Development of whale optimization neural network for daily water level forecasting

Abstract

Development of water level forecasting model is essential in flood prediction and water resources planning and management. Through accurate water level forecasting models, high efficiency in the usage of water resources as well as minimization of flood damage with proper management of future development can be achieved. Therefore, the objective of this paper was set to develop a novel artificial neural network (ANN) for predicting the water level of Batu Kitang river via the implementation of a metaheuristic algorithm, Whale Optimization Algorithm (WOA). WOA was used to train and optimized the ANN. To compare the reliability of Whale Optimization Neural Network (WONN) in predicting the water level at Batu Kitang river, WONN is compared against a conventional neural network, Levenberg-Marquardt Neural Network (LMNN). The predicted water level showed that WONN outperformed LMNN in various evaluation criterion. However, inaccurate predictions occurred on both WONN and LMNN, which shows that further improvements are required to boost the prediction performance.