Particle swarm optimization with composite particles in dynamic environments.

Area of Science:

• Computational intelligence
• Optimization algorithms
• Swarm intelligence

Background:

• Particle Swarm Optimization (PSO) is increasingly studied for dynamic environments.
• Dynamic optimization problems present unique challenges for standard algorithms.
• Existing PSO methods may struggle with rapidly changing search spaces.

Purpose of the Study:

• To introduce a novel PSO model, PSO with Composite Particles (PSO-CP).
• To enhance the efficiency and speed of PSO in dynamic environments.
• To address the limitations of current PSO algorithms in handling changing conditions.

Main Methods:

• PSO-CP partitions swarms into composite particles based on similarity using a 'worst first' principle.
• Utilizes a velocity-anisotropic reflection scheme for information integration within composite particles.
• Incorporates a scattering operator and an integral movement strategy to maintain swarm diversity.

Main Results:

• Experiments on a dynamic test benchmark demonstrate PSO-CP's effectiveness.
• Provides parameter setting guidelines for PSO-CP.
• PSO-CP shows superior efficiency compared to state-of-the-art PSO algorithms for dynamic problems.

Conclusions:

• PSO-CP offers an efficient approach for dynamic optimization problems.
• The composite particle structure and interaction mechanisms are key to its performance.
• This model represents a significant advancement in PSO for dynamic environments.