Improvement And Application Of Particle Swarm Optimization Algorithm

Optimization problem is widespread in production and life of human society. With the development of science and technology, and the improvement of complexity of related issues, higher request to the optimization techniques is put forward. Particle Swarm Optimization(PSO) algorithm is a new kind of optimization tool,which originates from the research of foraging behavior of birds,and it belongs to the category of Swarm Intelligence Algorithm. Because of its simple principle, easy implementation, stronger generality, the algorithm has been researched widely by lots of experts and applied well in many areas.Particle swarm optimization algorithm is an effective optimization technique, but both the theory basis and practical applications are not mature enough, there are still many issues worthy of further study. This paper analyzes the basic principle of PSO algorithm, and makes a research mainly on the improvement and application of PSO algorithm on the base of two sub-swarms particle swarm optimization algorithm and quantum-behaved particle swarm optimization algorithm. The main works can be showed as follows:(1) Based on two sub-swarms particle swarm optimization algorithm, a Two Sub-swarms Particle Swarm Optimization algorithm Based on Multi-phase Exchange (TSME-PSO) is proposed. In TSME-PSO, the particles of two sub-swarms update their position and velocity with different models. In the whole searching phase, the particles will be exchanged between two sub-swarms, and the exchange amount of particles continues to fall. The experiment results of some functions show that the overall performance of TSME-PSO algorithm is much better than other algorithms,especially in high-dimension and complicated problems.(2) By studying quantum-behaved particle swarm optimization algorithm and combining with the exchange strategy of TSME-PSO, a Two Sub-swarms Quantum-behaved Particle Swarm Optimization algorithm based on Exchange Strategy (TS-QPSO) is proposed. In TS-QPSO, the particle has the feature of quantum behavior and it can appear everywhere of the searching space in a certain probability. For this reason,the space can be expanded greatly. By setting the exchange amount and exchange mode reasonably, the population diversity can be enhanced effectively, and the performance of global optimization can be also improved. Simulation results have proved the feasibility of TS-QPSO.(3) The improved algorithms: TSME-PSO and TS-QPSO proposed in this paper are applied to solve Vehicle Routing Problem with Time Windows (VRPTW).The specific coding method of VRPTW is given in this paper, and the detailed process of solving VRPTW problems is also described. Experiment results show that the convergence performance of improved PSO algorithms is superior to PSO algorithm, and the effect of optimization is satisfactory.