A new optimization algorithm to solve multi-objective problems.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristics

Background:

• Simultaneous optimization of competing objectives necessitates advanced algorithms.
• Existing multi-objective optimization methods face challenges in capability and solution spread.

Purpose of the Study:

• To propose the novel multi-objective moth swarm algorithm (MOMSA) for solving multi-objective problems.
• To enhance synchronization and maintain a good spread of non-dominated solutions.

Main Methods:

• Developed a new definition for pathfinder moths and moonlight within the swarm algorithm.
• Employed a crowding-distance mechanism for efficient solution selection.
• Utilized an archive to store non-dominated solutions for improved exploration.

Main Results:

• MOMSA demonstrated superior performance across multi-objective benchmark problems (7-30 dimensions).
• Compared to MOEA/D, PESA-II, and MOALO, MOMSA showed better results in generational distance, spacing, spread, and maximum spread.
• Achieved competitive CPU time with high-quality results.

Conclusions:

• The proposed MOMSA is a robust and reliable model for multi-objective optimization.
• MOMSA offers enhanced capability in handling complex, competing objectives.
• The algorithm effectively maintains a good spread of non-dominated solutions.