摘要
多种下垫面类型的出现,使得城市地区径流系数的变化日趋复杂.本文通过模拟降雨实验,分析了降雨强度等因素对径流系数变化的影响,运用LM-BP算法的神经网络模型,采用S型/S型模型结构,针对不同降雨条件下每种下垫面的径流系数进行了数学模拟,并与其他方法的模拟结果进行了比较.结果表明,该模型预测结果准确性较高.最后,以某办公楼为例采用该模型进行了产流量预测,验证了该模型的实用性.
The runoff coefficient change in time and space has become more and more complex because of the diversification of the underlying surface of the city. On the basis of experiments, the law of rainfall coefficient changed by a single factor is analyzed, the S/S mathematial model of relationship of different underling surfaces and influence factors is developed based on LM calculation method. The runoff coefficients of every underlying surface are forecasted in different rainfall conditions by using ANN models and comparison with other methods. The results show that the model is of high forecast precision. The runoff-producing amount in an office building is forecasted by using the model. A case study shows that the rain water resource has the value of reusing.
出处
《水利水运工程学报》
CSCD
北大核心
2008年第3期35-39,共5页
Hydro-Science and Engineering
基金
国家自然科学基金资助项目(编号40271022,50279041)
国家“863”计划研究资助项目(编号2005AA113150)
关键词
神经网络
径流系数
降雨强度
雨水利用
neural network
runoff coefficient
rainfall intensity
rainwater utilization
