基于NSGA-Ⅱ的水文模型参数多目标优化研究

Study of multi-objective optimization of parameters of HYMOD model using NSGA-Ⅱ algorithm

为了对水文模型中难以直接测算的参数进行调试和优化,将带精英策略的非支配排序遗传算法(NSGA-Ⅱ)应用于水文模型(HYMOD)参数多目标优化计算中,得到最优解Pareto集合。通过多目标距离函数法从Pareto集中求出一组协调集。采用非支配解集覆盖度和非支配解的空间分布两个性能度量指标,对NSGA-Ⅱ算法与多目标粒子群算法(MOPSO)的优化结果进行比较分析。结果表明,NSGA-Ⅱ优化得到的非支配集比MOPSO算法得到的支配比例高;但前者的非支配解的空间分布较MOPSO算法相对均匀。

In order to debug and optimize the parameters of the hydrological model （HYMOD） that can not be calculated di- rectly, an application of elitist non - dominated sorting genetic algorithm （ NSGA -Ⅱ） for multi - objective optimization of param- eters of HYMOD is presented. The Pareto set of optimal solution is obtained, from which, the multi -objective distance function is adopted to identify a compromise solution. The coverage ratio and spatial distribution of non - dominated solution are used to analyze and compare the optimization results by NSGA -Ⅱ algorithm and multi -objective particle swarm optimization （MOPSO） algorithms. It is shown that the dominating ratio of non - dominated solutions of NSGA - II is larger than that of MOPSO; however, the non - dominated solutions of NSGA - Ⅱ are more homogeneously distributed than that of MOPSO.