摘要
以镇江市某小区雨水管为研究对象,基于水动力学机理模型,指导设定BP神经网络的输入与输出,并在Google的TensorFlow平台下进行程序设计以及神经网络的训练和验证,完成对管网下一时刻充满度的预测。结果表明,在一定边界条件下,BP神经网络能够较好地完成学习与训练,且预测误差较小,满足实际应用所需的精度要求。
Using the storm sewer data of a residential district in Zhenjiang,the input and output of BP neural network were set based on the hydrodynamic mechanism model,and the program design and neural network training and verification were conducted under Google's TensorFlow platform to predict the next moment capacity of the storm sewer.The results indicated that the BP neural network could complete the learning and training satisfactorily under certain boundary conditions.The forecast error was small,which could meet the requirement of practical application.
作者
盛政
王浩正
胡坚
SHENG Zheng;WANG Hao-zheng;HU Jian(Jiangsu Manjiangchun Urban Planning and Design Co.Ltd.,Zhenjiang 212000,China;Zhenjiang Housing and Urban -Rural Development Bureau,Zhenjiang 212000,China)
出处
《中国给水排水》
CAS
CSCD
北大核心
2018年第23期130-133,共4页
China Water & Wastewater