Optimal Parameter Estimation for Nonlinear Muskingum Model based on Artificial Bee Colony Algorithm

Parameter estimation of the nonlinear Muskingum model is a highly nonlinear optimization problem. Although various techniques have been applied to optimize the coefficients of the nonlinear Muskingum flood routing models, but an efficient method for this purpose in the calibration process is still lacking. The accuracy of artificial bee colony (ABC) algorithm is investigated in this paper to optimize the coefficients of nonlinear Muskingum model. The performance of this algorithm was compared with other optimization techniques. For evaluating the ability of the ABC algorithm, several statistical criteria such as sum of the square error, sum of the absolute error, mean absolute error and mean relative error were used in the present study. ABC is an intelligent algorithm, which can effectively overcome the prematurity and slowed convergence speed of the traditional evolution algorithms. It determines the best parameter values in terms of the sum of square residual between the observed and routed outflows. The simulation results show that the performance of ABC algorithm with the sum of the square of the deviations between the computed and observed outflows (SSQ) of 35.62 m3 s-1, the sum of the absolute value of the deviations between the computed and observed outflows coefficients (SAD) of 23.2 m3 s-1, the mean absolute errors between the routed and observed outflows (MAE) of 1.05 m3 s-1 and the mean relative errors between the routed and observed outflows (MRE) of 2.9% is comparable to those of other algorithms. Thus ABC provided an efficient way for parameter optimization of the nonlinear Muskingum model.