摘要
针对城市日用水量非线性变化问题,为实现水资源的优化调度和合理利用,提出一种深度学习的水量预测方法,建立多因素长短时神经网络模型预测日用水量。该方法选取影响日用水量的因素作为输入特征,日用水量时间序列数据作为训练样本,利用数据挖掘,输出用水量预测值。结合杭州示范区实际案例,与传统的人工神经网络方法进行对比,结果表明,长短时神经网络的预测结果优于传统的人工神经网络,并且基于多因素长短时神经网络模型的预测结果优于单因素长短时神经网络模型,预测结果具有较强的精度和稳定性。
For non-linear change in daily water demand,optimize the scheduling and rational use of water resources,this paper presents a method for predicting water depth of learning,a neural network model to predict daily water when the length of multiple factors.The daily water methods factors as the input vector,daily water consumption time series data as training samples,data mining,water coming day forecast water supply network.Compared with the traditional artificial neural network method and the actual case of the demonstration area in Hangzhou,the results show that the prediction results of long and short time neural networks are better than traditional artificial neural networks.The prediction results based on the multi-factor long-term neural network model are better than the single-variable long-term neural network model,and the prediction results have strong accuracy and stability.
作者
陆维佳
朱建文
叶圣炯
毛哲凯
信昆仑
Lu Weijia;Zhu Jianwen;Ye Shengjiong;Mao Zhekai;Xin Kunlun(College of Environmental Science and Engineering,Tongji University,Shanghai 200092 China;Hangzhou Water Holding Group Co.,Ltd.,Hangzhou 310009,China)
出处
《给水排水》
CSCD
北大核心
2020年第1期125-129,共5页
Water & Wastewater Engineering
