A New Variable Step-size and Updating Search Domains Glowworm Swarm Optimization Algorithm

Abstract

To address slow convergence of Glowworm Swarm Optimization (GSO) algorithm in the late stage of optimizing multimodal functions, a new variable step-size glowworm swarm optimization (VSGSO) algorithm is proposed, which improves convergence performance by adaptively adjusting step-size based on luciferin carried by each glowworm. Based on VSGSO, VSGSO-D algorithm, which is capable of updating search domains, is proposed. It dramatically improves optimization precision of GSO, especially in the case of optimization of complex multimodal functions. Optimization experiments are carried out by ten benchmark functions. The experimental results show that the improved algorithms proposed in this paper improve the global optimization speed, precision, and stability of GSO. At the same time, the enhanced algorithms protect the diversity of glowworm population and improve the multi-local optimization ability of GSO.