Scholarlink

<p>Multi-objective optimization algorithms play a critical role in addressing complex problems, which often involve the simultaneous optimization of multiple conflicting objectives. In recent years, significant progress has been made in the field of multi-objective optimization. However, numerous challenges still remain. For instance, when solving problems with non-convex and discontinuous Pareto fronts, how to avoid the oscillation of solutions in the objective space; how to balance between diversity and convergence; and the issue of maintaining Pareto boundary solutions. To address these challenges, this paper employs filtering preprocessing to restrict the fluctuations of abnormal solutions, thereby preventing abnormal oscillations in the objective function values of the solutions. To enhance the global search capability, we adopt two competitive learning strategies. One is the convergence-driven individual learning strategy, which facilitates knowledge transfer from high-performance solutions. The other is the near-optimal elite learning strategy. Under this strategy, the algorithm learns from the nearest elite solution, making full use of the advantages of neighboring elite solutions to promote the algorithm’s positive development. The synergistic optimization of these two competitive learning strategies can not only improve the diversity of solutions but also accelerate the convergence speed of the algorithm. The proposed multi- mechanism search method effectively balances the exploration-exploitation trade-off, thus optimizing the algorithm convergence and population diversity throughout the evolutionary process. Comparative evaluations against ten state-of-the-art algorithms on 21 test problems validate the promising performance of the proposed algorithm in terms of optimization quality as well as convergence speed. It is evident that incorporating a multi-mechanism joint search can significantly improve the algorithm’s performance while reducing its IGD value. Furthermore, the proposed algorithm can not only find a set of well-distributed points on the entire Pareto-optimal front but also maintain boundary solutions very well.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-objective optimization based on filtering preprocessing and Near-Optimal individual learning Strategy

  • 期刊论文

  • 发表日期：

    2025年10月9日

查看全文
  • Wei Gan,
  • Zhou Du,
  • Hongye Li,
  • Yingdi Sun

摘要

Multi-objective optimization algorithms play a critical role in addressing complex problems, which often involve the simultaneous optimization of multiple conflicting objectives. In recent years, significant progress has been made in the field of multi-objective optimization. However, numerous challenges still remain. For instance, when solving problems with non-convex and discontinuous Pareto fronts, how to avoid the oscillation of solutions in the objective space; how to balance between diversity and convergence; and the issue of maintaining Pareto boundary solutions. To address these challenges, this paper employs filtering preprocessing to restrict the fluctuations of abnormal solutions, thereby preventing abnormal oscillations in the objective function values of the solutions. To enhance the global search capability, we adopt two competitive learning strategies. One is the convergence-driven individual learning strategy, which facilitates knowledge transfer from high-performance solutions. The other is the near-optimal elite learning strategy. Under this strategy, the algorithm learns from the nearest elite solution, making full use of the advantages of neighboring elite solutions to promote the algorithm’s positive development. The synergistic optimization of these two competitive learning strategies can not only improve the diversity of solutions but also accelerate the convergence speed of the algorithm. The proposed multi- mechanism search method effectively balances the exploration-exploitation trade-off, thus optimizing the algorithm convergence and population diversity throughout the evolutionary process. Comparative evaluations against ten state-of-the-art algorithms on 21 test problems validate the promising performance of the proposed algorithm in terms of optimization quality as well as convergence speed. It is evident that incorporating a multi-mechanism joint search can significantly improve the algorithm’s performance while reducing its IGD value. Furthermore, the proposed algorithm can not only find a set of well-distributed points on the entire Pareto-optimal front but also maintain boundary solutions very well.