海水水质评价的人工神经网络模型研究

通过在各类海水水质污染指标浓度区间内生成随机分布样本的方法，生成足够多用于人工神经网络模型训练和检验用的样本，并应用基于误差反传原理的前向多层神经网络建立了用于海水水质评价的人工神经网络模型。并根据海水水质标准给出了区分不同类型水质的模型分界样本和模型输出分界值。讨论了确定合理隐层及其节点数的方法，使得训练后的神经网络模型具有更强的泛化能力，不受网络连接权值初始取值的影响。经训练的评价模型应用于实例的评价结果表明，新的评价模型具有较好的客观性、通用性和实用性。

Sea water quality assessment model using artificial neural networks

A new approach producing training set data, testing set data and critical values set data randomly distributed between the critical values was established in this paper. The principle of determining the number of hidden layers and their neurons on each layer was also discussed. And,the sea water quality assessment model using multi layer feedforward neural networks with error back propagation algorithm (NN based model) was set up. The NN based model possessed the capacity of higher generalization and not impacted by the initial values of connection weight. The assessed results of cases shown that the new presented NN based model was objective, reliable, practicable, and fault tolerant.