A Novel Distributed Gravitational Search Algorithm With Multi-Layered Information Interaction

Abstract

Population structures play a crucial role in individuals' evolution. Gravitational search algorithm (GSA) inspired by physical laws is a population-based algorithm. Its population structure is able to influence the individuals' search behavior. In this paper, we propose a distributed GSA with multi-layered information interaction, termed as MGSA, to offer a good balance between exploitation and exploration. A historical information layer and an elite top layer are designed to improve individuals' interaction. A distributed structure maintains the population diversity. Experimental results on CEC2017 benchmark functions and a real-world static economic dispatch problem confirm the effectiveness of MGSA in comparison with several state-of-the-art algorithms. In addition, landscape search trajectory and population diversity analyses verify the excellent convergence of MGSA.

Umbrellalike Hierarchical Artificial Bee Colony Algorithm

“…Nevertheless, the issues of how to further improve their performance are still very important and challenging. Many scientific works have shown that the population structure has a significant effect on improving the performance of an algorithm [23], [24]. In general, it uses different layers to place different individuals, and the information interaction between layers enables individuals to obtain better cooperation [25], thereby achieving more efficient evolution [26].…”