BP神经网络在洪水地区组成随机模拟中的应用

Application of BP Network Time-Sequence Models to the Integration of Flood Stochastic Simulation

将BP人工神经网络引入随机模型中建立BP网络模型,将其应用于澜沧江流域中下游的洪水地区组成研究中。通过模拟澜沧江流域戛旧站和戛旧站与允景洪站区间的日流量过程,进行实例分析和计算,并与传统自回归模型进行比较。结果表明该模型在澜沧江流域中下游洪水地区组成随机模拟中的精度较高。在研究的同时也发现,由于序列白噪声可能带来的误差,还需进一步研究如何更加合理的将该模型用于洪水地区组成随机模拟当中,以便能对生产实践提供更多的选择方案。

In this paper a Back-Propagation （BP） Network Time-Sequence Model, which ameliorates the traditional stochastic model with ANN functions, is proposed. This newly proposed model is used to analyze the flood spatial pattern of the middle and lower reaches o{ the Lantsang River and the experiment results are also compared with the real historical record. But how to compactly integrate this method with the stochastic simulation of the flood spatial pattern can tremendously accelerate the development of the hydrology industry.