عنوان مقاله [English]
Flood routing is one of the most complex problems that is investigated in open channel hydraulics and river engineering. It can help design engineers to recognize the impacts of riverine projects. Among the different flood routing methods, the Muskingum model, as the best hydrologic method of flood routing, has been widely used with high accuracy in river flood studies. The Parameters estimation of the nonlinear Muskingum flood-routing model has been considered by different researchers and several methods have been utilized to this purpose. In this paper, the DragonFly Algorithm (DA) was used to this end. The algorithm applies the penalty function for avoiding the negative values of output and storage and to find the global optimum values regardless of the parameter’s initial values in a quick convergence manner. The results of the DragonFly Algorithm (DA) were compared with GA and HS algorithms. The results showed that the DragonFly algorithm (DA)) are capable to provide satisfactory estimates of nonlinear Muskingum parameters. For this purpose, the proposed method was first used in Wilson River flood routing, after which the flood flow analysis of the Kardeh River was investigated. In the case of the Wilson flood which the o function was considered as observational and computational discharge the sum squares deviations (SSQ), the value of the objective function of DA was equal to 128/7861.
The results showed that the DragonFly Algorithm (DA) can provide an appropriate estimation of the optimal values of nonlinear Maskinging model parameters, so that for the sum squares deviations (SSQ) and RMSE the values for rainfed algorithm are 4/5551 and 0/711 respectively for the DragonFly Algorithm (DA). The DragonFly Algorithm (DA) can be used for any continuous engineering problem. Also, this algorithm are superior to other primitive supersonic algorithms, such as the Harmonic Search (HS)،Genetic Algorithm (GA). the higher accuracy and speed in controlling the optimal values of the nonlinear coefficients of the Maskingem method.