基于TriBA结构和改进迁徙策略的并行遗传算法

为了提高并行遗传算法解决大规模问题的寻优效率,本文基于粗粒度并行遗传算法模型,提出了一种基于TriBA结构和改进迁徙策略的并行遗传算法,给出了TriBA并行遗传算法的分配策略和迁徙方式,并将改进的迁徙策略引入评价算子,通过评价算子,反映当前种群的局部收敛程度,当收敛程度满足设定的评价算子时,进行迁徙操作,并使用并行计算机对基于TriBA拓扑结构和改进迁徙策略的并行遗传算法进行模拟仿真.仿真结果表明,改进后的迁徙策略的并行遗传算法的寻优效率比传统迁徙策略的并行算法效率更高.该算法有效地提高了算法的寻优速度,增强了算法的寻优稳定性.

TriBA-Topological Parallel Genetic Algorithm Based on a New Migration Strategy

In order to improve the parallel genetic algorithm＇s efficiency to solve large-scale optimization problems, the paper proposes a TriBA（triplet based architecture）-topological parallel genetic algorithm based on a new migration strategy. TriBA structure is a scalable, fully connected topology, which can pro- vide a variety of migration patterns; the improved migration strategy introduces evaluation operator, through the evaluation operator, reflected the degree of convergence of the current local population. If the degree meets the set evaluation degree, then the algorithm does the migration. Then the simulation is done for TriBA-Topological parallel genetic algorithm based on a new migration strategy with parallel comput- er. By the algorithm test function test, the experimental results show that the traditional migration paral- lel gentic algorithm and the algorithm can effectively improve the searching speed of the algorithm, and en- hance the stability of the optimization algorithm.