Evolutionary techniques versus swarm intelligences: application in reservoir release optimization

Abstract

In this paper, a nonlinear reservoir release optimization problem has been solved by using four optimization tools with various combinations of input parameters that are generally used in this research field. A comparison has been made between evolutionary methods [genetic algorithm (GA)] and swarm intelligences [particle swarm optimization (PSO) and artificial bee colony (ABC) optimization] in searching the optimum reservoir release policy. From the historical recorded data, the monthly inflow was categorized into three states: high, medium and low. As a guideline for the decision maker, an optimum release curve was generated for each month showing the release options with a variety of different storage conditions. GA (real and binary), ABC optimization and PSO algorithm have been used as optimization tools with the same formulation and objective function for all the methods. For verification of the models, a simulation is done by using 264 monthly historical inflow data. Different indices such as reliability, vulnerability and resiliency were calculated in order to check the performance and risk analysis purposes. The results show that the most recently developed ABC optimization technique provides the best results in meeting demands, avoiding wastage of water and in handling critical period of low flows.