Balancing Constraints and Objectives by Considering Problem Types in Constrained Multiobjective Optimization.

Area of Science:

• Optimization
• Computer Science
• Engineering

Background:

• Constrained multiobjective optimization problems are prevalent in real-world applications.
• Balancing constraints and objectives is critical but challenging due to problem type variations.
• Existing methods lack awareness of how problem types affect constraint handling.

Purpose of the Study:

• To propose a novel constraint handling technique (CHT) that accounts for different problem types.
• To develop a method that balances constraints and objectives effectively across diverse problem structures.
• To improve the performance of multiobjective optimization algorithms on problems with varying constraint-objective relationships.

Main Methods:

• Problems are classified into three types based on the relationship between constrained and unconstrained Pareto-optimal fronts (PF).
• A tailored mechanism is employed for each problem type, adjusting constraint and objective priorities.
• The new CHT is tested within decomposition-based and non-decomposition-based frameworks.

Main Results:

• Experimental studies on 38 benchmark problems and one real-world problem confirm the CHT's effectiveness.
• The proposed CHT achieves a good tradeoff across different problem types.
• The new CHT demonstrates superior performance compared to several state-of-the-art constraint handling techniques.

Conclusions:

• The novel CHT effectively handles constrained multiobjective optimization problems by considering problem types.
• This technique offers a robust solution for balancing constraints and objectives in diverse optimization scenarios.
• The findings suggest a significant advancement in constraint handling for multiobjective optimization.