Water Demand Prediction Based On BP-Neural Network And Optimal Allocation Of Water Resources In Jinzhong

Based on the current situation of regional water resources,the future supply and demand of water resources are predicted in Jinzhong City,Shanxi Province.On the basis of the predicted results,the allocation of water resources with different guarantee rates in 2025 and 2030 are optimized.Rationality of the research results are finally analyzed to provide a scientific basis for water resources planning of Jinzhong.The main research contents and results of this article are as follows:(1)The social,economic,physical and geographical conditions and the situation of water resources of Jinzhong City are analyzed.With the year of 2018 as the actuality year,the prediction of available water supply under different guarantee rates for each planning level year(2025,2030)are analyzed and calculated.(2)The improved BP-neural network is used to forecast the water demand.Its forecast results are compared with those obtained by other methods(quota method,trend extrapolation,grey model prediction).The results show that the improved BP-neural network is more accurate and more accord with the actual situation of the local society than other methods in predicting water demand.Combining gray correlation analysis with BP-neural network can effectively reduce the complexity of the network by filtering the influence factors of water demand.Selecting the advantageous Levenberg-Marquardt algorithm as the core algorithm of the model can speed up the network convergence speed.The model prediction accuracy is high,and the results are accord with expectations.(3)The surplus and shortage of water resources under different guarantee rates in the planned level year 2025 and 2030 are calculated and analyzed.(4)A multi-objective optimization model with economic,social and ecological benefits as the goals are es Tablelished to optimize the allocation of water resources.Relevant parameters are analyzed and determined,and the Matlab optimization toolbox is used to solve the multi-objective programming problem.The results of water resources allocation under different guarantee rates in Jinzhong in 2025 and 2030 are obtained.The rationality of the results is analyzed,to give reasonable suggestions based on the future water supply and use conditions.The results show that the research methods adopted and results obtained in this thesis are in according with the law of local social development.The results of water resources allocation can be applied to the actual water resources development planning of Jinzhong City,which is in line with the research objectives.