CD Tesis
Komparasi Model Peramalan Debit Sungai Menggunakan Ann – Som Ann Pada Sub Das Tapung Kiri
Tapung Kiri Sub river Basin is part of siak river basin which is categorized as a critical basin river and a flood prone area. The potential for flooding is caused by high rainfall in the middle, upstream and along basin river that have undergone land use changes. So it is necessary to do debit forecasting to find out when floods will occur in the future.
There have been many hydrological models that have been developed for this discharge prediction using both physical and softcomputing models. One of the softcomputing models that have received attention in hydrological modeling recently is the Artificial Neural Network model. The backpropagation algorithm is an algorithm commonly used in the ANN method to solve problems related to forecasting. However, the drawback of this algorithm is the determination of the initial random weight so that the network training is not optimal, so the Self Organizing Map is used so that the result obtained are more optimal.
This research is expected to provide an overview of which method has more advantages for use as an early warning system.The result showed that the SOM-ANN method had a correlation coefficient of 0,987, better than the ANN method which has a coefficient value of 0,897 in the network formation and modeling process. The use of the SOM – ANN method is more stable than the ANN method because in the SOM – ANN method the data comes from the source ( AWLR ) in the cluster first so that the data pattern becomes good for training, testing and validation.
Keywords : Artificial Neural Network, correlation coefficient, discharge forecasting, Self Organizing Map
Tidak tersedia versi lain