新安江模型参数多目标优化研究  

Study on Multi-objective Parameter Optimization of Xinanjiang Model

作　　者：郭俊[1] 周建中[1] 邹强[1] 宋利祥 张勇传[1] 

机构地区：[1]华中科技大学水电与数字化工程学院,湖北武汉430074 [2]珠江水利科学研究院,广东广州510623

出　　处：《水文》2013年第1期1-7,26,共8页Journal of China Hydrology

基　　金：国家973重点基础研究发展计划项目(2007CB714107));水利部公益性行业科研专项(200701008);高等学校博士学科点专项科研基金(20100142110012)

摘　　要：水文模型的参数优化率定一直以来是水文预报领域的重要研究内容,当水文模型的结构确定后,水文模型参数的选择对水文模型整体性能和水文预报结果的好坏有着至关重要的影响。针对传统水文模型参数优选采用单一目标不能充分全面挖掘水文观测资料中蕴含的水文特征信息的缺陷,本文以新安江三水源模型为例,尝试采用多目标优化算法优化率定水文模型,算例应用分析表明,通过合理的选择目标函数的种类和数目,采用多目标进化算法优化率定模型参数,可以获得相对于单目标率定模型参数更优的结果。进一步,研究工作针对模型参数优化的结果进行分析,可以明显看出模型参数优化中存在"异参同效"现象,为后续模型参数不确定性分析等相关研究工作的开展做好了铺垫。Parameter estimation of hydrological models is an important matter of hydrological forecasting. As the structure of the model is established, the calibration of parameters has great influence on the performance of the hydrological model. Practice experience suggests that the conventional calibration of hydrological models with single objective function is often inadequate to properly measure all of the characteristic of the observed data deemed to be important. To deal with this defect, the multi-objective evolution algorithm was employed to optimize the parameters of the Xinanjiang model with three runoff components in this paper. The results of the case study indicated that with well chosen objective functions, the multi-objective optimization can achieve better results than the single objective optimization. Fur- thermore, by analyzing the achieved the parameter combination, it is obvious that the phenomenon of same effect of different parameters exists, so as to do some preparations for the uncertainty analysis of the model parameters.

关 键 词：水文预报 新安江三水源模型 多目标 NSGA-II 混沌差分进化算法 

分 类 号：P338[天文地球—水文科学]