摘要
为深入对比研究典型参数优化算法在新安江模型中的应用情况,选用4种典型优化算法:自适应遗传算法(AGA)、改进粒子群算法(IPSO)、SCE-UA和贝叶斯优化算法(BOA),每种算法重复运行50次,每次迭代300次,以安徽省黄山市呈村流域为研究区域对日尺度新安江模型参数进行率定。结果表明:IPSO优化新安江模型参数得到的流量模拟值与实测值拟合程度高且收敛速度较快;目标函数收敛值集中,有90%集中在0.158149~0.156727范围内;算法稳定性好,优化后的参数方差均值仅为0.049404。AGA和SCE-UA优化参数表现较IPSO差,较BOA好。BOA计算量小,但其收敛过程出现明显波动;目标函数收敛值分散在5个范围内;算法稳定性差,优化后的参数方差均值高达0.073751。
In order to compare and study the application of typical parameter optimization algorithms in Xin'anjiang model in detail,four typical optimization algorithms were selected:Adaptive Genetic Algorithm(AGA),Improved Particle Swarm Algorithm(IPSO),SCE-UA,and Bayesian Optimization Algorithm(BOA).Each algorithm was operated independently 50 times,with 300-time-iteration.The parameters of the Xin'an River model were calibrated in the Chengcun Watershed of Huangshan City,Anhui Province.The results show that the simulated and measured flow values obtained by IPSO optimization of Xin'anjiang model parameters have a high degree of fit and a fast convergence speed;the convergence values of the objective function are concentrated,and 90%are concentrated in the range of 0.158149-0.156727;the algorithm has good stability and is optimized for the mean variance of the parameters is only 0.049404.The optimized parameters of AGA and SCE-UA are worse than IPSO and better than BOA.The amount of BOA calculation is small,but its convergence process fluctuates significantly;the convergence value of the objective function is scattered in 5 ranges;the algorithm has poor stability,and the mean value of the optimized parameter variance is as high as 0.073751.
作者
向鑫
敖天其
肖钦太
XIANG Xin;AO Tianqi;XIAO Qintai(College of Water Resources and Hydropower,Sichuan University,Chengdu 610065,China;State Key Laboratory of Hydraulics and Mountain River Engineering,Sichuan University,Chengdu 610065,China)
出处
《水文》
CSCD
北大核心
2023年第3期16-22,共7页
Journal of China Hydrology
基金
四川省科技计划重点研发项目(2021YFS0285)
科技部国家国际科技合作专项(2012DFG21780)。
关键词
新安江模型
自适应遗传算法
改进粒子群算法
SCE-UA
贝叶斯优化算法
Xin'anjiang model
Adaptive Genetic Algorithm
Improved Particle Swarm Optimization
SCE-UA
Bayesian Optimiza-tion Algorithm