Multi-objective Optimal Design Of Water Supply Systems Via Improved NSGA2 Algorithm

City water supply pipe network is an important part of urban water supply system which also is one of the important public infrastructure of city construction. With the continuous development of social economy, the urbanization process is accelerating, the existing water supply network facilities have been unable to meet the growing demand for both quantity and pressure. Therefore, urban multi-objective optimization design of water distribution network has become urgent problems of urban infrastructure construction. Due to the cost of the laying of pipeline system accounts for sixty to eighty percent of the whole project investment. In Early construction, reliability factor is easily overlooked. Urban water supply network multi-objective optimization design not only can greatly reduce the project investment, but also improve the reliability of the pipeline system. This paper analyzes the hydraulic characteristic of pipe network, proposed the network reliability and verified the mathematical deduction. Finally, a kind of new type of multi-objective optimization model of pipe networks based on reliability and economy is established. The mathematical model of pipe network is discrete, nonlinear, large solution space dimension which is very difficult to solve. In order to improve the quality and precision of solutions, an improved non dominated sorting genetic algorithm(NSGA2) is designed, multi-objective optimization design of water supply pipe network is taken out by using the improved algorithm to verify the effectiveness of the proposed algorithm. Finally, in order to improve the degree of information management of pipe network, a water supply pipe network management system is developed using VS2010, combining with ArcGIS and MATLAB development tools which provides the basis for the operation of the pipeline in the late maintenance. The main research work in this paper is as follows:1. Traditional water distribution network optimization design only consider the economic indicators, the objective function is simple, a single optimization results cannot provide a reasonable construction plan for the builder. We designed a new type of multi-objective optimization model which based on economy and reliability of the water supply network. By analysis the hydraulic of pipe network, we propose a reliability evaluation index by calculating the variance of head node. According to the requirements and constraints of the actual operation of the pipe network hydraulic conditions, we use Monte Carlo method to analyze network reliability model calibration, the final verification of multi-objective model mentioned in accuracy can reach the practical requirements. It applies to multi-objective optimization design of water distribution network.2. For the water supply network in the presence of multi-objective solution has many variables and a large amount of pre-processing, so, the solving process is complex and computing slowly. An improved non dominated sorting genetic algorithm(NSGA2) is designed. The improved algorithm introduced differential mutation operator to enhance the ability of local search. By optimizing the function of four benchmark test problems, results show that the improved algorithm can avoid the local optimal. All these improvements can enhance the ability of local search, but the search accuracy of improved algorithm is not high. Therefore, in order to take into account the speed and accuracy of algorithm, a variable scaling factor strategy is proposed, by dynamically adjust the algorithms of the search speed and accuracy of post-scaling factor and make a good balance between the convergence speed and reliability. By optimizing the two classic pipe case: bicycled pipeline network and Tower of Hanoi, experimental results show that the improved algorithm can jump over local optimum, obtain better solution with smaller computational cost. The improved algorithm can effectively solve the multi-objective optimization of water supply network problems.3. The existing traditional pipe network management model unable to achieve information management, the management efficiency is low. We developed a water distribution network management system. This software is under VS2010 platform, combine with ArcGIS secondary development based on C / S structure of the single plane version system, it’s easy to install. Programming with C# and MATLAB, multiobjective optimization module is embedded in the system, the intelligent optimization module design is completed, so the optimal results are intuitive display to users. By using ArcGIS control, two-dimensional map image of the pipe network is show out and customers can query channel pipe network and node information at any time.