A multi-population method for optimization algorithms inspired by platelet plug formation mechanism

Abstract

In primary hemostasis, platelets around the injury are activated, and an aggregation of them forms a platelet plug. The substances released by activated platelets control the activation, making it rapid and just enough. In this research, we propose a multi-population method for optimization algorithms inspired by the aggregation and the activation control of platelets. Each of individuals has a state of activated or inactivated, like platelets. Inactivated individuals explore globally whereas activated ones do locally. The states are changed dynamically to improve the efficiency of exploration. The conventional algorithms, such as particle swarm optimization (PSO) and differential evolution (DE), are used for each explorations of activated or inactivated individuals. We conducted experiments for the proposed algorithms using PSO or DE and compared with the standard PSO and DE. The results showed that some of the proposed algorithms converge to the global optimum more rapidly than the standard ones.