No abstract

Evolutionary computation algorithms (EC) and swarm intelligence have been widely used to solve global optimization problems. The optimal solution for an optimization problem is called by different terms in EC and swarm intelligence. It is called individual in EC and particle in swarm intelligence. Self-adaptive differential evolution (SaDE) is one of the promising variants of EC for solving global optimization problems. Adapting self-manipulating parameter values into SaDE can overcome the burden of choosing suitable parameter values to create the next best generation’s individuals to achieve optimal convergence. In this paper, a fully informed particle swarm (FIPS) is hybridized with SaDE to enhance SaDE’s exploitation capability while maintaining its exploration power so that it is not trapped in stagnation. The proposed hybrid is called FIPSaDE. FIPS, a variant of particle swarm optimization (PSO), aims to help solutions jump out of stagnation by gathering knowledge about its neighborhood’s solutions. Each solution in the FIPS swarm is influenced by a group of solutions in its neighborhood, rather than by the best position it has visited. Indirectly, FIPS increases the diversity of the swarm. The proposed algorithm is tested on benchmark test functions from “CEC 2005 Special Session on Real-Parameter Optimization” with various properties. Experimental results show that the FIPSaDE is more effective and reasonably competent than its standalone variants, FIPS and SaDE, in solving the test functions, considering the solutions’ quality.