A Modified Particle Swarm Optimization for Solving Global Optimization Problems

This paper proposes a modified particle swarm optimization based on the combining attractive and repulsive operator with function stretching technique (for short MPSOwARS). This new algorithm utilizes adequately the characters that the attractive and repulsive operator can efficiently ensure diversity of swarm and make algorithm prevent premature convergence, and the characters that function stretching technique can decrease efficiently the complexity of objective function. The results of the experiment on benchmark function are presented. Conclusions show that the MPSOwARS algorithm performs better than previous works and adapts to solve very complex multidimensional and multi-modal global optimization problems especially