An Improved Glowworm Swarm Optimization Based on Various Mutation Operators

Glowworm Swarm Optimization (GSO) is a population-based optimization algorithm that successfully solves numerous optimization problems. Nonetheless, the convergence speed required to reach optimal solutions can be made more efficient by skipping local optima. Also, considerable attention to tuning the parameter of the algorithm is crucial to improve the convergence speed. In this study, three variants of GSO are proposed using various mutation operators (Gaussian, Cauchy, and L & eacute;vy) to improve its convergence speed and prevent it from getting stuck in a local optimum. The small and random changes provided by the Gaussian mutation help in fine-tuning the position of the Glowworms. Meanwhile, the Cauchy mutation offers large changes that can assist the movement operator of the GSO in exploring wide area of the search space. Also, L & eacute;vy mutation is characterized by occasional large jumps, which have the potential to explore the problem space effectively. The performance and accuracy of the proposed methods are studied based on famous multimodal and unimodal benchmark test functions, as well as the CEC2014 test suite. Additionally, we have experimented with the proposed algorithm on a set of engineering problems. The effects of the parameter settings on the improved GSO are discussed using Response Surface Methodology (RSM). Results revealed that the suggested GSO algorithms, offer better solutions than the basic GSO algorithm and other GSO variants. In comparison to the state-of-the-art algorithms, the proposed GGSO obtained the best results for 68.75%, and 63.33% of the benchmark test functions and CEC2014, respectively. Additionally, statistical tests show the superiority of GGSO over other modified algorithms.