期刊文献+

基于分段LassoLarsCV算法的水位预测研究--以三峡库区长寿站为例 被引量:2

Research on water level prediction based on piecewise LassoLarsCV algorithm:case of Changshou hydrological station in Three Gorges Reservoir Region
下载PDF
导出
摘要 三峡水库蓄水以来,由于入库流量统计和地形资料测量等存在系统性误差,通过求解圣维南方程组计算的长寿站水位精度有时难以满足实际应用需求。对此,采用分段套索最小角回归交叉验证(Lasso Least angle regression Cross Validation,LassoLarsCV)算法预测长寿站水位。首先收集了2009~2019年相关水位和流量数据并进行标准化处理,然后根据坝前水位和入库流量对数据进行分类;最后针对每一个类别,从2009~2018年的数据中随机抽取80%用于训练线性模型,剩余20%数据用于检验并调整分段方案,以2019年的数据进行测试,采用LassoLarsCV算法模型求解。结果表明:采用LassoLarsl算法得到的长寿站水位误差可以控制在1.2 m以内,99.47%的水位误差在0.5 m以内,效果优于传统水动力学方法。LassoLarsCV可直接研究出、入库流量和坝前水位与长寿站水位的映射关系,降低了模型对输入数据准确性的要求,同时提高了长寿站水位的计算精度。 Since the impoundment of the Three Gorges reservoir,due to the systematic errors in statistics or measurement of inlet flow and topographic data,the accuracy of water level at Changshou station calculated by Saint Venant equations is sometimes hard to meet the practical application requirements.In this paper,the piecewise LassoLarsCV algorithm is used to study the water level prediction of Changshou hydrological station.The method directly studies the mapping relationship between the inflow and outflow discharge and the water level in front of the dam to the water level of Changshou station,which reduces the accuracy requirement of the input data and improves the calculation accuracy of the water level of Changshou station.Firstly,the data from 2009 to 2019 are collected and standardized,and then the data are classified according to the values of water level and inflow flow in front of the dam.Finally,for each category,80%of the data from 2009 to 2018 are randomly selected to train the linear model,and the remaining 20%is used to test and adjust the segmentation scheme,and then tested with 2019 data.The model is solved by LassoLarsCV algorithm.The results of the experiment show that the water level error of Changshou station can be controlled within 1.2 m and 99.47%within 0.5 m,which is better than the hydrodynamic method.The model directly studies the relation between inflow,discharge flow and dam water level of Three Gorges Reservoir and the level of Changshou station,reducing the requirement on accuracy of input data,also it improves the calculation accuracy of this station.
作者 刘涛 周曼 胡挺 纪国良 LIU Tao;ZHOU Man;HU Ting;JI Guoliang(River Basin Hub Administration,China Three Gorges Corporation,Yichang 443100,China)
出处 《人民长江》 北大核心 2021年第5期60-65,共6页 Yangtze River
基金 “十三五”重点研发计划项目“水库库区淤积对防洪的影响研究”(2016YFC0402306-01) 中国长江三峡集团有限公司自主科研项目“基于多方法的三峡水库回水规律研究”(WWKY-2020-0451)。
关键词 水位预测 LassoLarsCV算法 圣维南方程组 线性模型 水动力学 三峡库区 water level prediction LassoLarsCV algorithm Saint Venant equations linear model hydrodynamics Three Gorges Reservoir area
  • 相关文献

参考文献7

二级参考文献41

共引文献28

同被引文献25

引证文献2

二级引证文献2

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部