基于神经网络的北京市水体水华短期预报系统

Water-bloom short-time predicting system of Beijing based on neural network

采用算法改进型的BP神经网络,选择叶绿素含量、磷、氮磷比、电导率和水温五个参数作为模型输入,以预测1日、3日和5日后的叶绿素含量为目标,构建了北京市长河水系水华短期预报系统。该系统三个周期的预测精度分别达到了97.2%、94%、88.3%,并且具有较好的泛化能力。相比于其它智能算法,BP神经网络结构简单、方便实用,仍然具有很强的应用性。

A water-bloom short-time predicting system of Beijing CHANGHE water system is founded with ameliorated BP neural network,in which content of chlorophyll,phosphor,ratio of nitrogen and phosphor,conductance and temperature of water are chose as the inputs of model and the target is to predict the contents of chlorophyll after 1 day,3 days and 5 days.The precisions of three periods,which the system get,reached separately 97.2%,94%,88.3%,and the system has good universality.Compared with other intelligent methods,BP neural network is simple,convenient and practicality,and has good appliance.