A Multi-Role Particle Swarm Algorithm for Numerical Optimization

the paper proposes a stochastic differential search operator (SDSO) which can traverse the numerical object space (eg. R ''), and a Multi-Role particle swarm which can be differentiated into the different roles to search the objective space using the different strategies. Based on the SDSO and the Multi-Role particle swarm, a Multi-Role particle swarm algorithm (MRPSA) for numerical optimization is proposed. In the test experiment, 6 unconstrained benchmark functions are used to demonstrate the performance of MRPSA. The results show that MRPSA can find the optimal or close-to-optimal solutions of those benchmark functions efficiently.