Zarezadeh, M., Bozorg Haddad, O. (2010). Inflow Simulation and Forecasting Optimization Using Hybrid ANN-GA Algorithm. , 24(5), -. doi: 10.22067/jsw.v0i0.5295
M. Zarezadeh; O. Bozorg Haddad. "Inflow Simulation and Forecasting Optimization Using Hybrid ANN-GA Algorithm". , 24, 5, 2010, -. doi: 10.22067/jsw.v0i0.5295
Zarezadeh, M., Bozorg Haddad, O. (2010). 'Inflow Simulation and Forecasting Optimization Using Hybrid ANN-GA Algorithm', , 24(5), pp. -. doi: 10.22067/jsw.v0i0.5295
Zarezadeh, M., Bozorg Haddad, O. Inflow Simulation and Forecasting Optimization Using Hybrid ANN-GA Algorithm. , 2010; 24(5): -. doi: 10.22067/jsw.v0i0.5295
Inflow Simulation and Forecasting Optimization Using Hybrid ANN-GA Algorithm
1Department of Water Resources Engineering, Tarbiat Modares University, Tehran, Iran
2Faculty of Agricultural Engineering and Technology, College of Agriculture & Natural Resources, University of Tehran, Karaj, Tehran, Iran
Abstract
Abstract
One of the major factors on the amount of water resources is river flow which is so dependent to the hydrologic and meteorologic phenomena. Simulation and forecasting of river flow makes the decision maker capable to effectively manage the water resources projects. So, simulation and forecasting models such as artificial neural networks (ANNs) are commonly used for simulation and predicting the exact value of such factors. In this research, the Dez River basin was selected as the case study. This paper investigates the effectiveness of temperature, precipitation and inflow factors and the lag time of those factors in inflow simulation and forecasting. Genetic algorithm (GA) has been thus used as an optimization tool, determining the optimum composition of the effective variables. Thus, in a flow simulation and forecasting model, the number of hidden layers, effective neurons in each layer, effective meteorologic and hydrologic parameters and also the lag time of each factor of flow simulation and forecasting has been considered as decision variables, and GA has been used to obtain the best combination of those variables. In this study, minimization of the total mean square error (MSE) has been considered as the objective function. Results show GA's effectiveness in flow simulation and forecasting with consistent accuracy. The value of R2 criterion has been obtained 0.86 and 0.79 in the simulation and forecasting models, respectively. The results also showed superiority replies obtained from the simulation model to the prediction model. One of the reasons for this superiority can be considering the meteorological factors in the current month in river flow simulation.
