Bare bones particle swarm optimization with scale matrix adaptation.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Swarm Intelligence

Background:

• Bare Bones Particle Swarm Optimization (BBPSO) is effective for continuous optimization but prone to premature convergence in multimodal problems.
• Local optima trap BBPSO, limiting its performance on complex search spaces.
• Existing methods struggle to balance exploration and exploitation effectively.

Purpose of the Study:

• To introduce a novel variant of BBPSO, SMA-BBPSO, designed to overcome premature convergence.
• To enhance the exploration capabilities of BBPSO using a scale matrix adaptation strategy.
• To improve the overall performance and solution-finding effectiveness of BBPSO.

Main Methods:

• Developed SMA-BBPSO, incorporating a multivariate t-distribution for particle position selection.
• Employed a scale matrix adaptation rule based on maximum likelihood estimation of neighborhood best positions.
• Utilized the hierarchical form of the t-distribution, allowing it to encompass normal distributions.

Main Results:

• SMA-BBPSO demonstrated superior effectiveness in finding optimal solutions across various benchmark problems.
• The use of the t-distribution enhanced particles' ability to escape local optima.
• Nonparametric statistical tests confirmed a statistically significant improvement over other swarm algorithms.

Conclusions:

• SMA-BBPSO effectively addresses the premature convergence issue in BBPSO.
• The scale matrix adaptation strategy enhances exploration-exploitation balance.
• SMA-BBPSO represents a significant advancement in swarm intelligence for optimization.