Black-Winged Kite Algorithm Integrating Opposition-Based Learning and Quasi-Newton Strategy.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• The black-winged kite algorithm (BKA) suffers from limited global search capability and declining population diversity.
• Addressing these limitations is crucial for improving the algorithm's effectiveness in complex optimization tasks.

Purpose of the Study:

• To propose an enhanced black-winged kite algorithm (OQBKA) by integrating opposition-based learning and quasi-Newton strategies.
• To improve global search capability, population diversity, local optimization precision, and convergence performance of the BKA.

Main Methods:

• Introduced a mirror imaging strategy based on convex lens imaging (MOBL) for enhanced population distribution and local optima escape.
• Incorporated the quasi-Newton method in later iterations for improved local optimization and convergence.
• Conducted ablation studies on the CEC2017 benchmark set.
• Performed comparative experiments on the CEC2022 benchmark suite.
• Evaluated performance on three constrained engineering design problems.

Main Results:

• OQBKA achieved an average ranking of 1.34 across 29 test functions on the CEC2017 benchmark.
• On the CEC2022 benchmark, OQBKA obtained average rankings of 2.5 (10-dimensional) and 2.17 (20-dimensional), outperforming ten state-of-the-art algorithms.
• Demonstrated superior exploration-exploitation balance and optimization accuracy.
• Successfully generated feasible solutions for complex constrained engineering design problems.

Conclusions:

• The integrated strategies in OQBKA show strong complementarity, significantly enhancing BKA's performance.
• OQBKA offers a robust and practical approach for solving complex optimization problems, including engineering design.
• The proposed OQBKA exhibits superior performance compared to existing metaheuristic algorithms.