通过分析模式定理及建筑块理论,提出一种基于建筑块迁移策略并行遗传算法。算法根据种群的收敛情况,从其他种群中获取非重叠的建筑块,采用模拟退火思想防止优良模式的浓度过快地增大引起早熟。理论分析和对多峰函数的仿真结果均表明,该算法减少了无效迁移次数,降低了通信开销,而且发生成熟前收敛的概率明显下降,保证了遗传算法的全局收敛性。
Through analyzing schema and building blocks thesis, this paper proposed a parallel genetic algorithms based on building block migration. Relying on convergence situation, got unlapped building block from other population, and used simulated annealing method to prevent the density of good schema increase greatly resulting in premature convergence. Theory analysis and experimental results show that the method not only reduces ineffective migration and decrease communication costs, but also lowers the possibility of occurring premature and assures the capability of global convergence.