An immune inspired multi-agent system for dynamic multi-objective optimization

In this research, an immune inspired multi-agent system (IMAS) is proposed to solve optimization problems in dynamic and multi-objective environments. The proposed IMAS uses artificial immune system metaphors to shape the local behaviors of agents to detect environmental changes, generate Pareto optimal solutions, and react to the dynamics of the problem environment. Apart from that, agents enhance their adaptive capacity in dealing with environmental changes to find the global optimum, with a hierarchical structure without any central control. This study used a combination of diversity-, multi-population- and memory-based approaches to perform better in multi-objective environments with severe and frequent changes. The proposed IMAS is compared with six state-of-the-art algorithms on various benchmark problems. The results indicate its superiority in many of the experiments.