Research On Hybrid Particle Swarm Optimization With Genetic Algorithm

Application of particle swarm optimization algorithm and genetic algorithm in the field has a lot of common ground, such as pattern recognition, machine control, electronic communications, and so on. Particle swarm algorithm is simple and easy to use. Its search is with high efficiency and a good local search capability, but a more general global search capability, so it is easy to fall into a local optimum. Particle swarm optimization algorithm is suitable to the processing of real value data, not to be good at the processing of discrete data. Genetic algorithms compared to the particle swarm optimization algorithm is more complex, as compared with many other intelligent algorithms, genetic algorithm’s local search capabilities is more general, the time of converging to the optimal solution is relatively long, but its capability of global search is very good. This effective combination of particle swarm optimization algorithm and genetic algorithm mixed algorithm is designed to keep the original advantages of each algorithm, efficient algorithm to avoid the existing deficiencies. Via the solution of function optimization problem and the traveling salesman tour problem to verify the validity of this algorithm. Major work includes:(1)To hold the advantages of the original algorithm in hybrid algorithm, and avoid the deficiency of the original algorithm, this strategy is to allow genetic algorithm and Particle Swarm Optimization algorithm to manipulate different objects, coupled with improved niche merit retention policies. Adopting self-adaptive dynamic inertia weight and asynchronous change factor in particle swarm algorithm aims at improving basic particle swarm optimization algorithm’s deficiencies. Modifying the crossover and mutation operators in genetic algorithm, to make its global search and local search ability can automatically adjust intensity and precision. Through several classic functions test to verify whether the hybrid algorithm is better than a single particle swarm optimization algorithm or genetic algorithm, via comparison with other algorithms to prove the effectiveness of the algorithm.(2) To use the mixed algorithm of this paper to solve the travelling commercial problem, needs some modification on this side of particle swarm optimization algorithm, this article mainly is through Ⅱ Yuan cross operation to solve this difficult problem, Ⅱ Yuan cross operation in operation means is similar to the cross operator of genetic algorithm, and at the same time it has some operation features of learning factor of particle swarm optimization algorithm. In order to further improve the convergence speed, introduce the factor about the far and near relation of city’s distance. A test of 3 examples proves the heuristic factor is able to improve the convergence of the algorithm in the end of this paper, especially for traveling salesman problem that has large number of cities, the difference becomes apparent.