Dual-Subpopulation Competitive Particle Swarm Optimization with Engineering Applications.

Area of Science:

• Computational Intelligence
• Swarm Intelligence
• Optimization Algorithms

Background:

• Particle Swarm Optimization (PSO) is a popular bio-inspired algorithm.
• Balancing exploration and exploitation in PSO is a persistent challenge.
• Existing PSO variants often use static strategies, limiting adaptability and risking premature convergence.

Purpose of the Study:

• To propose an adaptive PSO variant addressing exploration-exploitation balance.
• To enhance optimization performance and robustness through dynamic resource allocation.
• To introduce Dual-Subpopulation Competitive Particle Swarm Optimization (DCPSO).

Main Methods:

• Partitioning the PSO population into distinct exploration and exploitation subpopulations.
• Implementing a dynamic competition mechanism to assess search progress.
• Utilizing adaptive particle migration between subpopulations based on performance.

Main Results:

• DCPSO demonstrated superior performance compared to standard PSO and other state-of-the-art algorithms.
• Statistically significant improvements were observed across most CEC2017 benchmark functions, especially hybrid and composition problems.
• DCPSO showed robustness, stable convergence, and practical effectiveness in engineering design problems.

Conclusions:

• DCPSO effectively balances exploration and exploitation through dynamic subpopulation management.
• The proposed adaptive migration strategy enhances computational resource allocation.
• DCPSO offers a robust and effective alternative for complex optimization tasks.