首页|PA-DDS算法在HBV模型参数优化中的应用

PA-DDS算法在HBV模型参数优化中的应用

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
  • 维普
影响水文模型预报精度的因素有很多,其中模型参数的优化对模拟结果起到至关重要的作用,目前用于参数优化的方法包括单目标优化和多目标优化两种.随着参数率定方法研究的深入,多目标分析问题越来越受到关注.Pareto存档动态维度搜索(Pareto-Archived Dynamically Dimensioned Search,PA-DDS)作为多目标优化算法,通过在求解过程中动态存储Pareto前沿以防止最优解的丢失,在寻优速度以及解的稳定性方面比较有优势.精英非支配排序遗传算法(NSGA-Ⅱ)具有寻优速度快、解集收敛性能好等优点,已经成为检验其他多目标优化算法性能的标准;AMALGAM算法通过对四种相关算法分配权重从而实现信息交换同时寻优,解的收敛性能较好.因此本文将PA-DDS算法与AMALGAM算法和精莫非支配排序遗传算法(NSGA-Ⅱ)在收敛性能方面进行了对比,并将非劣解分布的均匀性及解的相似性方面与AMALGAM进行比较,利用尼泊尔巴格玛蒂河流域2005-2011年期间实测洪水日径流过程资料作为HBV模型参数率定系列,运用PA-DDS算法对模型参数进行优化,得出Pareto最优解,并利用2013年5场洪水日径流过程进行模型检验.结果表明:PA-DDS算法比AMALGAM算法能够更快地得到Pareto最优解且解的质量较好,拟合历史洪水平均确定性系数达到0.86,模型预报精度高,表明PA-DDS优化算法在解决多参数多目标优化问题中具有优势.
The Application of PA-DDS Algorithm in Parameter Optimization of HBV Model
There are many factors affecting the accuracy of hydrological model forcasting.The optimization of model parameters plays an important role in simulation results.At present,the parameter optimization methods include two types of single objective optimization and multi-objective optimization.With the development of parameter calibration method,more and more attention has been paid to the problem of multi-objective analysis.As a multi-objective optimization algorithm,Pareto-Archived Dynamically Dimensioned Search (PA-DDS) has advantages like optimization speed and stability by dynamically storing the Pareto front in order to prevent the loss of optimal solution.Elitist non-dominated sorting genetic algorithm (NSGA-Ⅱ) has advantages like searching speed and convergence performance,so NSGA-Ⅱ has become a standard to test the performance of other multi-objective optimization algorithms.AMALGAM algorithm has better convergence performance by distributing the weight of four related algorithms to realize the information exchange PA-DDS algorithm is compared with AMALGAM algorithm and NSGA-Ⅱ algorithm in terms of convergence performance.The uniformity of Pareto distribution and the similarity of solutions are compared with AMALGAM.,PA-DDS algorithm is used to optimize the model parameters and obtain Pareto optimal solutions by using historical runoff data from 2005 to 2011 of Baghmati River Basin,finally tested the parameters calibration result byusing five floods in 2013.According to the result,PA-DDS algorithm can quickly get optimal solutions and the Pareto quality is better than AMALGAM algorithm,the average certainty coefficient of historical floods fitting reaches 0.86 and the forecasting accuracy of HBV model is high,it shows that the PA-DDS optimization algorithm has the advantage of solving the multi-objective optimization problem of multi-parameters.

HBV modelparameters optimizationPareto optimal solutioncertainty coefficientPA-DDS multi-objective optimization algorithmAMALGAM multi-objective optimization algorithmNSGA-Ⅱ multi-objective optimization algorithm

代旭、陈元芳

展开 >

河海大学水文水资源学院,南京210098

HBV模型 参数优化 Pareto最优解 确定性系数 PA-DDS多目标优化算法 AMALGAM多目标优化算法 NSGA-Ⅱ多目标优化算法

国家自然科学基金

51479061

2017

中国农村水利水电
水利部中国灌溉排水发展中心 水利部农村水电及电气化发展局 武汉大学 中国国家灌溉排水委员会

中国农村水利水电

CSTPCD北大核心
影响因子：0.655
ISSN：1007-2284
年,卷(期)：2017.(12)
  • 3
  • 5