基于SA-NSGA-Ⅱ算法的水库多目标优化调度研究

Study on Multi-objective Optimal Scheduling of Reservoirs Based on SA-NSGA-ⅡAlgorithm

针对带精英策略的快速非支配排序遗传算法NSGA-II求解多目标优化问题时存在局部搜索能力较差的缺陷,提出一种基于逐次逼近方法改进的快速非支配排序遗传算法(SA-NSGA-Ⅱ),该算法通过不断调整搜索空间来减小不可行域和支配解占比,从而增强局部搜索能力,快速逼近Pareto真实前沿。以平均出力最大和下游河道适宜生态流量改变度最小为目标,建立小浪底水库多目标优化调度模型,分别采用SA-NSGA-Ⅱ和NSGA-Ⅱ求解模型并比较优化效果。结果表明,在两种算法求解模型生成的混合Pareto前沿中,SANSGA-Ⅱ、NSGA-Ⅱ所生成的Pareto前沿点分别占比80.45%、19.55%;SA-NSGA-Ⅱ、NSGA-Ⅱ生成的Pareto前沿中,分别有86.90%、52.67%的点处于非支配地位,且SA-NSGA-Ⅱ较NSGA-Ⅱ的算法运算时间减少了15.32%。因此,在相同初始条件下,SA-NSGA-Ⅱ的优化效果优于NSGA-Ⅱ,验证了SA-NSGA-Ⅱ在水库多目标优化调度中的适用性。

Aiming at the defect of poor local search ability of NSGA-Ⅱ,a fast non-dominated sorting genetic algorithm with elite strategy,in solving multi-objective optimization problems,a fast non-dominated sorting genetic algorithm based on the improvement of the successive approximation method(SA-NSGA-Ⅱ)was proposed.The algorithm reduced the ratio of infeasible domains and dominated solutions by continuously adjusting the search space,so as to enhance the local search ability and quickly approximate the true Pareto frontier.With the goals of maximum average output and the minimum change of suitable ecological flow of the downstream river,a multi-objective optimal scheduling model of Xiaolangdi Reservoir was established.The SA-NSGA-Ⅱ and NSGA-Ⅱ were used to solve the model respectively and the optimization effects were compared.The results demonstrate that among the hybrid Pareto frontier generated by the two algorithms,the Pareto frontier points generated by SA-NSGA-Ⅱ and NSGA-Ⅱ account for 80.45% and 19.55%,respectively.Among the Pareto frontiers generated by SA-NSGA-Ⅱ and NSGA-Ⅱ,86.90% and 52.67% of the points are in the non-dominated position,respectively,and the algorithm operation time of SA-NSGA-Ⅱ is reduced by 15.32% compared to NSGA-Ⅱ.Therefore,under the same initial conditions,the optimization effect of SA-NSGA-Ⅱ is better than that of NSGA-Ⅱ,which verifies the applicability of SA-NSGA-Ⅱ in multi-objective optimal operation of reservoirs.