The enhancement of elitist whale optimization algorithm with nonlinear parameter for solving real-world single-objective constrained optimization

ABSTRACT: The Elitist Whale Optimization Algorithm with Nonlinear Parameter (EWOANP), despite advancements over the standard Whale Optimization Algorithm, faces challenges in solving complex real-world constrained optimization problems due to inadequate constraint handling, inconsistent performance, lower accuracy, and non-adaptive search mechanisms. This study aims to enhance EWOANP by integrating: (1) €-dominance with dynamic epsilon adaptation for robust constraint handling; (2) an SSA-inspired local search strategy to improve solution accuracy and avoid local optima; and (3) adaptive scaling factors and strategy selection probabilities based on historical performance data. The enhanced EWOANP was evaluated on a subset of benchmark problems from the Real-World Single-Objective Constrained Optimization test suite. Results indicate that the enhancements significantly improved constraint management and solution consistency on several problems, achieving high feasibility rates. The adaptive parameters and local search also contributed to better accuracy where feasible solutions were found. However, challenges in achieving feasibility persisted on extremely difficult, higher-dimensional instances. In conclusion, the proposed enhancements substantially advance EWOANP’s capabilities for constrained optimization. Future work should focus on refining the synergy between the adaptive mechanisms and constraints handling, and on developing more potent diversification strategies for highly complex and high-dimensional problems.

Filipiniana