An improved sparrow search algorithm

Abstract

The sparrow search algorithm (SSA) is a relatively new swarm intelligence heuristic algorithm. It has fast convergence speed, strong optimization ability and more extensive application scenarios compared with traditional heuristic search methods. And thus, the SSA is attracting the attention of researchers in different fields. However, there are deficiencies of initial population quality, search ability, and population diversity in the SSA. Therefore, this paper proposes an improved sparrow search algorithm (ISSA). The ISSA uses skew tent map-based chaotic method to produce initial population for a higher quality of convergence. For the location update of the producer sparrows during the iterations, the ISSA introduces a non-linear decreasing weight, promoting both exploration and exploitation of the search space, to improve the convergence and search precision. And the mutation strategy is employed to update the location of the scrounger sparrows with lower energy and the chaotic search is combined with the local exploitation for the scroungers with higher energy, which can enhance the diversity and avoid trapping in local optimum. Simulation experiments are carried out on 26 benchmark test functions. And the results show that the ISSA is superior to or at least competitive to the SSA in the convergence properties of accuracy, speed, and stability.