Scholarlink

<p>In costly engineering experiments, there is frequently a greater interest in regions potentially containing extreme values, including local extrema. This requires the selected samples to not only represent the global response but also focus on regions of interest. This paper proposes a novel adaptive exploration method for parallel optimization called Adaptive Sampling based on Local Penalization (ASLP). The adaptive exploration procedure comprises three stages. Initially, a space-filling design is employed to construct the surrogate model. Subsequently, the adaptive penalization term is incorporated to elevate the sampling priority in regions characterized by extreme values and rapid trend variations. Finally, adaptive sampling is implemented via the Minimum Energy Design (MED) approach. Through strategic parallel optimization, this methodology effectively concentrates on extremal regions while facilitating efficient evaluation of global performance. The effectiveness of the proposed method is verified by several numerical benchmark experiments and simulation experiments of radar anti-jamming performance evaluation.</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

A novel adaptive exploration method for parallel optimization

  • 期刊论文

  • 发表日期：

    2026年6月3日

查看全文
  • Liu Jiawei,
  • Xiao Yike,
  • Duan Xiaojun,
  • Jia Yuanyuan

摘要

In costly engineering experiments, there is frequently a greater interest in regions potentially containing extreme values, including local extrema. This requires the selected samples to not only represent the global response but also focus on regions of interest. This paper proposes a novel adaptive exploration method for parallel optimization called Adaptive Sampling based on Local Penalization (ASLP). The adaptive exploration procedure comprises three stages. Initially, a space-filling design is employed to construct the surrogate model. Subsequently, the adaptive penalization term is incorporated to elevate the sampling priority in regions characterized by extreme values and rapid trend variations. Finally, adaptive sampling is implemented via the Minimum Energy Design (MED) approach. Through strategic parallel optimization, this methodology effectively concentrates on extremal regions while facilitating efficient evaluation of global performance. The effectiveness of the proposed method is verified by several numerical benchmark experiments and simulation experiments of radar anti-jamming performance evaluation.