Implementation of Univariate Paradigm for Streamflow Simulation Using Hybrid Data-Driven Model: Case Study in Tropical Region

• Published in: IEEE access Vol. 7; pp. 74471 - 74481
• Main Authors: Yaseen, Zaher Mundher, Mohtar, Wan Hanna Melini Wan, Ameen, Ameen Mohammed Salih, Ebtehaj, Isa, Razali, Siti Fatin Mohd, Bonakdari, Hossein, Salih, Sinan Q, Al-Ansari, Nadhir, Shahid, Shamsuddin
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive systems; Artificial neural networks; Biomimetics; evolutionary algorithm; Evolutionary algorithms; Evolutionary computation; Forecasting; Fuzzy logic; Genetic algorithms; Geoteknik; Mathematical models; Optimization; Particle swarm optimization; Predictive models; River flow; Rivers; Root-mean-square errors; Soil Mechanics; Stream flow; Streamflow forecasting; Streamow forecasting; tropical; tropical environment; Tropical environments; uncertainty analysis
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2019.2920916

The performance of the bio-inspired adaptive neuro-fuzzy inference system (ANFIS) models are proposed for forecasting highly non-linear streamflow of Pahang River, located in a tropical climatic region of Peninsular Malaysia. Three different bio-inspired optimization algorithms namely particle swarm optimization (PSO), genetic algorithm (GA), and differential evolution (DE) were individually used to tune the membership function of ANFIS model in order to improve the capability of streamflow forecasting. Different combination of antecedent streamflow was used to develop the forecasting models. The performance of the models was evaluated using a number of metrics including mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (<inline-formula> <tex-math notation="LaTeX">R^{2} </tex-math></inline-formula>), and Willmott's Index (WI) statistics. The results revealed that increasing number of inputs has a positive impact on the forecasting ability of both ANFIS and hybrid ANFIS models. The comparison of the performance of three optimization methods indicated PSO improved the capability of ANFIS model (RMSE = 7.96; MAE = 2.34; <inline-formula> <tex-math notation="LaTeX">R^{2}=0.998 </tex-math></inline-formula> and WI = 0.994) more compared to GA and DE in forecasting streamflow. The uncertainty band of ANFIS-PSO forecast was also found the lowest (±0.217), which indicates that ANFIS-PSO model can be used for reliable forecasting of highly stochastic river flow in tropical environment.