FORECASTING OF RUNOFF HYDROGRAPH USING NEURAL NETWORK ALGORITHMS

초록

In this paper, the runoff hydrographs in the stream are simulated using the neural network models. In the models, Back-Propagation(BP) algorithm using Levenberg-Marquardt (LM) techniques and Radial Basis Function Network(RBFN) theory are applied and the neural hydrographs at the Bocheong stream in Keum river basin in Korea are simulated to indicate the applicability of the neural network models. The models simulate the runoff hydrographs successfully using the two different algorithms of RBFN and BP algorithms with better results in the case of the RBFN algorithm. The models using RBFN and BP algorithms can be applied in the simulation of the runoff in the unlearned watershed.