混合门限回归模型在河道水位预报中的应用
发布时间：2018-09-17 13:13
【摘要】：针对松花江干流汛期洪水的特点以及松花江流域防洪减灾的需求,采用多元门限回归模型建立了松花江干流肇源、三家子、涝洲、木兰、富锦5个水位站的水位预报模型;在多元门限回归模型的基础上进行改进,得到混合门限回归模型,并以此建立松花江干流5个站的水位预报模型。两种模型的预报因子均通过AIC准则和DW检验法筛选确定,并用最小二乘法估算模型的参数。选取各水位站2008—2012年汛期的水位资料分别率定相应的水位预报模型,选取2013年汛期的水位资料对各个率定的模型进行验证。率定和验证的结果表明:多元门限回归模型的预报精度偏低,而混合门限回归模型的预报精度高,且有一定的通用性,适用于水位预报。
[Abstract]:In view of the characteristics of flood in flood season of Songhua River and the demand of flood control and disaster reduction in Songhua River basin, the water level forecast model of 5 water level stations of Songhua River mainstream, Sanjiazi, waterlogging Island, Mulan and Fujin is established by using multivariate threshold regression model. Based on the multivariate threshold regression model, the mixed threshold regression model is obtained, and the water level prediction model of 5 stations in Songhua River is established. The prediction factors of the two models are determined by AIC criterion and DW test, and the parameters of the model are estimated by the least square method. The corresponding water level prediction models are selected from the water level data of the flood season from 2008 to 2012, and the water level data of the 2013 flood season are selected to verify the models. The results of rate determination and verification show that the prediction accuracy of multivariate threshold regression model is low, while that of mixed threshold regression model is high and universal, so it is suitable for water level prediction.
【作者单位】： 河海大学水文水资源与水利工程科学国家重点实验室;黑龙江省水文局;
【基金】：国家重点研发计划项目(2016YFC0402704) 水文水资源与水利工程科学国家重点实验室专项经费项目(1069-514031112)
【分类号】：P338
本文编号：2246027
[Abstract]:In view of the characteristics of flood in flood season of Songhua River and the demand of flood control and disaster reduction in Songhua River basin, the water level forecast model of 5 water level stations of Songhua River mainstream, Sanjiazi, waterlogging Island, Mulan and Fujin is established by using multivariate threshold regression model. Based on the multivariate threshold regression model, the mixed threshold regression model is obtained, and the water level prediction model of 5 stations in Songhua River is established. The prediction factors of the two models are determined by AIC criterion and DW test, and the parameters of the model are estimated by the least square method. The corresponding water level prediction models are selected from the water level data of the flood season from 2008 to 2012, and the water level data of the 2013 flood season are selected to verify the models. The results of rate determination and verification show that the prediction accuracy of multivariate threshold regression model is low, while that of mixed threshold regression model is high and universal, so it is suitable for water level prediction.
【作者单位】： 河海大学水文水资源与水利工程科学国家重点实验室;黑龙江省水文局;
【基金】：国家重点研发计划项目(2016YFC0402704) 水文水资源与水利工程科学国家重点实验室专项经费项目(1069-514031112)
【分类号】：P338
【相似文献】
相关期刊论文 前4条
1 程银才;范世香;;基于门限回归的稳定水位流量关系曲线定线方法[J];水文;2013年04期
2 金菊良,魏一鸣,丁晶;投影寻踪门限回归模型在年径流预测中的应用[J];地理科学;2002年02期
3 马志瑾;;多元门限回归短期径流预报模型在黄河上游的应用[J];甘肃水利水电技术;2012年06期
4 ;[J];;年期
，本文编号：2246027
本文链接：https://www.wllwen.com/kejilunwen/shuiwenshuili/2246027.html
