Research On Optimization Of Water Allocation And Multi-Level Partitioning Algorithm For Multi-Source Water Supply System

With the increasing demand for urban water supply and the expansion of the scale of the pipe network,the multi-source water supply system has gradually become the main water supply mode of the city.Due to the complex relationship between the water plant and the pipe network,in actual production and operation,because the impact of water plant cannot be quantified,water quantity scheduling and management is more extensive,relying on experience and repeated exploration,it has theoretical and practical value to carry out scientific decision-making and refined management of multi-source water supply.In this paper,cluster analysis,optimization algorithm,graph theory,and other methods are used to study the theoretical and practical problems of water interaction influence,block division,water allocation optimization after water plant reconstruction and expansion,and multi-level partitioning management in the multi-source water supply system.Explore hydraulic tools,analyze pipe networks,strive to simplify complex water supply problems,and promote the application of information technology in engineering practice,the main work content is as follows:1)The water supply block segmentation and cross-district water transfer in multisource pipe network was studied:the water supply range of water plants was analyzed based on particle forward tracking;According to the proportion of water sources for the node,the improved two-point K-means clustering is used to divide water supply blocks,and the water supply division mode of independent control area and common influence area is proposed.Based on the results of water supply block division,the water transfer capacity of boundary pipelines is analyzed,and the identification method of cross-regional important pipelines is proposed;According to the exchange of water volume between blocks,the water transfer efficiency between water supply blocks is proposed,which provides a rapid analysis method for easy engineering promotion.2)Aiming at the water allocation after the reconstruction and expansion of the water plant in the multi-source pipe network,the optimization model of water quantity scheduling of the multi-source pipe network is established:the problem is analyzed according to the Pareto optimization and the law of diminishing marginal utility,and an optimization model with the lowest overall energy consumption is established,which is solved by hybrid local search.Taking G City as the object,the impact of different quantity of water plants involved in regulation and increased production requirements on the change of overall energy consumption was explored.It is found that when a single water plant increases production,(about 1～2)water plants in the common impact area or adjacent independent control area can achieve good results by participating in water allocation,which greatly simplifies the complexity of production allocation3)The multi-level partitioning method of water supply system was explored:the key pipeline judgment index of water supply pipe network was proposed,the widthfirst search algorithm was used to simplify the pipe network,and the main pipe network was extracted for subsequent pipe network partitioning.Combined with graph theory and clustering ideas,a multi-level partitioning method for water supply network based on control point and stack structure is proposed:preliminary partitioning based on plant-pump relationship and complete partitioning based on diversion.As an example,the multi-level partitioning algorithm is verified,and the influence of river crossing points and diversion conditions on the partitioning results is explored,and the configuration can be flexibly selected according to needs in practical applications.