Research On Convergence Analysis Of PSO And Its Application To Project Optimization

With the development of project management science and technology,optimization technology, the management of construction project and optimizationtechnology are becoming more and more integrated. However, the application of thecurrent optimization technologies to project management requires so manybackgrounds and resources including time that the practical needs can not besatisfactorily met.Particle swarm optimization (PSO), a kind of evolutionary computation foundedby Dr. Eberhart and Dr. kennedy, was motivated by the simulation of social behavior.Unlike other artificial intelligence technology, Particle swarm optimization (PSO) isapt to be carried out and is easy to reach the best resolutions. And hence, hardly hadparticle swarm optimization been brought forward when the algorithm wasextraordinarily focused on by scientists at home and abroad. Not only was thealgorithm applied to scientific study but was devoted to project optimization.After carrying on a detailed review and introducing the basic relevant theory, thisdissertation proposes a theoretic convergence analysis of particle swarm optimization,has a view on parameter selection, and in the end attempts to apply the algorithm tobetter solve project optimization problem. First, this dissertation introduces theinvention, development and improvement of particle swarm optimization, and thecorresponding version of algorithm, i.e. the original version, the inertia weight version,the constriction factor version and other improvements based on differentcomprehension of different scientists. However, to this day, many correlativeparameters are decided not by science and knowledge but by experience. As we cansee, different sets of parameters can result in different results, that is, the algorithmperhaps be easy to be convergent or maybe be difficult to reach the best point, or evencan never reach the destination. In the meanwhile, parameter selection can have aneffect on the balance of the exploration and exploitation ability of the algorithm.Therefore this dissertation is devoted to the convergence analysis of particle swarmoptimization. Based on the current study of the convergence analysis of particleswarm optimization in one-domain-problem, this dissertation innovates in discussingthe convergence analysis of particle swarm optimization in two-domain-problem andalso in multiple-domain-problem, and then supplies the theoretical ground ofparameter selection. After these discussiones, this dissertation introduces PSO toproject optimization. The example quoted indicates that particle swarm optimizationcan deal with this kind of problem easily and exactly.The convergence analysis of PSO and the application of PSO to the projectoptimization problem proposed in this dissertation have some theoretic and practicalsignificance.