Scholarlink

This study investigates the stochastic resource allocation (SRA) problem, commonly found in complex systems, where the success rates of resources completing tasks are described probabilistically. An adaptive variable neighborhood search (AVNS) algorithm is proposed to address the SRA problem, featuring a permutation encoding method based on pseudo-resources and pseudo-tasks. The AVNS algorithm employs multiple permutation-based operators in its shaking and intensification phases, regulated by an adaptive mechanism to determine the sequence of operator selection. Additionally, a marginal benefit-based constructive algorithm is utilized for initializing the AVNS’s initial solution. To validate the effectiveness of the proposed AVNS, we designed 24 test cases with varying distributions of task values and resource success probabilities, and analyzed the optimality gaps of different algorithms. The computational experiments demonstrate that the AVNS algorithm provides superior allocation schemes compared to existing algorithms in most instances, highlighting its robustness and efficiency in solving complex SRA problems.

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

Solving the Stochastic Resource Allocation Problem Through an Adaptive Variable Neighborhood Search Algorithm

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Weijie Ma,
  • Heng Wang,
  • Jingyu Zhang,
  • Yujue Wang,
  • Rui Yu

摘要

This study investigates the stochastic resource allocation (SRA) problem, commonly found in complex systems, where the success rates of resources completing tasks are described probabilistically. An adaptive variable neighborhood search (AVNS) algorithm is proposed to address the SRA problem, featuring a permutation encoding method based on pseudo-resources and pseudo-tasks. The AVNS algorithm employs multiple permutation-based operators in its shaking and intensification phases, regulated by an adaptive mechanism to determine the sequence of operator selection. Additionally, a marginal benefit-based constructive algorithm is utilized for initializing the AVNS’s initial solution. To validate the effectiveness of the proposed AVNS, we designed 24 test cases with varying distributions of task values and resource success probabilities, and analyzed the optimality gaps of different algorithms. The computational experiments demonstrate that the AVNS algorithm provides superior allocation schemes compared to existing algorithms in most instances, highlighting its robustness and efficiency in solving complex SRA problems.