Abstract—The aims of this study are to improve the model perforemance with input selection and comparision fast learning and automated learning algorithms. Therefore, the methodology was, first, investigation cross correlation, stepwise regression, cross correlation with stepwise regression, genetic algorithm, supervise and all input) then comparing the learning algorithms: Levenberg Marquardt-LM and Baysian Regularization-BR). The ANN was used to forecast the water level at a gauge station M.7. The results showed the selecting input from genetic algorithm gives the best result for forecasting the flood peak and BR provided better results than LM particulaly at the flood peak.

Index Terms—Neural network, flood forecasting, Levenberg-Marquardt, Bayesian Regularization, Mun Basin.

Tawee Chaipimonplin is with the Department of Geography, Chiang Mai University, Chiang Mai, Thailand (e-mail: tawee.c@cmu.ac.th, chaipimonplin@hotmail.com).
Thaveesak Vangpaisal is with the Civil Engineering Department, Ubon Ratchathani University, UbonRatchthani, Thailand (e-mail: thaveesak.v@bub.ac.th).

[PDF]

Cite:Tawee Chaipimonplin and Thaveesak Vangpaisal, "Comparison of the Efficiency of Input Determination Techniques with LM and BR Algorithms in ANN for Flood Forecasting, Mun Basin, Thailand," International Journal of Computer and Electrical Engineering vol. 6, no.2, pp. 90-94, 2014.