Interval Multiobjective Optimization With Memetic Algorithms.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Operations Research

Background:

• Interval multiobjective optimization problems (IMOPs) are prevalent in real-world applications.
• Existing evolutionary algorithms for IMOPs (IMOEAs) require extensive objective function evaluations and yield uncertain Pareto fronts.
• Improving convergence and distribution of Pareto fronts in IMOPs remains a challenge.

Purpose of the Study:

• To propose a novel memetic algorithm (MA) that incorporates local search procedures to enhance IMOP solving.
• To address the convergence, distribution, and uncertainty issues associated with Pareto fronts in IMOPs.
• To improve the efficiency of evolutionary algorithms for interval multiobjective optimization problems.

Main Methods:

• Integration of multiple local search procedures into a state-of-the-art evolutionary algorithm for IMOPs.
• Development of an activation strategy for local search based on hypervolume increment.
• Utilizing hypervolume contribution as a fitness function for local search initialization and genetic operators.

Main Results:

• The proposed memetic algorithm (MA) demonstrated superior performance compared to three state-of-the-art algorithms.
• Empirical evaluation on ten benchmark IMOPs and a solar desalination problem showed significant improvements.
• The MA achieved better convergence and distribution of Pareto fronts with reduced uncertainty.

Conclusions:

• The proposed memetic algorithm effectively tackles interval multiobjective optimization problems.
• Incorporating local search strategies significantly enhances the performance of evolutionary algorithms for IMOPs.
• The MA offers a promising approach for solving complex optimization problems with uncertainty.