As a new swarm intelligence algorithms, particle swarm optimization algorithm, since its simple rules, rapid calculation and a strong global searching and optimization capabilities etc., recent years attracted more and more attention and research of the scholars in many related fields. Currently, the PSO algorithm has been successfully applied in the fuzzy system control, function optimization and other optimization algorithms fields.In this paper, the theory of particle swarm optimization will be systematically introduced, and the defects and problems of algorithm will be analyzed as well. On this basis, the opposition-based learning which has the faster learning speed and optimization capabilities is used to increase the particle swarm optimization ability and convergence speed. The main work of this paper are summarized as follows:(1) The background, model and the mathematical principles of the PSO algorithm are systematically introduced. In recent years, some typical improvements of the PSO algorithm are simply introduced. At last the defects and the cause of these problems of the PSO algorithm are analyzed.(2) The background, principles of the opposition-based learning and the basic theory of the Opposition-based Particle Swarm Optimization are presented.(3) The mathematical model of Opposition-based Particle Swarm Optimization is established and analyzed through the stochastic processes. The convergence of OPSO is reviewed through the martingale method. The transfer process of OPSO maximum fitness is described as a sub martingale. The almost everywhere convergence conditions of OPSO is established by the martingale convergence theorem.(4)The Plow Operator mechanism is applied into the particle swarm optimization. The blind detection of Plow Operator is used to improve the problem of the OPSO which easily falls into local optimum. Experiments show that:the algorithm has strong optimization ability and convergence speed.(5) The dynamic cross-factor OPSO was proposed. The cross-factor is employed to increase the particle diversity and reduce the search space, with the purpose of the improving the convergence rate of the algorithm in the evolution process. Lastly a linear decreasing probability is used to adjust the OPSO to have a faster closing speed.Finally, the main works of this paper are summarized, and further research questions are indicated. |