ABSTRACT
ABSTRACT: An accurate water stage forecasting is very important because it allows the pertinent department to issue a forewarning of the impending flood and to implement early evacuation arrangement when necessary. Existing methods including rainfall-runoff modeling or statistical techniques entail exogenous input together with a variety of assumptions. In general kinds of these models perform not so good in the Karst area for the runoff process is much more complicated than that in the other region. The recent researches show that the use of Artificial Neural Networks (ANN) has been approved to be a cost-effective technique. But their training, usually with back-propagation algorithm or other gradient algorithms, is featured with certain drawbacks such as very slow convergence and easy entrapment in a local minimum. In this paper, an Accelerated Particle Swarm Optimization (APSO) model is adopted to train RBF NN. The approach is applied to predict water levels in Houzhaihe River of Pu Ding. And the results indicate that the RBF model improved by APSO performs much better than the canonical RBF model.
