Symbiosis-Based Alternative Learning Multi-Swarm Particle Swarm Optimization

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Swarm Intelligence

Background:

• Particle Swarm Optimization (PSO) is a widely used metaheuristic.
• Maintaining population diversity is crucial for effective optimization.
• Existing PSO variants may struggle with complex multimodal functions.

Purpose of the Study:

• To propose a novel PSO variant, Symbiosis-based Alternative Learning Multi-swarm Particle Swarm Optimization (SALMPSO).
• To enhance population diversity and optimization performance through symbiotic principles.
• To evaluate the effectiveness of SALMPSO on various benchmark functions.

Main Methods:

• Developed SALMPSO incorporating a learning probability for exemplar selection (center, local best, historical best).
• Implemented two levels of social interaction: within and between multiple sub-swarms.
• Instantiated four variants of SALMPSO based on different learning strategies.
• Tested SALMPSO variants on 15 benchmark functions across 10, 30, and 50 dimensions.

Main Results:

• SALMPSO variants demonstrated superior performance compared to other PSO variants.
• The alternative learning strategy significantly improved convergence speed.
• Enhanced optimal values were achieved on most tested multimodal functions.
• The multi-swarm approach with inter-swarm communication boosted exploration and exploitation.

Conclusions:

• SALMPSO effectively leverages symbiotic cooperation for improved optimization.
• The proposed learning strategy and social interaction mechanisms enhance population diversity and search efficiency.
• SALMPSO offers a promising alternative for solving complex optimization problems, particularly multimodal ones.