An Adaptive Mutated Multi-objective Particle Swarm Optimization with an Entropy-based Density Assessment Scheme

摘要

In order to solve the problem of premature convergence in MOPSO and attain solutions with good diversity and distribution, a new algorithm has been proposed in this study. The algorithm adopts a new density assessment scheme on the basis of particles' entropy information, which helps to obtain a Pareto set with uniformly distributed solutions. Also an adaptive chaotic mutation operator is designed to avoid premature convergence and help particles explore more efficiently in search space. In addition, the proposed algorithm is validated through comparisons with two existing state-of-the-art multi-objective algorithms using established benchmarks and metrics. Results show that the proposed algorithm shows better distribution performance than the compared algorithms while maintains a good convergence performance.