An adaptive oppositional grey wolf optimizer for complex engineering problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• Metaheuristic optimization algorithms face challenges like premature convergence and imbalanced search phases.
• The Grey Wolf Optimizer (GWO) shows potential but is limited by rigid parameters and elite stagnation in complex landscapes.

Purpose of the Study:

• To introduce the Adaptive Oppositional Grey Wolf Optimizer (AOGWO) to overcome limitations of existing metaheuristic algorithms.
• To enhance scalability and performance in complex, high-dimensional, and deceptive optimization problems.

Main Methods:

• Hybrid opposition-based learning (OBL) for maximum initial spatial diversity.
• Adaptive cosine control strategy with decaying Jumping Rate and Lévy flight for dynamic exploration-exploitation balance.
• Selective Leading Opposition (SLO) mechanism applied to the Alpha leader to prevent elite traps.

Main Results:

• AOGWO demonstrated superior performance on 41 benchmark functions (CEC2017 & CEC2022 suites), achieving top mean performance on 24/29 CEC2017 and all 12 CEC2022 functions.
• Statistically significant improvements over competing algorithms confirmed by Wilcoxon signed-rank and Friedman tests.
• Competitive optimal solutions for constrained engineering design problems, satisfying all constraints.

Conclusions:

• AOGWO is a robust and dynamically adaptive optimization framework.
• The proposed enhancements effectively address premature convergence and elite stagnation.
• AOGWO represents a promising innovation for practical optimization, economic productivity, and resource efficiency.