Adaptive Gradient Multiobjective Particle Swarm Optimization.

Area of Science:

• Computational intelligence
• Optimization algorithms
• Swarm intelligence

Background:

• Multiobjective optimization problems (MOPs) present challenges in balancing convergence and diversity.
• Existing particle swarm optimization (PSO) variants may struggle with efficient exploration and exploitation in complex search spaces.

Purpose of the Study:

• To introduce an Adaptive Gradient Multiobjective Particle Swarm Optimization (AGMOPSO) algorithm.
• To enhance computational performance, convergence speed, accuracy, and solution diversity in multiobjective optimization.

Main Methods:

• Developed an AGMOPSO algorithm integrating a multiobjective gradient (stocktickerMOG) method for archive updating.
• Incorporated a self-adaptive flight parameters mechanism to balance convergence and diversity based on particle information.
• Analyzed the convergence properties of the proposed AGMOPSO algorithm.

Main Results:

• The AGMOPSO algorithm demonstrated faster convergence speed and higher accuracy compared to other multiobjective PSO algorithms.
• The proposed method achieved a better spread of solutions and improved diversity.
• AGMOPSO showed superior performance against two state-of-the-art multiobjective algorithms in terms of convergence to the Pareto-optimal front.

Conclusions:

• The AGMOPSO algorithm effectively improves computational performance in multiobjective optimization.
• The integration of stocktickerMOG and self-adaptive mechanisms enhances both convergence and diversity.
• AGMOPSO offers a promising approach for solving complex multiobjective problems efficiently.