An Improved Reptile Search Algorithm with Ghost Opposition-Based Learning for Global Optimization Problems
Heming Jia,Chenghao Lu,Di Wu,Changsheng Wen,Honghua Rao,Laith Abualigah
DOI: https://doi.org/10.1093/jcde/qwad048
2023-06-17
Journal of Computational Design and Engineering
Abstract:In 2021, a meta-heuristic algorithm, Reptile Search Algorithm (RSA), was proposed. RSA mainly simulates the cooperative predatory behavior of crocodiles. Although RSA has a fast convergence speed, due to the influence of the crocodile predation mechanism, if the algorithm falls into the local optimum in the early stage, RSA will probably be unable to jump out of the local optimum, resulting in a poor comprehensive performance. Because of the shortcomings of RSA, introducing the local escape operator can effectively improve crocodiles' ability to explore space and generate new crocodiles to replace poor crocodiles. Benefiting from adding a restart strategy, when the optimal solution of RSA is no longer updated, the algorithm's ability to jump out of the local optimum is effectively improved by randomly initializing the crocodile. Then joining Ghost opposition-based learning to balance the IRSA's exploitation and exploration, the Improved Reptile Search Algorithm with Ghost Opposition-based Learning for the Global Optimization Problem (IRSA) is proposed. To verify the performance of IRSA, we used nine famous optimization algorithms to compare with IRSA in twenty-three standard benchmark functions and CEC2020 test functions. The experiments show that IRSA has good optimization performance and robustness, and can effectively solve six classical engineering problems, thus proving its effectiveness in solving practical problems.
computer science, interdisciplinary applications,engineering, multidisciplinary