An adaptive multi-swarm optimizer for dynamic optimization problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Dynamic optimization problems (DOPs) require tracking multiple, shifting optima simultaneously.
• Existing multipopulation methods face challenges in adapting population numbers and diversity to unpredictable environmental changes.
• Effective solutions for DOPs necessitate algorithms that can handle complex and undetected environmental dynamics.

Purpose of the Study:

• To propose an adaptive multi-swarm algorithm for solving dynamic optimization problems (DOPs).
• To enable populations to adapt to dynamic environments without relying on explicit change detection mechanisms.
• To address the challenges of adapting population size and diversity in complex, unpredictable dynamic environments.

Main Methods:

• Development of an adaptive multi-swarm algorithm designed for dynamic environments.
• Implementation of adaptive population mechanisms that do not require prior change detection.
• Experimental validation using the moving peaks problem benchmark.
• Comparative analysis against existing multipopulation-based algorithms in evolutionary computation.

Main Results:

• The proposed adaptive multi-swarm algorithm demonstrates effective population adaptation in dynamic environments.
• The method successfully tracks changing optima without explicit environmental change detection.
• Experimental results on the moving peaks problem validate the algorithm's behavior and performance.

Conclusions:

• The adaptive multi-swarm algorithm offers a robust solution for dynamic optimization problems (DOPs).
• The proposed approach effectively manages population adaptation and diversity in challenging dynamic scenarios.
• This work contributes a novel method for enhancing multipopulation strategies in evolutionary computation for DOPs.