Development of forecasting model for sungai muda, kedah by utilizing artificial neural neywork (ann) / Nurul Hasniza Mohd Sopi

Faculty of Civil Engineering and Earth Resources

Project Paper (Bachelors of Civil Engineering) -- Universiti Malaysia Pahang – 2017

Bibliography : p. 71-72

This report deals with flood problem which is eventually happened in Malaysia when it coincides with monsoon and gave harm and damages of human life, as it had took many lives each time it happens. A case study of flood is going to be conduct to analyse the pattern of water level and determine other causes that contributes to the flood. The main aim of the study is to minimize the effect of the flood problems. It is also used to develop high accuracy model utilizing Artificial Neural Network (ANN) in predicting flood. Furthermore, it used to forecast flood occasion in the study area of station number of 5606410 of Sungai Muda (Jabatan Syed Omar) which is the main river that supplies water to Kedah and Penang. Besides, it used to investigate whether water level data alone can be used to produce modelling and to determine whether ANN is functioning in the forecasting. In this case study, a computational model will be used to stimulate the input data and generate the result, which is called Artificial Neural Network, ANN, which are modeled on the operating behavior of the brain, are brain, are tolerant of some imprecision and are especially useful for classification and function approximation or mapping problems, to which hard and fast rules cannot be applied easily. The terminology of artificial neural networks has created form an organic biological model of neural system, which it comprises an asset of joined cells, the neurons. The neurons receive impulses or response from either input cells or any other neurons. It will perform some kind of transformation of the input and then, it will transfer the outcome to other neurons or known as output cells. The neural networks are developed from many layers of connected neurons. The results with RMSE value of 38.414 for 1 hour interval time, while input 6+1 had the highest NSC value of 0.999. Besides that, with RMSE value of 78.692 for 5+1 input and had highest NSC value of 0.997 for 3 hour interval time. Lastly, with RMSE value of 205.404 for 6 hour interval, this time interval had highest value of NSC OF 0.997 for 4+1 input. In conclusion, this research contributes toward the development of forecasting using Artificial Neural Network for flood problem.