期刊文献+

六种预测模型在北京市城市生态环境用水短期预测中的比较 被引量:7

Comparison of six prediction models for short-term prediction of urban eco-environmental water use in beijing
下载PDF
导出
摘要 针对城市水资源短缺和城市生态环境需水研究的迫切性,选用6种预测模型开展城市生态需水预测研究。为探究适合北京市城市生态环境用水预测的模型,以中国城市统计年鉴2004—2019年的北京市相关统计数据为基础,在时间序列预测方法上比较了灰色模型、分数阶灰色模型,在系统分析预测方法上比较了多元线性回归、主成分回归、BP神经网络和灰色神经网络方法。总结并归纳了城市生态环境用水量的概念和计算方法,分析和选取6种不同模型的输入变量,分别开展模型的训练和测试模拟,2004—2016年为训练期,2017—2019为测试期。通过交叉检验得出各类模型预测结果的得分,结果显示分数阶灰色模型得分为0.081 4,优于其他5种模型的得分,说明该模型在六种方法中对北京市城市生态环境用水短期预测的适用性较好。利用分数阶灰色模型对北京市未来2 a的城市生态环境用水进行预测,2021和2022年生态环境用水量分别为19.79亿m^(3)和22.16亿m^(3)。北京市的城市生态环境用水在未来短期时间内将保持高速增长的趋势。 Aiming at the urgency of the study on the urban water resources shortage and eco-environmental water demand, 6 mathematical models are selected for carrying out the prediction of the urban eco-water demand. In order to explore a model suitable for the prediction of the urban eco-environmental water use in Beijing and on the basis of the relevant statistical data of Beijing between 2004—2019 from China Urban Statistical Yearbook, the multi-gray model and fractional gray model are compared in the aspect of time series prediction method, while the multiple linear regression, principal component regression, BP neural network and gray neural network are compared in the aspect of the systematic analysis prediction method. The concept and calculation method of the amount of the urban eco-environmental water use are summarized, and then the input variables of the six different models are analyzed and selected to carry out the model trainings and testing simulations respectively with the training period of 2004—2016 and the testing period of 2017—2019. The prediction scores of all kinds of the models are obtained through the relevant cross validations and the results show that the score from the fractional gray model is 0.081 4 and better than those of the other five models, which indicates that within the six methods, this method has a better applicability for the short-term prediction of the urban eco-environmental water use in Beijing. The prediction of the urban eco-environmental water use in Beijing obtained from the fractional gray model shows that the eco-environmental water uses in 2021 and 2022 are to be 1.979 billion m^(3) and 2.116 billion m^(3)respectively;that is to say, the urban eco-environmental water use in Beijing is going to maintain a trend of rapid increase within a short-period of time in the days to come.
作者 黄天意 周晋军 李雅君 李英杰 HUANG Tianyi;ZHOU Jinjun;LI Yajun;LI Yingjie(Faculty of Architecture,Civil and Transportation Engineering,Beijing University of Technology,Beijing 100124,China)
出处 《水利水电技术(中英文)》 北大核心 2022年第3期119-133,共15页 Water Resources and Hydropower Engineering
基金 北京市自然科学基金青年项目(8214046) 清华大学水沙科学水利水电工程国家重点实验室及宁夏银川水联网数字治水联合研究院联合开放研究基金资助课题(sklhse-2021-Iow07) 国家重点研发计划项目(2016YFC0401401,2017YFC1502701)。
关键词 城市生态环境用水 需水预测 分数阶灰色模型 多元线性回归模型 BP神经网络模型 灰色神经网络模型 水资源 urban eco-environment water use water demand prediction fractional grey model multiple linear regression model BP neural network model grey neural network model water resources
  • 相关文献

参考文献15

二级参考文献168

共引文献280

同被引文献119

引证文献7

二级引证文献19

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