基于BP神经网络模型的河道洪水反向演算研究  

Study on Flood Inverse-routing in River Channel Based on BP Neural Network

作　　者：刘欢[1] 陆宝宏[1,2] 陆建宇[1] 朱从飞[1] 臧冬伟 甄亿位 左建[1] 

机构地区：[1]河海大学水文水资源学院,江苏南京210098 [2]河海大学水文水资源与水利工程科学国家重点实验室,江苏南京210098

出　　处：《水电能源科学》2016年第3期52-54,共3页Water Resources and Power

基　　金：国家自然科学基金项目(50979023);水利部公益性科研专项项目(201201026);中国博士后科学基金资助项目(2013M531270);江苏省博士后基金资助项目(1302029C)

摘　　要：河道洪水反向演算在河库联合调度和推求设计洪水的区间组成等方面具有重要意义,直接使用传统的马斯京根法反演效果较差。基于河道上下断面洪水的非线性关系,采用BP神经网络模型训练历史洪水,并用训练得到的网络对下游洪水进行反演。实例应用结果表明,该模型反演结果与实际洪水过程更为接近,具有一定的精度和实用性。The flood inverse-routing of river channel is significant for joint dispatching of river and reservoir and calculating section composition of design flood.The result directly calculated by Muskingum model is not satisfying.Based on the nonlinear relationship between the upstream and downstream floods in river channel,the historical floods were trained by using BP neural network.With the obtained neural network,the downstream floods were inverse-routed.The application results show that the inversion results of the model are close to the actual flood process,and the model has certain precision and applicability.

关 键 词：洪水反演 BP神经网络 河道 非线性关系 

分 类 号：P338[天文地球—水文科学] TP183[水利工程—水文学及水资源]