基于多层神经元网络的随机自适应径流预报模型

FORECAST OF ADAPTIVE STREAMFLOW USING MULTILAYER NEURAL NETWORKS

针对水文系统的非线性特性，提出了一种基于多层神经元网络的自适应径流预报模型，并对其预报机理进行了分析．该预报模型由两个同构的多层神经元网络──训练网络和预报网络实现，实例验证了本文方法的有效性．从而为解决水文系统的径流预报提供了一条可行的途径．

This paper develops a novel multilayer neural networks based on adaptive streamflow forecast technique for complicated nonlinear water systems, and the forecast mechanism using multilayer neural networks is analyized. Two isomorphic multilayer nets are used together to implement the streamflow forecast model. One is called learning network while the other is called forecast network. It provides a new way to deal with nonlinear streamflow forecast problems of water systems.