Hybrid Particle Swarm Algorithm for Packing of Unequal Circles in A Larger Containing Circle

This paper describes an algorithm based on particle swarm optimization (PSO) for the unequal circles packing problem. Three strategies are used to enlarge the searching space of PSO. First, the positions of particles are periodically reinitialized to help the algorithm in escaping from local minimums, where the circles with more overlap gain more chance to be reinitialized. Second, a policy similar to simulated annealing approach is adopted to decide whether or not to accept the reinitializing results. Finally, the affection of the global best particle is linearly increased to delay the convergence. Four packing experiments are performed to test the hybrid PSO. The results show that the algorithm has better performance when the scale of the problem is not very large