A comprehensive study on predicting river runoff
2017 9th International Conference on Knowledge and Systems Engineering (KSE) - Trang 251-256 - 2017
Tóm tắt
In this study, MIKE NAM, artificial neural networks (ANNs), and a hybridization of ANNs and Particle Swarm Optimization (ANN-PSO) are utilized to predict the Dak Nong runoff. ANNs are trained by the back-propagation (BP) procedure [1] which is based on the gradient descent algorithm and an incorporating algorithm of PSO and BP [2]. Moreover, to improve the performance of ANNs, a common method of time series analysis, so-called partial autocorrelation function (PACF), is collaboratively used. The experimental results are conducted on a dataset collected in the Dak Nong basin for the duration of 1981–2007. The experiments demonstrate that PACF significantly impacts the performance of ANNs. Although ANN-PSO is not superior to ANNs trained by BP (ANN-BP) in this study, ANN-PSO outperforms ANN-BP in terms of capturing the peaks of the Dak Nong runoff. In addition, both ANN-BP and ANN-PSO outperform MIKE NAM.
Từ khóa
#Artificial Neural Networks #Particle Swam Optimization #MIKE NAM #River RunoffTài liệu tham khảo
frank, 1958, The perceptron: A probabilistic model for information storage and organization in the brain, cornell aeronautical laboratory, Psychological Review, 65, 386, 10.1037/h0042519
haykin, 2009, Neural Networks and Learning Machines
sarangi, 2013, A hybrid differential evolution and back-propagation algorithm for feed forward neural network training, International Journal of Computer Applications, 84, 641
tawatchai, 2000, Application of tank, nam, arma and neural network models to flood forecasting, Hydrological Processes, 14, 2473, 10.1002/1099-1085(20001015)14:14<2473::AID-HYP109>3.0.CO;2-J