| Summary: | 碩士 === 國立高雄第一科技大學 === 系統與控制工程研究所 === 95 === The inspiration of Particle Swarm Optimization is from observing birds’ forage. In fact, all particles evolve in a swarm at the same time, so it use parallel processing manner to handle. However, most of the computers almost use sequent processing manner to work nowadays, thus it can not satisfy all particles’ evolution in parallel processing manner. The differences of two methods are worth study.
Use Mutli-processor or Hyper Cluster to achieve parallel processing method is too difficult for users. Because of the cost or technique is great consideration. For the reason, the thesis is used in thread to imitate the computing form of parallel processing manner. Then, discuss the differences between the way and sequent processing manner.
Use parallel processing for algorithm to compute because of a computer’s computing. When a particle swarm evolves, it disperses to many swarms. On the communication among the swarms, there are several articles using network topology to be the way of communication of the swarms. The thesis applies four common network topology forms to be the way of communication of the particle swarms. Through the communication of the swarms, they send and receive the information to each other.
The experiment is separated into three parts, searching single static target, variable targets and building the models of obstacle. Adjust parameters of the experiment and observe the evolving results on design simulator. Finally, analyze and discuss the results of experiment completely.
|
|---|
