A New Boundary Condition for Particle Swarm Optimization

Boundary conditions are often used in particle swarm optimization (PSO) in order to enhance the entire solution space of particles as far as possible. However, most of them are not categorized in detail, the boundary conditions used for comparisons are not in the same category and their performances vary in different engineering fields. In order to address these issues, this paper presents a comprehensive learning velocity boundary condition (CLBC), which is verified by Rastrigrin and Rosenbrock function with 30 dimensionalities in three types of search range. The CLBC shows comparatively faster convergence ability and obtains more precise results, compared with other boundary conditions of the same kind.