Gravitational search algorithm is an effective population-based algorithm. It simulates the law of gravity to implement the interaction among particles. Although it can effectively optimize many problems, it generally suffers from premature convergence and low search capability. To address these limitations, a gravitational search algorithm with hierarchy and distributed framework is proposed. A distributed framework randomly groups several subpopulations and a three-layered hierarchy manages them. Communication among subpopulations finally enhances the search performance. Experiments discuss parameters and strategies of the proposed algorithm. Comparison between it and sixteen state-of-the-art algorithms demonstrates its superior performance. It also shows the practicality for two real-world optimization problems.

引力搜索算法是一种有效的基于种群的算法。它模拟重力定律以实现粒子之间的相互作用。尽管它可以有效地优化许多问题，但通常会出现过早收敛和搜索能力低下的问题。为了解决这些局限性，提出了一种具有层次结构和分布式框架的引力搜索算法。分布式框架将几个子种群随机分组，并由三层层次结构对其进行管理。子群体之间的交流最终提高了搜索性能。实验讨论了该算法的参数和策略。它与16种最新算法的比较证明了其卓越的性能。它还显示了两个实际优化问题的实用性。