Research And Application Of Multi Objective Indirect And Direct Methods In Optimization Of Water Supply Network

In recent years, with the development of urbanization, the city size and population are increasing, and the demand for water resources is more and more. The water supply network system is an important infrastructure for the survival and sustainable development of the city, and the cost of the water supply network accounts for about 70%-85% of the total cost of the system. The safety and reliability of the water supply network is an important guarantee for the water consumption of urban residents. The water supply network has the characteristics of high investment cost and great social significance, so it is of great significance to study the optimal design of water supply network. The main research contents and achievements are as follows:For a single objective design of the traditional water supply network optimization, only with the economic costs as the objective and ignoring the importance and reliability of construction scheme, this thesis proposed multi-objective optimization mathematical model of water supply network based on the total cost of annual minimum and maximum reliability. The network construction cost include the laying of a construction cost,the annual depreciation overhaul cost and annual operating cost of power station, and consider the impact of interest rate fee; the comprehensive value of water supply pipe network reliability by node surplus water head and pipe network resilience to measure.According to the design optimization problem for complex non-deterministic polynomial problem, it can be divided into indirect method and direct method. The indirect method uses linear weighting method to transform it into a single objective mathematical model and then solves it, otherwise the direct method is directly used to solve multiple objectives. But the indirect method using weighting method is not reasonable to measure the competitive relationship of multiple objectives, so this thesis uses maximum and minimum method to value economy and reliability of the unified [0,1] range,and the objective function of indirect method is improved reasonably. In order to realize the indirect method, this thesis designs a new improved genetic algorithm, genetic algorithm, which can enhance the local search ability of the algorithm and avoid the local optimum. For direct method for network optimization, this thesis uses the classic NSGA-? algorithm to realize the algorithm and elitist non dominated sorting method based on the optimal Pareto can get uniform distribution of Pareto solution set in the forefront of the solution.In this thesis,two kinds of calculation methods of pipe network optimization are developed by MATLAB, tested by the two loop network model and Hanoi network model and applied to practical engineering network model. The results show that the two algorithms converge to the Pareto optimal solution in the small network, and for the large scale network, breeding genetic algorithm is better than NSGA-II algorithm in convergence effect. In addition, the optimal solution of genetic algorithm is less, the distribution is poor, and the NSGA-II algorithm has better distribution, so the NSGA-II algorithm is better than the genetic algorithm in the distribution of the solution.