eBee: An elite-guided artificial bee colony algorithm with stochastic mask-driven exploration for numerical and engineering optimization
Applied Soft Computing, cilt.203, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 203
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.asoc.2026.116234
- Dergi Adı: Applied Soft Computing
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Compendex, INSPEC
- Anahtar Kelimeler: Elite-guided search, Swarm intelligence, The eBee optimization algorithm, UAV coverage path planning, Wind power plant micro-siting
- Erciyes Üniversitesi Adresli: Evet
Özet
Swarm intelligence algorithms have become essential tools for solving non-differentiable optimization problems. The canonical Artificial Bee Colony (ABC) algorithm, however, often suffers from slow convergence and limited exploitation, owing to its single-dimension perturbation mechanism and its purely random scout activation. To overcome these limitations, this study proposes the Elite-Guided Artificial Bee Colony Algorithm (eBee), a structurally enriched reformulation of the canonical ABC framework that integrates stochastic multi-dimensional perturbation, adaptive heavy-tailed scale generation, and an elite-guided search strategy into a single cohesive design. eBee employs a stochastic binary mask to perturb subsets of variables simultaneously and thereby capture inter-variable dependencies, while a heavy-tailed step-size generator, obtained by applying a random power transformation to random variates, provides a self-adaptive balance between local refinement and global exploration without additional parameter tuning. The onlooker phase is further strengthened by a power-law selection mechanism that biases the search towards elite solutions and accelerates convergence in promising regions. The proposed method was evaluated on the IEEE CEC 2005 and CEC 2022 benchmark suites and compared with ten state-of-the-art algorithms, namely ABC, CLPSO, EJAYA, IMODE-TMS, JADE, jDE, PSO, SADE, TSA, and UMODEAsIV. Its practical value was then demonstrated on two structurally distinct engineering problems: an energy-aware UAV coverage planning task and a levelized-cost micro-siting problem for a wind power plant, the latter solved by combining the eBee search operators with Deb's parameterless feasibility rules. The experimental results show that eBee is statistically significantly superior on most complex multimodal and hybrid landscapes while remaining computationally competitive. Statistical validation based on the Wilcoxon signed-rank test with multiple-comparison correction confirms the robustness and consistency of the observed improvements. These findings establish eBee as an efficient and precise optimization framework for challenging numerical and engineering problems.