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

作　　者：黄晓敏[1] 雷晓辉[2] 王宇晖[1] 蒋云钟[2] 

机构地区：[1]东华大学环境科学与工程学院,上海201620 [2]中国水利水电科学研究院流域水循环模拟与调控国家重点实验室,北京100038

出　　处：《人民长江》2012年第2期16-21,共6页Yangtze River

基　　金：国家自然科学基金创新研究群体基金项目(50721006);"十一五"国家科技支撑计划重大项目(2008BAB29B08)

摘　　要：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.

关 键 词：水文模型 多目标参数优化 遗传算法 非支配排序 

分 类 号：P334.92[天文地球—水文科学]