Empirical study of a hybrid algorithm based on Clonal Selection and Small Population Based PSO

摘要

In this paper, a hybrid algorithm, based on Clonal Selection Algorithm (CSA) and Small Population Based Particle Swarm Optimization (SPPSO) is introduced. The performance of this new algorithm (CS2P2SO) is observed for four well known benchmark functions. The SPPSO is a variant of conventional PSO (CPSO), introduced by the second author of this paper, where a very small number of initial particles are used and after a few iterations, the best particle is kept and the rest are replaced by the same number of regenerated particles. On the other hand, CSA belongs to the family of Artificial Immune System (AIS). It is an evolutionary algorithm, where, during evolution, the antibodies which can recognize the antigens proliferate by cloning. With the hybridization of these two algorithms, the strength of CPSO is enhanced to a great extent. The concept of SPPSO helps to find the optimum solution with less memory requirement and the concept of CSA increases the exploration capability and reduces the chances of convergence to local minima. The test results show that CS2P2SO performs better than CPSO and SPPSO for the Sphere, Rosenbrock’s, Rastrigin’s and Griewank’s functions.