基于BP、Elman、PSO-SVR三种预报模型在石羊河流域的应用比较

详细信息    查看全文 | 推荐本文 |

英文篇名：Comparison and Application of Three Prediction Models Based on BP, Elman and PSO - SVR in Shiyang River Basin 作者：雷莉 ; 王超 英文作者：LEI Li;WANG Chao;Shiyang River Basin Administration of Gansu Provincial Water Resources Department;China Institute of Water Resources and Hydropower Research; 关键词：石羊河流域 ; 西营水库 ; 主成分分析 ; 径流预报 英文关键词：Shiyang River Basin;;Xiying Reservoir;;principal component analysis;;runoff forecast 中文刊名：中国农村水利水电 英文刊名：China Rural Water and Hydropower 机构：甘肃省石羊河流域水资源局;中国水利水电科学研究院; 出版日期：2019-09-15 出版单位：中国农村水利水电 年：2019 期：09 语种：中文; 页：32-36 页数：5 CN：42-1419/TV ISSN：1007-2284 分类号：P338

摘要

针对石羊河流域开展水量调度对中长期径流预报的迫切需求,以西营水库为研究对象,在分析石羊河流域径流特性的基础上,构建基于主成分分析法筛选预报因子的BP、Elman和PSO-SVR三种预报模型对西营水库进行年径流预报。结果表明,基于主成分分析的Elman和PSO-SVR中长期径流预报模型在率定期和检验期的合格率均满足相关规定对作业预报模型的精度要求,可为石羊河流域中长期径流预报提供实际支撑,为石羊河流域开展水资源优化配置和水量调度提供依据。
        Aiming at the urgent demand of medium and long term runoff forecast for water regulation in Shiyang River Basin and based on the analysis of runoff characteristics of Shiyang River Basin, this paper takes the Xiyanyang Reservoir as the research object and introduces 130 remote factors such as the atmospheric circulation index as the forecasting factors. Principal component analysis is used to optimize the factors, and the mid-long-term runoff forecast model based on physical cause analysis is combined with BP, Elman and PSO-SVR respectively to predict the annual runoff of Xiying Reservoir. The results show that the Elman and PSO-SVR medium and long-term runoff forecast models based on the principal component analysis meet the accuracy requirements of operation forecasting model in "hydrological information forecasting rules", which can be used as a support model for mid-long term runoff forecasting in Shiyang River Basin, providing the basis for optimizing the allocation of water resources and water regulation in Shiyang River Basin.

引文

[1] 桑燕芳,王中根,刘昌明.水文时间序列分析方法研究进展[J].地理科学进展,2013,32(1):20-30.
    [2] 王琪,张亭亭,游海林,等.基于多元回归分析的大伙房水库径流中长期预报[J].水力发电,2014(5):17-20.
    [3] 白庆芹,邢君杰.基于组合回归方法的尼洋河流域中长期径流预报[J].水电能源科学,2016(12):53-55,24.
    [4] 刘佩瑶,郝振纯,王国庆,等.新安江模型和改进BP神经网络模型在闽江水文预报中的应用[J].水资源与水工程学报,2017,28(1):40-44.
    [5] 王丽学,杨军,孙靓,等.基于灰色系统与RBF神经网络的中长期水文预报[J].人民长江,2015,46(17):15-17.
    [6] 汤成友,刘晔.基于小波分析的最近邻抽样回归预报模型研究[J].人民长江,2011,42(6):53-56.
    [7] 杨继鉝,邵利民,郭超,等.混沌理论在水文系统中的应用分析及海浪预报探讨[J].海洋技术,2008(1):96-100.
    [8] 胡昌军.基于改进相空间重构原理的支持向量机月径流模拟[J].水资源与水工程学报,2013,24(4):210-216.
    [9] 雷杰,彭杨,纪昌明.基于改进灰色-周期外延模型的中长期水文预报[J].人民长江,2010,41(24):28-31.
    [10] 纪昌明,俞洪杰,阎晓冉,等.基于联合互信息的水文预报因子集选取研究[J].水力发电学报,2017,36(8):12-21.
    [11] 屈亚玲,周建中,刘芳,等.基于改进的Elman神经网络的中长期径流预报[J].水文,2006(1):45-50.
    [12] 农振学,王超,雷晓辉.基于主成分分析和BP神经网络的赣江流域中长期径流预报[J].水电能源科学,2018(1):16-19.
    [13] 李薇,周建中,叶磊,等.基于主成分分析的三种中长期预报模型在柘溪水库的应用[J].水力发电,2016(9):17-21.
    [14] 吴超羽,张文.水文预报的人工神经网络方法[J].中山大学学报(自然科学版),1994(1):79-90.
    [15] Vapnik V.The Nature of Statistical Learning Theory[M].New York:Springer,1995.