A hybrid particle swarm optimization algorithm for solving engineering problem.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Nature-Inspired Computing

Background:

• Particle Swarm Optimization (PSO) can suffer from premature convergence and local optima.
• Developing robust optimization algorithms is crucial for complex problem-solving.

Purpose of the Study:

• To propose an improved Particle Swarm Optimization (PSO) algorithm, named NDWPSO, to address limitations of premature convergence and local optima.
• To enhance global search speed and convergence rate through hybrid strategies.

Main Methods:

• Utilized elite opposition-based learning for particle initialization.
• Implemented dynamic inertial weight parameters for early-stage global search.
• Introduced a novel local optimal jump-out strategy.
• Incorporated Whale Optimization Algorithm's (WOA) spiral shrinkage and Differential Evolution's (DE) mutation strategies for late-stage convergence.

Main Results:

• NDWPSO outperformed 3 PSO variants across all 49 datasets.
• Achieved 69.2% to 84.6% of best results on benchmark functions across dimensions (30, 50, 100).
• Secured 80% of best optimal solutions for 10 fixed-multimodal benchmark functions.
• Delivered optimal solutions for 3 practical engineering problems.

Conclusions:

• The NDWPSO algorithm effectively overcomes premature convergence and local optima.
• NDWPSO demonstrates superior performance compared to other nature-inspired algorithms on benchmark and engineering tasks.
• The hybrid strategies significantly enhance both global exploration and local exploitation capabilities.