摘要
将枯水期河段槽蓄水量表示为上断面多个时刻流量的线性组合,考虑河段引水、加水和损失等因素的影响,建立了枯水流量演进的线性方程模型,并用优化方法率定了模型参数;以河段上断面流量、河段用水量、区间加水量、河段损失量作为网络输入因子,河段下断面流量作为网络输出因子,构建了一个3层BP神经网络模型;将线性方程模型、BP神经网络模型和水力学方法应用于黄河下游河段枯水流量演进模拟计算,并对其模拟结果进行了比较.结果表明:线性方程模型稳定性好,便于演进计算和反馈控制计算;BP神经网络模型具有较好的泛化能力,计算精度也较高,但难以进行反馈控制计算;水力学方法在进行流量演进和反馈控制时,存在计算稳定性和收敛性问题,其应用有一定困难.
According to linear combination of different discharges during the low water period, a linear equation model for discharge routing of the low water period was established. The model parameters were calibrated by use of the optimization method. Then, a three-layer BP neural network with a hidden layer was developed, with the discharge of the upper profile, water consumption, regional water recharge, and water loss of the river section taken as the input factors, and the discharge of the lower profile taken as the output of the network. Finally, the hydrodynamic method was adopted for simulating the discharge process. Through comparison of the calculated results of the above three methods, some conclusions were drawn： the linear equation model is of high stability, and it is convenient for discharge routing and feedback control calculation; the artificial neural network is of high capacity of generalization and high accuracy of calculation, but it is difficult for discharge feedback control calculation; the hydrodynamic method is not a practical method due to some problems on stability and convergence when it is used for discharge routing and feedback control.
出处
《河海大学学报（自然科学版）》
CAS
CSCD
北大核心
2006年第3期267-271,共5页
Journal of Hohai University(Natural Sciences)
关键词
线性方程模型
BP神经网络模型
水力学方法
黄河下游河段
流量演进
反馈控制
model of linear equation
BP neural network model
hydrodynamic method
lower Yellow River
discharge routing
feedback control
