论文部分内容阅读
为了解决感潮河网地区水动力模型水位边界资料较短的问题,基于圣维南方程组定解问题分析,利用BP神经网络建立了缺资料地区短系列水位资料的智能延展模型。模型以邻近水文站水位作为输入,以缺资料地区水位作为输出,利用Levenberg-Marquardt算法对缺资料地区短系列水位资料进行学习,利用邻近水文站提供的长系列水位资料对缺资料区短系列水位资料进行延展。以某三叉感潮河道为例进行计算,结果表明,延展值与实际值的线性相关系数为0.944,相关性较好,达到工程应用的精度。该模型能够充分利用已有资料,获得足够的建立水动力模型所需的资料,以解决感潮河网地区建模边界资料较短的问题。
In order to solve the problem of short data of hydrodynamic boundary of water level in tidal river network, based on the analysis of fixed solution of Shengweinan equation set, an intelligent extension model of short series of water level data is established by using BP neural network. The model takes the water level of the adjacent hydrological station as input and the water level of the lacking data area as the output. Levenberg-Marquardt algorithm is used to study the short series of water level data in the lacking data area. Using the long series of water level data provided by the adjacent hydrological stations, Data is extended. Taking a trigeminal tidal channel as an example, the result shows that the linear correlation coefficient between the extension value and the actual value is 0.944, and the correlation is good, which achieves the accuracy of engineering application. The model can make full use of the existing data and obtain enough data to establish the hydrodynamic model to solve the problem of short modeling data in the tidal river network area.