A comparative study of population-based optimization algorithms for downstream river flow forecasting by a hybrid neural network model

摘要: Population-based optimization algorithms have been successfully applied to hydrological forecasting recently owing their powerful ability of global optimization. This paper investigates three algorithms, i.e. differential evolution (DE), artificial bee colony (ABC) and ant (ACO), determine the optimal one for downstream river flow. A hybrid neural network (HNN) model, which incorporates fuzzy pattern-recognition a continuity equation into network, is proposed forecast flow based on upstream flows areal precipitation. The algorithm employed premise parameters HNN model. Daily data from Altamaha River basin Georgia in analysis. Discussions performances, convergence speed stability various are presented. For completeness' sake, particle swarm (PSO) included as benchmark case comparison performances. Results show that DE attains best performance generalization forecasting. accuracy comparable PSO, yet presents weak superiority over ABC ACO. Diebold-Mariano (DM) test indicates each pair has no difference under null hypothesis equal accuracy. ACO both favorable searching including recession coefficient initial storage. Further analysis reveals drawback slow time-consumption algorithm. present reliability with respect control whole. It can be concluded considerably more adaptive optimizing problem Comparison performances population-based flow.Differential (ACO).Particle performances.Hybrid model incorporating equation.DE