基于数据驱动模型的季节性河流洪水预警研究

Research on seasonal river flood warning based on data-driven model

为提升季节性河流洪水预警精度,实现洪水灾害的防治,提出基于数据驱动模型的季节性河流洪水预警方法。利用径流曲线法计算河床汇流率,将汇流率作为辅助函数,采用BP神经网络和数据驱动相结合的方法搭建季节性河流洪水预警模型,将季节性河流洪水的时间序列输入模型中,得到季节性河流洪水预警结果。以传统的季节性河流洪水预警方法作为对比对象,通过实验得出基于数据驱动模型的季节性河流洪水预警方法在预警精度上优于传统预警方法,预警结果具有较高的可靠性,能够对季节性河流进行精准的洪水预警,在河流洪水预警领域具有重要的应用价值,可为洪水灾害的防范提供理论与实践指导。

In order to improve the accuracy of seasonal river flood early warning and to prevent flood disasters,a method of seasonal river flood early warning based on data-driven model is proposed.The runoff curve method is used to calculate the confluence rate of riverbed,which can be taken as an auxiliary function.A seasonal river flood warning model is built by combining BP neural network and data-driven method.The time series of seasonal river floods are input into the model to obtain the seasonal river flood warning results.Taking the traditional seasonal river flood early warning method as the contrast object,it can be concluded through experiments that the seasonal river flood early warning method based on the data-driven model is superior to the traditional method in the early warning accuracy,and has highly reliable warning results.Thus,it can provide accurate flood early warning for seasonal rivers,and has important application value in river flood early warning,which also provide theoretical and practical support for flood disaster prevention.