Scholarlink

<p>The facility layout design problem is a well-known challenge in computer science and operations research, aiming to determine the optimal arrangement of equipment to maximize operational efficiency. Given its computational complexity, effective optimization techniques are crucial for solving this problem. Metaheuristic methods have emerged as a powerful tool to address such NP-hard problems by reducing computational overhead and accelerating solution times. In this paper, a systematic approach to facility layout optimization using the Giza Pyramids Construction (GPC) algorithm is proposed. Furthermore, a novel and enhanced variant of GPC, the Reinforcement Learning-based GPC (RLGPC), which leverages reinforcement learning to refine search efficiency and solution quality, is introduced. To validate the approach, 12 benchmark problem instances are constructed, and comparative evaluations against seven state-of-the-art metaheuristic algorithms are conducted. Additionally, a generalized movement and rotation operator to further enhance solution precision is designed. The computational intensity of evaluating complex layout constraints and the inherent complexity of the RL-driven metaheuristic necessitate high-performance computing resources for practical application. The proposed algorithm is designed for parallelization, making it a suitable candidate for solving large-scale industrial layout problems on supercomputing infrastructures. Experimental results and rigorous statistical analyses demonstrate that the proposed reinforcement learning-based GPC algorithm outperforms competing methods in terms of solution quality and convergence speed. Statistical analysis also shows that the proposed RLGPC algorithm achieves the best average ranking in the Friedman test (mean rank of 1.04) and outperforms all competing algorithms in 11 out of 12 benchmark instances. Rigorous statistical analyses, including the Holm post hoc test, confirmed the significance of these improvements with a <i>p</i> value &lt; 0.00001 for most comparisons. The findings suggest that the GPC-based approach offers an effective, scalable, and computationally efficient framework for tackling facility layout design problems, making it a promising tool for real-world industrial applications.</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

Reinforcement Learning-based Giza Pyramids Construction algorithm for solving layout design problem: a systematic approach

  • 期刊论文

  • 发表日期：

    2025年12月8日

查看全文
  • Sasan Harifi

摘要

The facility layout design problem is a well-known challenge in computer science and operations research, aiming to determine the optimal arrangement of equipment to maximize operational efficiency. Given its computational complexity, effective optimization techniques are crucial for solving this problem. Metaheuristic methods have emerged as a powerful tool to address such NP-hard problems by reducing computational overhead and accelerating solution times. In this paper, a systematic approach to facility layout optimization using the Giza Pyramids Construction (GPC) algorithm is proposed. Furthermore, a novel and enhanced variant of GPC, the Reinforcement Learning-based GPC (RLGPC), which leverages reinforcement learning to refine search efficiency and solution quality, is introduced. To validate the approach, 12 benchmark problem instances are constructed, and comparative evaluations against seven state-of-the-art metaheuristic algorithms are conducted. Additionally, a generalized movement and rotation operator to further enhance solution precision is designed. The computational intensity of evaluating complex layout constraints and the inherent complexity of the RL-driven metaheuristic necessitate high-performance computing resources for practical application. The proposed algorithm is designed for parallelization, making it a suitable candidate for solving large-scale industrial layout problems on supercomputing infrastructures. Experimental results and rigorous statistical analyses demonstrate that the proposed reinforcement learning-based GPC algorithm outperforms competing methods in terms of solution quality and convergence speed. Statistical analysis also shows that the proposed RLGPC algorithm achieves the best average ranking in the Friedman test (mean rank of 1.04) and outperforms all competing algorithms in 11 out of 12 benchmark instances. Rigorous statistical analyses, including the Holm post hoc test, confirmed the significance of these improvements with a p value < 0.00001 for most comparisons. The findings suggest that the GPC-based approach offers an effective, scalable, and computationally efficient framework for tackling facility layout design problems, making it a promising tool for real-world industrial applications.