Zebra Optimization Algorithm Incorporating Multiple Improvement Strategies

Abstract

Given the limitations of the Zebra Optimization Algorithm in terms of both the ability to jump out the local optimum solution and convergence speed, this study developed a zebra optimization algorithm incorporating multiple improvement strategies (MI-ZOA). In order to enhance the global search capability and improve the uniformity of the population distribution within the search space, the algorithm initially introduces Kent chaotic mapping to produce random sequences for population initialization. Moreover, the algorithm capitalizes on the long-tailed attribute of the Lévy flight and puts in a factor that has a non - linear variation with the iteration number, with the aim of increasing the search space coverage while being in coordination with the algorithm’s local development capacity. Furthermore, the golden-sine update mechanism is introduced into the algorithm to improve search efficiency and optimization accuracy at a later stage. Subsequently, after the ZOA algorithm resists predator attacks, a Gaussian-Cauchy mutation operator is introduced to effectively avoid getting trapped in local optima and accelerate the algorithm’s convergence rate. Finally, using eight benchmark functions in the CEC2017 test set, comparative tests were conducted on MI-ZOA, ZOA, DBO, GA, and HHO. The results showed that the MI-ZOA had advantages in convergence speed and global search ability compared to other algorithms.