摘要
参数识别是水文模型应用的前提,在参数识别的同时需要充分考虑模型自身及参数的不确定性。引入了一种具有较强全局与局部搜索能力的多目标优化算法(MOCOM—UA),探讨了该算法在多目标基础上融合遗传算法和单纯形法两类算法不同搜索机制的模型优化方案,并以汉江上游江口流域新安江模型降水—径流模拟实践为例,将MOCOM—UA算法应用于新安江模型的参数识别,得到了4目标函数情形下的Pareto参数空间和模型的预测范围,并根据该算法模拟计算的结果,初步分析了参数和模型的不确定性。
Parameter determination is the prerequisite for the simulation of rainfall and runoff in a hydrological model. Model and parameter uncertainty should be considered during that procedure. This paper started with an introduction of the theory on the multi-objective complex evolution （MOCOM-UA） algorithm of excellent global and local optimization capability for dealing with the complex problem of hydrological model calibration. Based on the Xin＇anjiang model, different searching mechanisms between the multi-objective approach based method and genetic and simplex algorithm were respectively discussed by rainfall-runoff simulations for the Baohe watershed in the upper reaches of the Hanjiang River basin. With the application of the multi-objective genetic algorithm coupled with the simplex method as an optimal solution, the Pareto space of parameters and the prediction extent of the model were calculated and analyzed under four objective function conditions. The uncertainties of the model and its parameters were preliminarily investigated.
出处
《水土保持通报》
CSCD
北大核心
2008年第3期107-112,共6页
Bulletin of Soil and Water Conservation
基金
国家重点基础研究发展规划项目(2006CB400502
2001CB309404)
中国科学院“百人计划”择优支持项目(8-057493)
中国科学院大气物理研究所东亚区域气候-环境重点实验室开放基金
关键词
参数识别
水文模型
多目标
全局优化
遗传算法
单纯形法
新安江模型
parameter determination
rainfall-runoff model
multi-objective
global optimization
genetic algorithm
simplex method
Xin＇anjiang Model
