规则对网络进行训练，经过３７６８４次学习后，网络达到预先给定的收敛标准．使网络具备了识别湖泊富营养化程度的功能．应用该网络对我国１７个湖泊的富营养化程度进行评价，操作过程简便易行，评价结果切合实际，展示了这种方法的一系列优点．

Artificial neural network evaluation of lake eutrophication.

Taking chemical oxygen demand, total nitrogen, total phosphorus and transparency as artificial neural network evaluation parameters and after repeated attempts. the four layer structural Error Back Propagation Network(EBPN) was established to evaluate lake eutrophication. There are four neural units in input layer, four in both hidden layers, and one in output layer. Taking the eutrophication evaluation criterion of Taihu Lake as sample pattern, the network was trained in the light of learning rule of EBPN. After 37684 tries, the network reached the convergence standard given in advance, enabling it to possess the function of distinguishing the degree of lake eutrophication. This network was used to evaluate the eutrophication degree of 17 lakes in China. Its operation process was simple and convenient, and the results accorded with reality, showing that the approach has a series of advantages.