An enhanced slime mould algorithm based on adaptive grouping technique for global optimization

Abstract

When solving global optimization problems by metaheuristic algorithms (MAs), an important issue is how to keep a balance between convergence and diversity. This article develops an enhanced slime mould algorithm called AGSMA to address the above issue. In AGSMA, the entire population is divided into two subpopulations. The size of each subpopulation is dynamically tuned based on the average fitness value. Afterward, different from the search equation in conventional SMA, a new search mechanism is raised to discover more promising regions. Furthermore, an efficient learning operator is devised to discourage premature convergence. The comparison of the raised AGSMA on several benchmark functions and practical engineering problems with selected state-of-the-art methods demonstrates that the developed algorithm performs effectively and competitively. The source code of AGSMA is publicly available at https://github.com/denglingyun123/Enhanced-slime-mould-algorithm.