A Target-Oriented Dynamic Foraging Artificial Hummingbird Algorithm

The Artificial Hummingbird Algorithm (AHA), inspired by the unique flight capabilities and intelligent foraging behavior of hummingbirds, has been widely applied to solve optimization problems in recent years. However, AHA has the problem of unbalanced exploration and exploitation capabilities during optimization, leading to premature convergence and suboptimal solution accuracy. Therefore, this paper proposes a Target-Oriented Dynamic Foraging Artificial Hummingbird Algorithm (TODF-AHA), which adopts a goal-attraction-based foraging strategy and dynamically updates visitation information to ensure the population maintains sufficient diversity and adaptability during iterations. Furthermore, by introducing an asymmetric foraging probability adjustment mechanism, which appropriately increases the guide foraging probability and reduces the domain foraging probability, TODF-AHA effectively balances the weight between guided foraging and local foraging, thus improving global and local exploitation capabilities. We evaluated TODF-AHA on 28 functions of CEC2013 and compared it with five state-of-the-art algorithms, namely SAHA, PSO, LCBA, TVAC and MEABC. TODF-AHA outperformed these five algorithms on 19, 25, 20, 21 and 15 functions, respectively. In the Friedman test, TODF-AHA had the lowest rank (2.34). The experimental results demonstrate that TODF-AHA exhibits superior convergence performance and solution accuracy when tackling complex optimization problems, demonstrating its competitive advantage.