RIVER WATER LEVEL PREDICTION WITH ARTIFICIAL NEURAL NETWORK AND GENETIC ALGORITHM

• Main Author: Sulaiman Marwan, Puteradarma
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/60618
• Institution: Institut Teknologi Bandung

The information about water level of a river is important to the local organization for water management and natural disaster predictions. This urgency push water management local organization to demand a reliable water level prediction model. The purpose of this final project is to build an Artificial Neural Network (ANN) wi. The data was taken from an enviromental organisation website, RiverLevelsUK and Metoffice in Durham, England. The performance of ANN depend on the characteristic configuration of ANN itself. There is a lot of hyperparameters and variable optimization methods was done by researchers before training the model. With genetic algorithm, every possible ANN configurations goes through elimination and recombination process so the best ANN configuration can be found in the configuration population. The configuration in question includes the number of layers, the number of neurons per layer, activation functions, input variables, and so on. The performance of the model is measured by the prediction againts the actual value of the data test using MSE and NSE. It is hoped that this final project can be used as material for further research for prediction of river discharge in other places, the development of ANN and the Genetic Algorithm itself.