A Promotive Particle Swarm Optimizer With Double Hierarchical Structures

Area of Science:

• Computational Intelligence
• Swarm Intelligence
• Optimization Algorithms

Background:

• Particle Swarm Optimization (PSO) is a widely used metaheuristic algorithm.
• Existing PSO variants face challenges in balancing exploration and exploitation, especially in complex search spaces.
• Fair competition and diversity maintenance are crucial for effective swarm intelligence.

Purpose of the Study:

• To propose a novel Promotive Particle Swarm Optimizer (PPSO) with double hierarchical structures.
• To enhance the performance of PSO by introducing mechanisms for fair competition and increased diversity.
• To improve accuracy and convergence speed in solving complex optimization problems.

Main Methods:

• Dividing the swarm into hierarchical, independent subpopulations for parallel search.
• Implementing a unidirectional communication strategy and a promotion operator for inter-subpopulation particle exchange.
• Constructing a hierarchical multiscale optimum within subpopulations using a tiered particle architecture.
• Utilizing double hierarchical structures to protect promising particles and increase search diversity.

Main Results:

• The proposed PPSO demonstrated improved accuracy and convergence speed on 30 benchmark problems.
• Performance gains were particularly notable in solving complex optimization tasks.
• Statistical analysis confirmed the effectiveness of the double hierarchical structures and promotion mechanisms.

Conclusions:

• The novel PPSO with double hierarchical structures offers significant advantages over existing PSO variations.
• The proposed method effectively balances exploration and exploitation, leading to superior optimization performance.
• This approach provides a robust framework for tackling complex computational intelligence challenges.