IAROA: An Enhanced Attraction-Repulsion Optimisation Algorithm Fusing Multiple Strategies for Mechanical Optimisation...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• Attraction-Repulsion Optimization Algorithm (AROA) is a novel metaheuristic for global optimization.
• AROA balances development and exploration but suffers from limited solution diversity, convergence precision, and local stagnation.

Purpose of the Study:

• To enhance the global search capability and applicability of the AROA algorithm.
• Introduce an Improved Attraction-Repulsion Optimization Algorithm (IAROA) by integrating multiple advanced strategies.

Main Methods:

• Implemented an elite dynamic opposite (EDO) learning strategy for initial solution enrichment.
• Incorporated dimension learning-based hunting (DLH) for increased solution diversity and exploration balance.
• Utilized a pheromone adjustment strategy (PAS) to accelerate convergence and extend search range.
• Introduced Cauchy distribution inverse cumulative perturbation (CDICP) for improved local search and avoidance of local optima.

Main Results:

• IAROA demonstrated superior optimization precision, solution stability, and convergence speed compared to AROA and 13 other classical algorithms.
• Performance was validated on CEC2017 test functions and six complex engineering design problems.
• The algorithm showed high competitiveness in solving constrained engineering design problems.

Conclusions:

• The proposed IAROA significantly improves upon the original AROA's performance.
• IAROA offers enhanced applicability, effectiveness, and robustness for complex optimization tasks.
• This enhanced algorithm is a promising tool for solving challenging real-world engineering design problems.