Research On Topology Of The Particle Swarm Optimization Algorithm

The adaptive capacity of organisms in nature is constantly evolving solving the complex optimization problem. In recent years, some bionic optimization algorithm such as ant colony algorithms, artificial immune algorithm, particle swarm algorithm appears, it not only enrichs the modern optimization techniques, but also provided a new solution for traditional technologies to solve the optimization problem. with the majority of practical problems complexity growing, in the field of industrial and scientific, it has been a large number problem can not be find the optimal solution of the problem in an acceptable period. Traditional planning techniques have been unable to meet the demand for solving such problems, so it needs a more efficient optimization algorithm. As a new group of intelligent methods, particle swarm optimization is a very useful tool in dealing with high dimensional problems and the lack of domain knowledge problems. The main idea of particle swarm optimization algorithm derived from natural biological groups of foraging activity, the algorithm is principle simple, convergence speed, high precision, since the introduction it has been widespread concern in the field of optimization.In this paper, topology particle swarm algorithm analyzed, topology is the connection between different particles, the connections ways of the particle swarm algorithm will affect convergence accuracy and convergence time. Therefore, using reasonable topology will get better convergence results for different test function . This paper studied on topology structure based on the basic particle swarm algorithm , and the same test functions validate the different topologies affect the particle swarm algorithm. There are two method for the improved particle swarm algorithm topology validation, one is the number of particles is adjustment so that particle swarm optimization has more choice in the search early, even if the particles have a stronger diversity, while in the latter part of the algorithm to reduce the number of particles, so that convergence time of the algorithm is shorter, Another strategy of the particle swarm algorithm for a fixed initial assignment of particles, so that particle swarm algorithm in the search space, a better particle distribution, and ultimately improve the convergence accuracy of particle swarm optimization algorithm. Finally, simulation test functions to verify all topologies in testing the effectiveness of different functions.