An innovative hybrid algorithm for bound-unconstrained optimization problems and applications

Abstract

Particle swarm optimization (PSO) and differential evolution (DE) are two efficient meta-heuristic algorithms, achieving excellent performance in a wide variety of optimization problems. Unfortunately, when both algorithms are used to solve complex problems then they inevitably suffer from stagnation, premature convergence and unbalanced exploration–exploitation. Hybridization of PSO and DE may provide a platform to resolve these issues. Therefore, this paper proposes an innovative hybrid algorithm (ihPSODE) which would be more effective than PSO and DE. It integrated with suggested novel PSO (nPSO) and DE (nDE). Where in nPSO a new, inertia weight and acceleration coefficient as well as position update equation are familiarized, to escape stagnation. And in nDE a new, mutation strategy and crossover rate is introduced, to avoid premature convergence. In order to balance between global and local search capability, after calculation of ihPSODE population best half member has been identified and discard rest members. Further, in current population nPSO is employed to maintain exploration and exploitation, then nDE is used to enhance convergence accuracy. The proposed ihPSODE and its integrating component nPSO and nDE have been tested over 23 basic, 30 IEEE CEC2014 and 30 IEEE CEC2017 unconstrained benchmark functions plus 3 real life optimization problems. The performance of proposed algorithms compared with traditional PSO and DE, their existed variants/hybrids as well as some of the other state-of-the-art algorithms. The results indicate the superiority of proposed algorithms. Finally, based on overall performance ihPSODE is recommended for bound-unconstrained optimization problems in this present study.