人工神经网络方法在黄河下游平滩流量计算中的应用  

Application of the method of artificial neural network in calculating bankfull discharge in the Lower Yellow River

作　　者：李文文[1,2] 吴保生[2] 夏军强[2] 邵景力[1] 

机构地区：[1]中国地质大学(北京)水资源与环境学院,北京100083 [2]清华大学水沙科学与水利水电工程国家重点实验室,北京100084

出　　处：《泥沙研究》2010年第3期17-23,共7页Journal of Sediment Research

基　　金：中荷战略科学联盟计划项目(2008DFB90240)资助

摘　　要：利用黄河下游1960-2007年实测水沙资料,选取花园口、高村、艾山及利津4个断面作为研究对象,采用人工神经网络方法,通过输入来流量和来沙系数确定各断面的平滩流量。不同输入因子的计算结果表明:对于平滩流量的计算,不仅应考虑来水来沙条件的累积作用,同时还应考虑汛期与非汛期的全部水沙条件,则能更加准确地计算各断面汛后的平滩流量。人工神经网络方法不仅能够全面地考虑多种因素对平滩流量计算的影响,而且还能够模拟平滩流量随水沙条件变化的动态调整过程。The bankfull discharge is an important indicator for the flood discharging capacity in the Lower Yellow River. It is a task to accurately predict bankfull discharges at key sections before each experiment of regulating water and sediment in the Yellow River. In this paper, the observed hydrological data from 1960 to 2007 in the Lower Yellow River were used to establish the relationships between flow discharge, incoming sediment coefficient and bankfull discharge at four hydrometric sections by using the artificial neural network. Model predictions of different input factors show that the model with the consideration of flow discharges and incoming sediment coefficients during flood and nonflood seasons can obtain relatively higher calculation accuracy. The artificial neural network method （ANN） has the advantage that all related factors can be fully accounted for in the model and it can be used to simulate the dynamic process of bankfull discharge.

关 键 词：黄河下游 平滩流量 人工神经网络方法 来沙系数 

分 类 号：TV147[水利工程—水力学及河流动力学]