Kriging Surrogate Model-Based Constraint Multiobjective Particle Swarm Optimization Algorithm.

Area of Science:

• Computational Intelligence
• Optimization Theory
• Engineering Mathematics

Background:

• Constrained multiobjective optimization problems (CMOPs) present significant challenges due to irregular search regions, leading to local optima and poor solution distribution.
• Existing methods struggle with high dimensionality and complex constraints, necessitating improved search efficiency and data structure utilization for nondominated vectors.

Purpose of the Study:

• To develop an effective search method for CMOPs that overcomes local optimization and uneven distribution issues.
• To introduce novel Kriging surrogate model-based operators for enhanced particle swarm optimization (PSO).

Main Methods:

• Design of Kriging surrogate model-based simplex crossover operator (KSCO) for speed update calculations.
• Development of Kriging surrogate model-based local search of simplex crossover operator (KLSSCO) for particle selection in speed updates.
• Implementation of a constrained multiobjective PSO algorithm (KCMOPSO) integrating KSCO and KLSSCO.

Main Results:

• The proposed KCMOPSO algorithm demonstrates improved accuracy in searching both infeasible and feasible regions of CMOPs.
• The algorithm accelerates convergence compared to traditional methods.
• Experimental results indicate KCMOPSO outperforms existing elite methods in solving CMOPs.

Conclusions:

• The novel KSCO and KLSSCO operators effectively enhance PSO for CMOPs.
• KCMOPSO provides a robust solution for complex optimization problems with irregular search spaces.
• The proposed method offers superior performance in terms of accuracy and convergence speed for constrained multiobjective optimization.