| With the acceleration of urbanization,smart city is gradually coming into people’s vision.Water supply engineering facilities,as an important part of urban infrastructure,needs to follow the pace of the times to develop smart water,and promote the informatization and intelligent development of water.The accuracy of water demand prediction is directly related to the design and planning of smart water supply and the planning and construction of water supply engineering facilities.Beijing-Tianjin-Hebei urban agglomeration is an acute shortage of water areas,although the diversion of south-to-north water diversion project in a certain extent,ease the pressure of water supply of the Beijing-Tianjin-Hebei urban agglomeration,but with the development of economy and the expansion of population,the water demand is still rising.The scientific and reasonable prediction of water demand is the premise and basis for the planning and construction of water supply projects in the future.In this thesis,the water demand prediction of Beijing-Tianjin-Hebei urban agglomeration in China is studied,and the gap between supply and demand of water resources in Beijing-Tianjin-Hebei urban agglomeration is clarified,which lays a foundation for the planning and construction of water supply projectsFirstly,the explanatory variables associated with economy,social,water use,and resources availability are identified.Eleven statistical and machine learning models are built,which uses data covering the 2004–2020 period of the Beijing-Tianjin-Hebei Urban Agglomeration.The interpolation and extrapolation scenarios are conducted to find the most suitable predictive model.Through the comparative analysis of the prediction results,it is verified that the Gradient Boosting Decision Tree model has the best prediction performance in the two scenarios.Secondly,the model is further tested for three other regions in China,and its robustness is validated.According to the optimal prediction model,key features affecting the water demand of Beijing-Tianjin-Hebei urban agglomeration are selected with the Recursive Feature Elimination technology.Seven key features are identified,reducing about one-third redundant features.And the GBDT model is used to predict the water demand of the Beijing-Tianjin-Hebei urban agglomeration in the next five years,and the water demand gap of the beijing-Tianjin-Hebei urban agglomeration is analyzed.Finally,based on the research content and results,rationalization suggestions are put forward for the future planning and construction of water supply projects in the Beijing-Tianjin-Hebei urban agglomeration,including the follow-up planning and construction of South-to-North Water Diversion Project,the planning and construction of green water-saving city and the optimization of water supply network.This thesis proves that machine learning models outperformed the statistical models,with the ensemble models being superior to the single predictors.The best predictive model can also be applied to other regions and help forecast water demand to ensure sustainable water resources management. |
