基于ConvAttLSTM-KF混合模型的城市洪涝实时校正方法研究
DOI:
作者:
作者单位:

1.中山大学土木工程学院、南方海洋科学与工程广东省实验室(珠海);2.黄河古贤水利枢纽有限公司;3.1、中山大学土木工程学院、南方海洋科学与工程广东省实验室(珠海),珠海,519082;4.2、广东省华南地区水安全调控工程技术研究中心,广州,510275;5.3、华南地区水循环与水安全广东普通高校重点实验室,广州,510275

作者简介:

通讯作者:

中图分类号:

P333;TP18

基金项目:

国家重点研发计划项目(2021YFC3001000),国家自然科学基金项目(U1911204),珠海市洪潮涝遭遇水情监测预报与群闸排涝优化调度研究(SML2023SP213)


Real-Time Correction for Urban Flooding Based on ConvAttLSTM-KF Model
Author:
Affiliation:

1.School of Civil Engineering, Sun Yat-sen University and Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai);2.Yellow River Guxian Water Conservancy Hub Co., Ltd.;3.1. School of Civil Engineering, Sun Yat-sen University and Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai 519082, China. 2. Guangdong Engineering Technology Research Center of Water Security Regulation and Control for Southern China, Sun Yat-sen University, Guangzhou 510275, China. 3. Key Laboratory of Water Cycle and Water Security in Southern China of Guangdong High Education Institute, Sun Yat-sen University, Guangzhou 510275, China.

Fund Project:

Acknowledgement: The research is financially supported by National Key R&D Program of China (2021YFC3001000), National Natural Science Foundation of China (Grant No. U1911204), Monitoring and forecasting of flood - tide - waterlogging and optimal regulation of flood drainage in Zhuhai City of China (SML2023SP213).

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    随着城市洪涝灾害频发,通过实时校正方法提高预报的精度成为城市防涝的关键。以广州市天河区车陂涌流域为研究对象,使用一种基于ConvAttLSTM-KF混合模型的实时校正方法对SWMM模型的预报结果进行校正并与传统卡尔曼滤波技术的校正结果对比。研究表明,ConvAttLSTM-KF模型在洪峰流量的误差范围小于卡尔曼滤波;在峰现时间误差方面,ConvAttLSTM-KF模型的平均误差为2.5min,而卡尔曼滤波为21.25min。ConvAttLSTM-KF模型的平均纳什效率系数为0.97,较卡尔曼滤波提高了5%。

    Abstract:

    With the increasing frequency of urban flooding, improving forecasting accuracy through real-time correction methods has become crucial for urban flooding control. Taking the Chebei River Basin in Tianhe District, Guangzhou as the study area, a real-time correction method based on the ConvAttLSTM-KF model was applied to correct the forecasting results of the SWMM model, and the correction results were compared with those of the Kalman filtering. The study shows that the ConvAttLSTM-KF model has a smaller error range in peak flow than the Kalman filtering. In terms of peak occurrence time error, the average error of the ConvAttLSTM-KF model is 2.5 minutes, while the Kalman filtering has an average error of 21.25 minutes. The ConvAttLSTM-KF model’s average Nash efficiency coefficient is 0.97, which is 5% higher than that of the Kalman filtering.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2024-09-27
  • 最后修改日期:2025-02-20
  • 录用日期:2025-02-21
  • 在线发布日期:
  • 出版日期: