【摘要】 为掌握丹江口库区水质未来的变化趋势以及预防污染事件的发生,建立了一个水质指标的预测模型。利用库区某断面自动检测站的水质指标实测参数作为学习样本,选取化学需养量(COD)、生化需养量(BOD)、pH值、氨氮(NH3-N)、总磷(TP)、总氮(TN)等指标作为预测参数,运用Levenberg-Marguardt优化算法对学习样本进行优化,建立基于反向传播(BP)神经网络的预测模型并应用于丹江口库区水质指标。结果显示,实际检测值与预测值相对误差小于7%,该模型具有良好的可行性和有效性。更多还原

【Abstract】 A predictive model was set up to grasp the future change tendency of water quality about Danjiangkou reservoir and prevent further pollution.The historical time series of water quality indexes in district border of Danjiangkou reservoir were taken as instructive samples,and six indexes were taken as predicted indexes,such as chemical oxygen demand(COD),total phosphor(TP),total nitrogen(TN).The samples were modeled and optimized with Levenberg Marguardt algorithm of back propagation(BP)network.The predicted results indicate that the prediction of water quality is precise and fast,and the relative errors of the predicted results indexes is lower 7% with a few exceptions.The BP neural network model has feasibility and validity.更多还原