系统理论水文模型结构与参数多目标优化  

Structure optimization and parameter calibration of empirical hydrological model under multi-objective framework

作　　者：郭俊[1] 周建中[2] 王浩[2,3] 邹强 

机构地区：[1]国网湖南省电力公司电力科学研究院,长沙410007 [2]华中科技大学水电与数字化工程学院,武汉430074 [3]中国水利水电科学研究院水资源研究所,北京100044 [4]长江水利委员会勘测规划设计院,武汉430000

出　　处：《水力发电学报》2014年第2期1-7,共7页Journal of Hydroelectric Engineering

基　　金：高等学校博士学科点专项科研基金(20100142110012);水利部公益性行业科研专项(201001080)

摘　　要：本文提出了一种水文模型结构和模型参数同时优化的方法,克服了经验确定模型结构方法未考虑模型结构与模型参数的交互影响且难以得到最优模型结构的缺陷;同时,针对传统基于单一目标优化率定模型参数不能全面反映水文情势多方面特性的不足,引入了多目标优化思想,将模型结构和模型参数优化置于多目标优化框架下,有效解决了单目标优化导致的模型预报误差均化效应。通过流域日径流预报的实验验证,结果表明本研究方法可以获得较优的模型结构和模型参数,为水文模型参数和结构优化提供了一种新的途径和方法。A novel method is presented to optimize the structure and parameters of hydrological model in a single run. This method can overcome the shortcoming of traditional method in determining the model structure, i.e. being unable to consider the interaction of structure parameters for an optimal structure. The use of single objective function for parameters calibration by traditional method is not suitable for hydrological system, showing another shortcoming. The new method adopts a multi-objective for the optimization and it was verified through daily runoff forecasting by a hydrological model based on support vector machine. The results show that the new method is superior in its structure and model parameters and it would provide a new approach to optimization and calibration of hydrological model.

关 键 词：水文学 径流预报 多目标优化 模型结构与参数优化 支持向量机 

分 类 号：TV121[水利工程—水文学及水资源]