Prediction of 4 hours ahead flood water level using improved ENN structure: Case study Kuala Lumpur

Recently, ANN models have been successfully applied in flood water level prediction system. However, most of publication on flood prediction only focusing on flood modelling and no element of prediction time was mentioned. Therefore, flood water level prediction is a new avenue to embark on in order to give early warning for evacuation purposes. This paper proposeda 4 hours ahead flood water level prediction using Improved ENN structure for Kelang River station which is located at Petaling Bridge, Kuala Lumpur. The model was developed using data obtained from the Department of Irrigation and Drainage, Malaysia upon special request. The prediction results of the original Elman neural network structure indicate unsatisfactorily performance results. Therefore, the Improved ENN structure was introduced. The performance indices results concluded that Improved ENN model was more versatile than the original ENN model and significant improvement from the original ENN model can be observed when the Improved ENN was introduced.