Orthogonal learning particle swarm optimization with variable relocation for dynamic optimization.