A Hybrid Model For Reservoir Operation Optimization Problem Using Inflow Forecasting Approach (Case Study:Dez Dam)

In this paper, a hybrid model is proposed to solve a reservoir operation optimization problem considering uncertain inflow conditions. In this model, the artificial neural network (ANN) and improved particle swarm optimization (IPSO) algorithm are used for inflow forecasting and reservoir operation optimization problem, respectively. The proposed IPSO is developed after applying some useful modifications to the original form of particle swarm optimization (PSO) algorithm. The modifications are proposed in order to reposition the infeasible particles. Two different conditions are considered in order to show the effect of inflow forecasting on reservoir operation optimization problem using ANN. It is worth noting that the ANN model is a powerful data driven model that can be used for real time inflow forecasting. In this research, in the first case, the actual measured inflow values are considered as input data to solve reservoir operation optimization problem. In the second case, the ANN model is used to forecast inflows while considering the effects of previous months inflows on the target month inflow. After determining the inflow, the reservoir operation optimization problem is solved using the forecasted inflows. In addition, in the proposed hybrid model, two different formulations are suggested to solve the optimization problem considering water release and reservoir storage volume as decision variables of the problem. The simple and the hydropower operation problems of Dez dam reservoir are solved for forecasted (5 year) time period considering all formulations and cases and the results are presented and compared with other available results. The results indicate the ability of ANN model to forecast the inflow of the Dez dam with acceptable accuracy. In addition, the improved particle swarm optimization algorithm shows to be an effective algorithm to solve reservoir operation optimization problem in which the results of first formulation is better than the second one.