A Hybrid Multi-Strategy Improved Dung Beetle Optimization Algorithm for Global Optimization Problems
Zheyi Wang, Aosheng Xing, Jie Zhang · 2024
Dung Beetle Optimizer (DBO) is a meta-heuristic algorithm formed by modeling inspired by living creatures and is known for its fast convergence. However, it still suffers from poor global exploration abilities, a proclivity to settle for local optimums and a lack of population variety. Therefore, this paper proposes a hybrid beluga whale optimization (BWO) and dung beetle optimization algorithm with somersault foraging (SRBDBO) using three strategies to improve the DBO. First, the interaction behavior between individuals in the hybrid BWO exploration phase enhances the global exploration ability of the rolling dung beetle, and the diversity of the dung beetle population is improved by mixing the BWO whale-fall phase at the end of each iteration of the population. Second, a somersault foraging strategy is introduced to expand the foraging area of small dung beetles. An adaptive somersault factor is also incorporated to balance global exploration and local exploitation and prevent the population from stopping updating at a locally optimal solution. Third, refraction opposition-based learning (ROBL) is used to reduce the likelihood that a thief dung beetle will stagnate updating in a locally optimal region. Finally, the widely recognized CEC2017 benchmark function set is utilized to assess the performance of SRBDBO. The simulation experiments show that SRBDBO demonstrates strong optimization-seeking performance and robustness when compared with the selected algorithm.