A Dynamic Neighborhood-Based Switching Particle Swarm Optimization Algorithm.

Area of Science:

• Computational Intelligence
• Swarm Intelligence
• Optimization Algorithms

Background:

• Particle Swarm Optimization (PSO) is a widely used metaheuristic algorithm.
• Premature convergence and limited exploration are common challenges in standard PSO.
• Existing PSO variants often struggle with complex, multimodal optimization landscapes.

Purpose of the Study:

• To introduce a novel dynamic-neighborhood-based switching PSO (DNSPSO) algorithm.
• To enhance the exploration and exploitation capabilities of PSO.
• To improve performance on challenging optimization problems, especially multimodal ones.

Main Methods:

• A dynamic-neighborhood-based velocity updating mechanism utilizing distance-based information.
• A switching learning strategy for adaptive selection of acceleration coefficients.
• Hybridization with Differential Evolution to mitigate premature convergence.

Main Results:

• The DNSPSO algorithm demonstrated superior solution accuracy compared to existing PSO methods.
• DNSPSO exhibited improved convergence performance across various benchmark functions.
• Significant advantages were observed for complex multimodal and rotated multimodal optimization tasks.

Conclusions:

• The proposed DNSPSO algorithm effectively addresses limitations of traditional PSO.
• DNSPSO offers a robust and efficient approach for complex optimization problems.
• The dynamic neighborhood and adaptive learning strategies contribute to enhanced search capabilities.