Cooperative multi-swarm particle swarm optimization based on adaptive and time-varying inertia weights

Optimization of particle swarms is a stochastic optimization method based on swarm intelligence applied in many fields of endeavor to solve technical, scientific and economic problems. Due to its ease of application, it has gained great importance in recent years. As the swarm may lose its diversity and lead to premature convergence, it is very easily trapped in local optima. To solve this problem, we propose, in this research work, an cooperative multi-swarm particle swarm optimization algorithm called cooperative multi-swarm particle swarm optimization (C-MsPSO). The introduced algorithm divides the entire population into four cooperative sub-swarms with an adaptive and time-varying inertia weight. The particles of each sub-swarm share the best overall optimum to ensure the cooperation between the four sub-swarms . On the other hand, the adaptive and time-varying inertia weight is used to create search potential and effectively maintain a balance between the local research (exploitation) and the global (exploration). To show the efficiency of the developed C-MsPSO algorithm, several uni-modal and multi-modal benchmark test functions are considered. The introduced algorithm demonstrates surprising efficiency and precision in identifying the optimal solution.The experimental results reveal that C-MsPSO outperforms the other PSO algorithms on twelve reference functions.