A New Crossed Particle Swarm Optimization Algorithm And Its Applications

Based on the swarm intelligence, Particle Swarm Optimization (PSO) algorithm is a kind of modern optimization method inspired by the research of the artificial life. As a kind of stochastic global optimization method, compared with the traditional optimization method, PSO does not need the higher analytic nature of the objective function, and is also used to the solution of some complex, large-scale, nonlinear optimization problem which may be nondifferentiable. Therefore, PSO attracts academic researchers'attention in recent years.The standard PSO algorithm and a number of modified PSO algorithms are discussed in this thesis. Thanks to the advantages of the standard PSO, we modify the PSO algorithm in some aspects. For example, the dynamic parameter is introduced to update the position equation, and the particles are limited in the search region. A new strategy for updating the speed is adopted, in which the speed is weakened linearly, and the number of updating speed is reduced. Then a new crossed PSO algorithm is proposed. The new algorithm intersects two different PSO algorithms, and the performance of the search process is boosted. Furthermore, the stretching technique is used in the neighbor of the local minimum to find the global minimum and another new crossed-stretching PSO algorithm is obtained. The new algorithm is used to solve some unconstraint nonlinear optimization problems with high dimensions, and the simulated results show that the performance of the new algorithm is very competitive with other similar PSO algorithms.In the end, we use the new algorithm to design the hexagon antenna array and to train the support vector machine in pattern recognition area, and obtain some favorable results, which also show that the new algorithm is effective.