Scholarlink

<p>Consideration is given to the resource leveling problem (RLP) in resource-constrained project scheduling (RCPS). Although RLP has gained an increasing research interest, multiple optimization criteria are rarely considered simultaneously in the literature. In this paper, a multi-objective version of RLP is investigated aiming to simultaneously minimize the <i>resource imbalance</i>, the <i>peak of the resource usage</i> as well as the <i>makespan</i>. A metaheuristic algorithm is presented devoted to the search for Pareto-optimal RLP solutions. This algorithm constitutes an adaptation of h-NSDE (the <i>hybrid non-dominated sorting differential evolution</i>) which has recently shown excellent performance over a particular class of machine scheduling problems. Using existing benchmark data sets, we test the performance of h-NSDE in comparison to three of the most famous in the literature multi-objective population-based metaheuristics namely NSGA-II, SPEA2, and PAES. The results obtained are quite promising demonstrating a clear superiority of h-NSDE in terms of both the solution quality and diversity in regard to Pareto-front.</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

h-NSDE—A Solution Algorithm for the Multi-objective Resource Leveling Problem

  • 期刊论文

  • 发表日期：

    2025年1月20日

查看全文
  • Marinos Aristotelous,
  • Andreas C. Nearchou

摘要

Consideration is given to the resource leveling problem (RLP) in resource-constrained project scheduling (RCPS). Although RLP has gained an increasing research interest, multiple optimization criteria are rarely considered simultaneously in the literature. In this paper, a multi-objective version of RLP is investigated aiming to simultaneously minimize the resource imbalance, the peak of the resource usage as well as the makespan. A metaheuristic algorithm is presented devoted to the search for Pareto-optimal RLP solutions. This algorithm constitutes an adaptation of h-NSDE (the hybrid non-dominated sorting differential evolution) which has recently shown excellent performance over a particular class of machine scheduling problems. Using existing benchmark data sets, we test the performance of h-NSDE in comparison to three of the most famous in the literature multi-objective population-based metaheuristics namely NSGA-II, SPEA2, and PAES. The results obtained are quite promising demonstrating a clear superiority of h-NSDE in terms of both the solution quality and diversity in regard to Pareto-front.