[1]桂晗亮,张春萍,武治国,等.人工神经网络和SWMM在降雨径流模拟中的应用对比[J].中国给水排水,2021,37(13):108-112.
　GUI Han-liang,ZHANG Chun-ping,WU Zhi-guo,et al.Comparison of Artificial Neural Network and SWMM Applied in Rainfall Runoff Simulation[J].China Water & Wastewater,2021,37(13):108-112.
点击复制

人工神经网络和SWMM在降雨径流模拟中的应用对比

中国给水排水[ISSN:1000-4062/CN:12-1073/TU] 卷: 第37卷 期数: 2021年第13期 页码: 108-112 栏目: 出版日期: 2021-07-01
Title:
Comparison of Artificial Neural Network and SWMM Applied in Rainfall Runoff Simulation
作者:
桂晗亮1，张春萍1,2，武治国2,刘翀2
(1.武汉工程大学光电信息与能源工程学院，湖北武汉430000；2.武汉新烽光电股份有限公司，湖北武汉430000)
Author(s):
GUI Han-liang1,ZHANG Chun-ping1,2,WU Zhi-guo2,LIU Chong2
(1. School of Optical Information and Energy Engineering, Wuhan Institute of Technology, Wuhan 430000, China; 2.Wuhan Newfiber Optics Electron Co. Ltd., Wuhan 430000, China)
关键词:
降雨径流模拟; 径向基神经网络; 小波神经网络; SWMM
Keywords:
rainfall runoff simulation; radial basis neural network; wavelet neural network; SWMM
摘要:
降雨径流的精确模拟一直是水文研究领域的重点问题之一，为探究人工神经网络在城市降雨径流模拟中的适用性，选取西宁市某小区为研究对象，以当前降雨量、累积降雨量和前一时段的流量为输入，以当前流量为输出，分别建立径向基神经网络和小波神经网络进行模拟，并与SWMM的模拟结果进行对比。另外，通过SWMM输出结果训练神经网络替代模型。结果表明，径向基神经网络和小波神经网络对降雨径流的模拟精度较高，能够真实模拟实际径流过程，适用于复杂的城市降雨径流过程模拟。另外，证实了径向基神经网络和小波神经网络模型作为SWMM替代模型的适用性。
Abstract:
Accurate simulation of rainfall runoff has always been one of the key issues in the field of hydrology. In order to explore the applicability of artificial neural network in urban rainfall runoff simulation, radial basis neural network model and wavelet neural network model were established for simulation by using current rainfall, cumulative rainfall and flow rate in the previous period as input and the current flow rate as output, and the simulation results were compared with those of SWMM in a community in Xining City. In addition, the SWMM output was employed to train the neural network models. The radial basis neural network and wavelet neural network had high accuracy in simulation of rainfall runoff, and could simulate the actual runoff process, which was suitable for complex urban runoff process simulation. In addition, the applicability of radial basis neural network and wavelet neural network models as alternative models of SWMM was verified.
更新日期/Last Update: 2021-07-01