Formalized model and analysis of mixed swarm based cooperative particle swarm optimization

Natural phenomenon of mixed flocks indicates such principles as cooperation and social symbiosis among various species. Inspired by the organization and collective intelligence of natural mixed flocks, a mixed swarm based particle swarm optimization (MCPSO) is proposed to efficiently handle the trade-off between the global and local search in PSO. The approach divides all particles into two species, i.e., exploration species and exploitation species. Exploration species undertakes the coarse search in the solution space to discover new potential area, while the exploitation species is instructed accordingly to conduct fine search in its activity territory. Information sharing plays a crucial role between the two species, through the cooperative mechanism, not only does MCPSO avoid the optimum missed in a coarse search, but also it significantly saves void fine search. The proposed MCPSO is validated with well-known benchmarks confirming that the cooperative mixed swarm is an effective model for the swarm based searching, further proving that MCPSO is a robust global technique for complex optimization problems.