基于 RF－LSTM 模型的城市洪涝积水模拟研究

打开文本图片集

关键词：RF 模型；LSTM 模型；SWMM 模型；洪水模拟；市中图分类号：P333．2；TV213．9 文献标志码：A doi：10．3969 ／ j．issn．1000－1379．2025．06．008引用格式：．基于 RF－LSTM 模型的城市洪涝积水模拟研究［J］．人民黄河，2025，47（6）：50－56．

Study on Urban Flood Simulation Based on RF⁃LSTM Model

HUANG Chuwen， GUAN Yongle，WANG Hongfa（School of Water Conservancy and Transportation， Zhengzhou University， Zhengzhou ， China）

Abstract： As a result of the heavy rainfall， flooding remains in the later stages of the rainfall and may continue to cause harm and impact． In order to accurately predict the depth and duration of urban flooding and waterlogging， the RF⁃LSTM model was proposed to address the diffi⁃ culty of simulating floods in the later stages of the heavy rainfall． Based on the SWMM model⁃simulated flood data in Zhengzhou City， China， the flood depths at three representative flooded points were simulated by using the proposed model， and the flooding process caused by rainfall under different recurrence periods was predicted． The results show that compared to the single LSTM model， the simulation accuracy of the RF⁃LSTM model has been improved， verifying the applicability of the model in flood simulation． The growth rates of flood duration and the maximum flood depth at flooded points are the highest under the 1-2 a return period， therefore the existing drainage system should be renova⁃ ted or redesigned．

Key words： RF model； LSTM model； SWMM model； flood simulation； Zhengzhou City

0 引言

近些年来，随着全球气候变暖以及城市化进程加快，我国城市遭受洪涝灾害的频率和强度都呈现提高的趋势［1］。（剩余8623字）