基于优选伸缩比SCE-UA算法的新安江模型参数率定研究
Study on parameter calibration of the Xinanjiang model based on SCE-UA algorithm combined with optimal scaling factor
投稿时间:2023-08-03  修订日期:2024-02-22
DOI:
中文关键词:  SCE-UA  下山单纯形搜索  伸缩比  优选法  动态
英文关键词:SCE-UA  downhill simplex search  scaling factor  optimal selection  dynamic
基金项目:中国水利水电科学研究院十四五“五大人才”计划(JZ0199A032021);光合基金A类(ghfund202302018283);
作者单位邮编
侯宇 中国水利水电科学研究院 100038
阚光远* 中国水利水电科学研究院 100038
梁珂 北京中水科工程集团有限公司 
摘要点击次数: 204
全文下载次数: 0
中文摘要:
      洗牌复合形进化(SCE-UA)算法已被广泛应用于水文模型参数率定领域,该算法鲁棒性好、全局寻优能力较强。经典SCE-UA算法中的下山单纯形搜索采用固定伸缩比实现反射和收缩操作,其寻优效率还有待提升。实际应用中发现变动伸缩比能够改善寻优效率,结合数值实验与分析,发现伸缩比取值存在最优区间。针对以上问题,在伸缩比最优区间内均匀采样,利用优选法在单纯形搜索过程中动态确定最适伸缩比,提出了基于优选伸缩比的SCE-UA算法。利用改进的算法,基于人工降雨—径流资料开展了新安江模型参数率定研究。结果表明,相较于固定伸缩比,优选伸缩比方法显著提升了优化效率,具有推广应用价值。
英文摘要:
      The Shuffled Complex Evolution-developed at University of Arizona (SCE-UA) algorithm has been widely applied in the field of hydrological model parameter calibration. The algorithm exhibits good robustness property and strong global searching capability. The reflection and contraction operations in the downhill simplex search of the traditional SCE-UA algorithm adopt a fixed scaling factor and its searching efficiency has potential to be further improved. In real-world applications, we find that the searching efficiency can be enhanced by adjusting the scaling factor. According to numerical experiment and analysis, it is recognized that there is an optimal interval for the scaling factor. In order to cope with the above-mentioned problems, the improvement of the SCE-UA algorithm by combining optimal scaling factor is proposed which uniformly sampling the scaling factor values in a discrete manner within the optimal interval and dynamically selecting the optimal factor value during the process of parameter optimization. Studies on Xinanjiang model parameter calibration are carried out based on the synthetic rainfall-runoff data and the proposed optimization algorithm. The research results indicate that the improved algorithm, which incorporates optimal scaling factor, significantly improves the optimization efficiency compared to the utilization of a fixed value. This improvement holds valuable potential for future general purpose real-world applications.
  查看/发表评论  下载PDF阅读器
关闭