BP神经网络在贵州喀斯特山区径流预报中的应用

Application of BP Neural Network to Runoff Forecasting in Karst Mountainous Area in Guizhou

以贵州六冲河、倒天河流域为例建立喀斯特山区径流预报BP神经网络模型。六冲河流域以七星关站丰水期流量过程为输出数据,以丰水期降雨过程、出口断面前期流量过程、蒸发量作为输入数据,倒天河流域以徐家屯站丰水期流量过程为输出因子,丰水期降雨过程、前期流量过程作为输入因子。预报结果确定性系数DC值分别为0.538、0.420。结果表明将蒸发量作为输入数据、流域面积比较大模型预报精度较大。

Taking Liuchong River basin and Daotian River basin of Guizhou province as examples.The paper establishes the runoff forecasting model in Karst mountainous area.In Liuchong River basin we establish the model by taking Qixingguan station's data of runoff process during high water period as the outputs and taking the data of its rainfall process and earlier period runoff and evaporation as the inputs.In Daotian River basin we establish the model by taking Xujiatun station's data of runoff process during high water period as the outputs and taking the data of its rainfall process and earlier period runoff as the inputs.The deterministic coefficient of the results is 0.538 and 0.420.The result shows it whill improve the precision while taking the evaporation as input and the area of the basin is biger.