Efficient Large-Scale Multiobjective Optimization Based on a Competitive Swarm Optimizer.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Large-scale multiobjective optimization problems (MOPs) present significant challenges due to their vast decision spaces.
• Existing algorithms, particularly those using decision variable division, struggle with complex problem landscapes, leading to inefficient or inaccurate solutions.

Purpose of the Study:

• To develop a more efficient and effective algorithm for solving large-scale MOPs.
• To address the limitations of current decision variable division-based approaches in handling complex MOPs.

Main Methods:

• Proposes a competitive swarm optimizer (CSO)-based approach for large-scale MOPs.
• Introduces a novel two-stage particle updating strategy to enhance search efficiency.
• Evaluates the algorithm on benchmark MOPs and a real-world application example.

Main Results:

• The proposed CSO-based algorithm demonstrates superior performance compared to several state-of-the-art multiobjective evolutionary algorithms.
• The two-stage particle updating strategy significantly improves search efficiency in large-scale MOPs.
• The algorithm effectively handles problems with complicated landscapes where other methods falter.

Conclusions:

• The novel CSO-based algorithm offers an effective solution for large-scale MOPs.
• The proposed strategy enhances search efficiency and accuracy, particularly for complex optimization problems.
• This work advances the field of multiobjective optimization by providing a more robust and efficient tool.