Enhanced variants of the Musk Ox Optimizer for global optimization problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• The original Musk Ox Optimizer (MO) demonstrates potential but faces challenges with premature convergence and insufficient exploration.
• Complex optimization problems require robust algorithms capable of effective global search and local exploitation.

Purpose of the Study:

• To develop and evaluate seven novel variants of the Musk Ox Optimizer (MO).
• To enhance the dual-phase search mechanism of MO for improved performance on diverse optimization tasks.

Main Methods:

• Seven new variants (AMOA, CMOA, OBMOA, DEMOA, HMOA-PSO, EOBMOA, QMOA) were algorithmically designed.
• Performance was assessed using 23 benchmark functions, 12 CEC 2022 test functions, and 6 engineering design problems.
• Statistical tests (Friedman, Wilcoxon) were employed to validate performance differences.

Main Results:

• CMOA excelled on unimodal benchmark functions.
• QMOA demonstrated superior performance on multimodal benchmark functions.
• HMOA-PSO yielded the most competitive results for engineering design problems.

Conclusions:

• The enhanced MO variants significantly improve convergence accuracy and solution quality.
• Algorithmic enhancements to MO's search mechanism are effective across various problem types.
• Specific variants show distinct strengths in unimodal, multimodal, and engineering optimization contexts.