Hierarchical multi-swarm cooperative teaching-learning-based optimization for global optimization

Hierarchical cooperation mechanism, which is inspired by the features of specialization and cooperation in the social organizations, has been successfully used to increase the diversity of the population and avoid premature convergence for solving complex optimization problems. In this paper, a new two-level hierarchical multi-swarm cooperative TLBO variant called HMCTLBO is presented to solve global optimization problems. In the proposed HMCTLBO algorithm, all learners are randomly divided into several sub-swarms with equal amounts of learners at the bottom level of the hierarchy. The learners of each swarm evolve only in their corresponding swarm in parallel independently to maintain the diversity and improve the exploration capability of the population. Moreover, all the best learners from each swarm compose the new swarm at the top level of the hierarchy, and each learner of the swarm evolves according to Gaussian sampling learning. Furthermore, a randomized regrouping strategy is performed, and a subspace searching strategy based on Latin hypercube sampling is introduced to maintain the diversity of the population. To verify the performance of the proposed approaches, 48 benchmark test functions are evaluated. Conducted experiments indicate that the proposed HMCTLBO algorithm is competitive to some existing TLBO variants and other optimization algorithms.