The Study Of Particle Swarm Optimization And Its Application To TSP

Particle Swarm Optimization (PSO) is a new kind of evolutionary computation and was originally introduced by Eberhart and Kennedy in 1995. It has since proven to be a powerful global optimization method. PSO has been widely applied in function optimization, neural network training, and fuzzy system control, etc. However, as PSO is a newly emerging optimization method, there are many research work should be substantiated. So it is very significant to seek more powerful improved algorithms based on PSO to solve concrete engineering problems.The paper analyses the elements of PSO. Based on the analysis on characteristics of PSO, the algorithm is summarized. The experienced settings of parameters are also given. In the paper, the present productions of PSO are summarized and compared.There is still no formal prove in theory on PSO because of the lack of theoretical research. The present research illustrates that the parameters setting plays an important role in PSO. The different improvements of PSO are all based on the improvements of setting of parameters. The paper studies the trajectories of particles in the search space and analyzes parameter settings lead to the swarm convergence.According to the investigation of PSO, the paper proposes a new fuzzy self-adapted particle swarm optimization. The attractive characteristic of the new PSO is that different values of the inertia weight were used in a same population. It speeds up the convergence of the swarm and thus improves the performance of PSO. It also keeps good diversity in particles and overcomes the evil of global inertia weight, which causes PSO trap into a local optimum. Four different benchmark functions were used to test the new PSO. The results illustrate that the new PSO has higher performance than the PSO with global inertia weight.Traveling Salesman Problem (TSP) is one of the classic problems of combinatorial optimization in graphic theory. TSP is a typical NP-hard problem and many practical problems can be transformed into TSP. The paper investigates the discrete PSO and designs corresponding operations in it for TSP. The new PSO was applied for TSP. The experiment results show that the new PSO perform much better than that with global inertia weight in resolving TSP.