摘要
BP神经网络与K-最近邻(KNN)算法相耦合所建立的BK(BP-KNN)模型是一种数据驱动模型,它克服了传统的BP神经网络模型必需前期实测流量、泛化能力不强的缺点。IHACRES模型是一种结构简单、应用广泛的以单位线为基础的集总式概念性模型。选择板桥、马渡王两个流域分别运用BK模型、IHACRES模型和新安江模型进行径流模拟。模拟结果表明,BK模型的模拟效果最好,IHACRES模型次之;说明数据驱动模型在水文模拟中有着巨大的运用空间。
The BK （BP-KNN） coupling model constituted by BP neural network model and K-nearest neighbor algorithm is a kind of data-driven model, which can overcome the shortage of the necessary of real-time forecasting discharge and the weakness of generalization ability in traditional BP neural network model. The IHACRES model which is widely applied and simple in structure is a kind of lumped conceptual model based on Unit Hydrograph （UH）. The BK model, IHACRES model and Xin＇anjiang model are applied to simulate the runoffs of Banqiao Watershed and Maduwang Watershed respectively. The results show that the BK model has a best simulation performance and the IHACRES model is second. It illustrates that the data-driven model has a ~reat application space in hydrology simulation.
出处
《水力发电》
北大核心
2013年第12期9-12,共4页
Water Power
基金
国家自然科学基金资助项目(41130639
51179045
41201028)
水利部公益项目(201201058-1)
江苏省普通高校研究生科技创新计划资助项目(CXZZ11_0435)
