Study On Joint Optimization Operation For Inter-basin Water Diversion And Supply:An Example From Shenzheng Western Diversion And Water Supply Reservoirs

With recent rapid growth of social economy and urban population,local water resources can not meet the city's water demands.The issue of water shortages is a serious problem for most large urban development.The water shortages mainly affected by development of urban city,season and water quality.In order to support the sustainable urban development,systematic research of urban water supply is an important uway to break through water supply bottleneck.For urban water supply,inter-basin diversion is the artificial method to increase water resources;this method is "open source" engineering measure.However,optimal operation of local water resources is "cutting" non-structural measure.The reasonable operation rules for water diversion and supply can effectively manage the local water resources and diversion water resources.Then the goals for optimal allocation,multi-source complementary,and development security are achieved.Thus,the reservoirs optimal operation researches for inter-basin water diversion and water supply have significance theoretical and practical value.The research results can provide important information for decision maker.In this paper,Shenzhen western diversion and water supply reservoirs are taken as study case.In this study,the research is mainly focus on four aspects,e.g.,medium-term inflow forecasts,water supply and distribution rules for multi-reservoirs supply water to one user,efficient algorithm for high dimensional and operation rule selection.The main research results obtained by the paper are as follows:In water supply and diversion reservoirs system,the relations of water supply are usually more complex and the inflows in future are unknown,the reservoirs are usually uncoordinated in water supply operation.Thus,the BP neural network is applied to construct medium-range inflow forecast model.And ensemble precipitation forecasts from European centre for medium-range weather forecasts(ECMWF)are used to forecast the medium-range ensemble inflows.The simulation results demonstrate that the average forecasts of ensemble inflows have high accuracy.Based on the medium-range inflow research results,the joint optimization operation for water diversion and supply in Gongming project is studied.Firstly,the aggregation method is applied to aggregate reservoirs in Gongming project into a "virtual reservoir",and joint operation rules are established.Then,the minimum amount of water diversion and maximum amount of water exchange are taken as operation objectives,and the multi-objective genetic algorithm(NSGA-II)is applied to optimize the operation model parameters.Finally,the solution set of operation rules are obtained.For water supply operation of multi-reservoirs supply water to one user,the reservoirs are aggregated into a "virtual reservoir",and the water supply rules based on "virtual reservoir" are defined and optimized.According to the rules,the amount of suppling water is determined.However,the amount of water is supplied by sveral reservoirs,and the distribution models are constructed and compared.The results show that the distribution model has higher efficiency by considering inflow forecasting information and water storage.For water supply operation of multi-reservoirs supply water to one user,the water supply rules are constructed by two methods,i.e.,aggregation method and seting operation rule in one of the reservoirs.In this section,the "virtual reservoir" is aggregated by combing inflow and storage.The joint optimation rules are constructed based on the "virtual reservoir".In this way,the dimension of the joint optimation rules is reduced,and the rules have a concise structure.In the joint operation rules,a large number of parameters need to be optimized.Conventional optimization algorithms are prone to premature convergence,and get local optimal solution.In this study,the Pareto technology and multi-dimensional dynamic search algorithms are combing to generate multi-objective multi-dimensional dynamic search algorithm(M-DDS).In M-DDS algorithm,a sensitivity analysis method is used to evaluate the sensitivity of parameters.Thus,the algorithm has a strong search capability.Comparing with NSGA-II algorithm,the M-DDS algorithm,based on sensitivity analysis method,can converge quickly,and higher precision.In this paper,the decision-making and operation rule selection model for multi-objective and multi-attribute is constructed based on Vague set theory.Based on this model,the Pareto solutions from M-DDS algorithm are analyzed and sorted.According to the sorting of solutions,the reasonable solution is selected.The research results provided a strong theoretical support for reservoirs operation of water diversion and water supply.Finally,a summary is given and some problems to be further studied are discussed.