Multi-objective Optimal Design Of Water Supply Systems Via Improved SPEA2 Algorithm

Urban water supply system is an important infrastructure to guarantee the fast development of city.Water supply network,one of the important parts of urban water supply system,accounts for 70% to 80% of the total investment of the whole water supply system.With the constant construction of modern cities in China,water demand for production and residents in the city is increasing constantly.Underground water supply network of most cities can not meet the demand of users,while network construction has not caught up with the pace of urban construction.Simultaneously,the network operation time is too long and the network operation pressure is too heavy.For these reasons,leakage,pipe explosion and other accidents often occur in the pipeline system,which cause huge economic losses.In order to meet the goal of "constructing a resource-saving society",some research is done in this thesis,including water supply system modeling and optimization as well as intelligent water supply network software design.In this thesis,firstly,a multi-objective optimization model of water supply network is proposed,then an improved Strength Pareto Evolutionary Algorithm 2 based on Reference Vectors(RVSPEA2 for short)is designed which is used in the multiobjective optimization design of water supply network.Finally,a water supply network intelligent management system is developed.These works achieve relatively satisfactory results.The main works are as follows:1.In the existing work about water supply network modeling,only the network economy is considered while the network reliability is ignored or considered as a constraint condition.In this situation,leakage,pipe explosion and other accidents often occur in the pipeline system.Thus,there is a huge security risk in the pipeline system.This thesis is aimed at solving the flaws of traditional single objective optimization design and based on the existing multi-objective optimization model of water supply network.In the existing works,the reliability evaluation functions can not reflect the distributions of node surplus head in the network.To solve this problem,an evaluation index of reliability based on the sum of node surplus head and the variance of node surplus head is proposed in this thesis.Finally,a new multi-objective optimization model of water supply network is proposed,including annual converted expenditure of network,the sum of node surplus head and the variance of node surplus head.2.For water supply network,the optimized model is multivariable,nonlinear and discrete,also it has many constraint conditions and large solution space.Therefore,SPEA2 is selected as the optimization algorithm.But in SPEA2,the distribution of solutions is not good and the local searching ability is weak.To solve these problems in SPEA2,a modified algorithm of SPEA2 named RVSPEA2 is proposed.In RVSPEA2,reference vectors are introduced to improve the distribution of solutions.Simultaneously,niche selection strategy is proposed in order to improve the local search ability of the algorithm.The proposed RVSPEA2 is compared with other similar algorithms in ten international standard multi-objective test functions.The simulation results show that the proposed RVSPEA2 outperforms these algorithms in terms of distribution and convergence.3.In order to verify the effectiveness of multi-objective optimization model of the network and the ability of RVSPEA2 to solve the problem of network optimization,the optimization process of multi-objective model for water supply network using RVSPEA2 is expounded.The proposed RVSPEA2 is compared with other algorithms by optimizing two classic cases of network.Experimental results verify the effectiveness of the multi-objective optimization model of network and RVSPEA2.Finally,RVSPEA2 is applied in the practical network design,and three alternative design schemes are presented.4.For water supply network,the current management mode is backward,the degree of informationization and the emergency capacity are low.In order to solve these problems,an intelligent management system of water supply network is developed.Based on Visual Studio 2010,ArcGIS,Google SketchUp and MATLAB software,the modular design method is used in the system,which contains four modules,including user management,graphics management,information management and intelligent decision.In the user management module,the functions of user login and registration are designed,which ensure the safety of system and user information.In the graphics management module,graphic editing,graphic browsing,graphical output and other functions are included.Users can conveniently see the location relationship between adjacent pipe lines in three-dimensional view,and graphics are outputted through twodimensional view.In the information management module,information editing,information browsing and information output are mainly included.Users can query,browse and edit the network information,and the forms with the results queried or edited are generated to output.In the intelligent decision module,the optimization of network and early warning analysis are designed,which provide people with the basis and help for network construction and management.The system realizes the higheffective informationization management of urban water supply network,and improves the ability of emergency.