Study Of Particle Swarm Optimization And Its Applications In Urban Water Supply Scheduling

As an important intelligent algorithm, PSO algorithm has many merits, for example, easy to be realized, fewer parameters which needs to be adjusted, fast convergence speed, strong currency and so on, so it always is paid more attention to by many scholars. Firstly, this thesis analyses the limitations of the standard algorithm, which based on the basic principle and the current research of PSO algorithm, and low-dimensional urban water supply scheduling. Secondly, two improved particle swarm algorithm is proposed.Finally, an improved particle swarm ADNPSO is used to solve water regulation problem of water supply of the urban water supply scheduling. The main contributions of this papar include the following aspects:(1) To further improve the convergence precision and convergence speed of the PSO algorithm which solves the low-dimensional optimization functions, the thesis proposes an improved particle swarm algorithm based on adaptive niche, multi-population and hybrid technology (ADNPSO for short). The algorithm determines the best location of the contemporary population according to adaptive niche technology, increases the diversity of population by multi-population, and uses hybrid to improve the characteristics of each particle in the population. Thereby, the global search ability of PSO is enhanced. The experimental results show that as far as the low-dimensional optimization function is concerned, convergence precision and convergence speed of the ADNPSO algorithm is better than the LinWPSO algorithm.(2) Because it is difficult for high-dimensional function to optimize, this thesis for a type of high-dimensional function, which each component of the global optimal solution vector is the same, gives ADNPSO1 algorithm based on ADNPSO algorithms. The experimental results show that ADNPSO1 algorithm can well optimize the type of high-dimensional functions. In addition, ADNPSO1 algorithm is also suitable for low-dimensional multi-peak function which each component of the global optimal solution vector is the same.(3) Urban water supply scheduling, which is a low-dimensional dynamic nonlinear optimization problem, was solved by ADNPSO algorithm. Experiments results show that the algorithm can better solves urban water supply scheduling than Genetic algorithm.