A reinforcement learning-guided artificial bee colony algorithm with adaptive operator selection for numerical 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.2026.116229
- Dergi Adı: Applied Soft Computing
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Compendex, INSPEC
- Anahtar Kelimeler: Adaptive operator selection, Artificial bee colony, CEC 2017, Global optimization, Reinforcement learning, Swarm intelligence
- Erciyes Üniversitesi Adresli: Evet
Özet
Artificial Bee Colony (ABC) has established itself as a prominent swarm intelligence algorithm for global optimization. However, its conventional single-dimensional search equation frequently leads to premature convergence and stagnation when confronting highly complex, multimodal, and non-separable optimization landscapes. To address these limitations, this paper proposes a Reinforcement Learning-guided Artificial Bee Colony algorithm, named QLABC. The proposed method integrates a discrete 6-state Q-learning agent that continuously monitors population diversity and algorithmic stagnation to dynamically orchestrate the search process. Instead of relying on static probabilities, the intelligent agent adaptively switches between standard partial-dimensional exploration, best-guided partial-dimensional exploitation, and a highly disruptive Differential Evolution (DE)-based mutation strategy. Additionally, a phase-end delayed Q-table update mechanism is introduced to stabilize the reinforcement learning process against noisy environmental feedback. The proposed framework is extensively evaluated on the challenging CEC 2017 benchmark suite and validated on the real-world Tension/Compression Spring Design engineering problem. Experimental results and comprehensive ablation studies indicate that while QLABC performs competitively on simpler topographies, it provides improved robustness and reduces severe stagnation in non-separable hybrid environments. Non-parametric Friedman and Nemenyi statistical tests further support the competitiveness of QLABC as a self-adaptive framework for complex continuous optimization problems.