Modeling of flood water level prediction using improved RBFNN structure

Recently, the applications of Artificial Neural Network (ANN) in various hydrologic problems have becoming popular. This is due to ability of ANN models to estimate nonlinear functions and hence become important tools to solve diverse water resources problems. Particularly, ANN models have been used in hydrological fields such as river flow forecasting, rainfall-runoff estimation, flood prediction and water quality prediction. Therefore, this paper proposed flood water level prediction model using Radial Basis Function Neural Network (RBFNN) and Improved RBFNN structure that using the water level data from Kelang river which is located at Jambatan Petaling, Kuala Lumpur. The models were developed by processing offline data over time using neural network architecture. The methodologies and techniques of the two models were presented in this paper and comparison of the long term runoff time prediction results between them were also conducted. The prediction results of the Radial Basis Function Neural Network architecture indicate fair performance for the one hour ahead of time prediction. The performance indices results also concluded that the Improved RBFNN model was more reliable than that of the original RBFNN model.