基于改进主成分分析法的城市需水量预测  

Prediction of urban water demand based on improved Principal Component Analysis method

作　　者：徐继红 XU Jihong(Xinjiang Tarim River Basin Xiner Reservoir Administration Bureau, Korla 841000, China)

机构地区：[1]新疆塔里木河流域希尼尔水库管理局,新疆库尔勒841000

出　　处：《水资源开发与管理》2019年第3期23-25,共3页Water Resources Development and Management

基　　金：水利部公益性行业科研专项资助(201501059)

摘　　要：为提高城市需水量预测精度,提出了基于主成分分析和长短时记忆神经网络的城市需水量预测模型。本文利用该模型对新疆阿克苏市城市需水量进行验证。结果表明:与BP神经网络等模型相比,该模型具有良好的预测性能和泛化能力,能够满足城市需水量精确预测的需要,可以为干旱区城市水资源精准调控提供参考。A prediction model of urban water demand based on PCA and LSTM is proposed in order to improve the prediction accuracy of urban water demand. The model is used to verify the water demand of Aksu in Xinjiang. The results show that the model has good prediction performance and generalization ability compared with BP model, and it can meet the needs of accurate prediction of urban water demand, and can provide reference for accurate regulation of urban water resources in arid areas.

关 键 词：城市需水量 预测 主成分分析 LSTM神经网络 

分 类 号：TU991.31[建筑科学—市政工程]