A Hybrid Particle Swarm Optimization-Genetic Algorithm for Multiobjective Reservoir Ecological Dispatching

Reservoir ecological dispatching is a complex system problem involving multiple objectives, multiple criteria and multiple phases. This study established a multiobjective ecological dispatching model of the Yinma River Basin in Changchun city based on the water demand, socioeconomic development, river ecology, and constraints on reservoir characteristic parameters. Taking advantage of particle swarm optimization (PSO) and genetic algorithm (GA), a PSO-GA hybrid algorithm is proposed and applied to solve the schemes of ecological dispatching models considering different ecological flow requirements. The annual mean scheduling results show that the three scheduling schemes basically achieve the objectives of river ecological base flow scheduling. For ecologically suitable flows, the guaranteed rates for the RGOS1, RGOS2, and RGOS3 schedules at the Dehui section were 78.35%, 86.36%, and 95.98%, respectively, whereas the rates were 81.77%, 90.13%, and 96.57%, respectively, at the Nong’an section. The scheduling results of typical years show that the water security situation in the study area is not optimal, but the river ecological environment can be considerably improved by reservoir ecological dispatching. Finally, the excellent performance of the hybrid PSO-GA proposed in this study is verified via comparison with other algorithms. The Pareto front optimized by the PSO-GA can dominate the Pareto front solutions of the other algorithms. The IGD (0.19) of the Pareto front optimized by the PSO-GA is the smallest, and the SP (0.83) and HV (0.93) are the largest, indicating better convergence and comprehensive performance.