Scholarlink

Combinatorial optimization problems (COP) constitute a complex set of problems characterized by discrete decision variables and a well-defined exploration space. Achieving optimality for combinatorial optimization problems typically demands exponential time, leading to their classification as NP-hard optimization problems. With recent advances in computational intelligence, metaheuristic (MH) algorithms have become widely employed for solving COPs. In this research we sought to adopt Machine Learning (ML) algorithm (k-means clustering) at the population initialization stage for multi-objective problems. The multi-objectives considered in this paper aims to minimize flowtime (FT) and energy consumption (EC). Further within each cluster, a Taguchi based parameter selection is being considered to achieve best parameters with in each cluster to enhance MH performance in solution identification with a reduced number of iterations. With the proposed methodology we achieved a performance improvement of 23.5% in flowtime minimization and 28.7% in the minimization of energy consumption. The knowledge gathered from this research provides a useful foundation for investigating novel avenues and refining the methodology for wider implementation and influence.

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

An Integrated Machine Learning Algorithm Based Metaheuristic with Taguchi Method in Solving Energy Efficient Multi-objective Flowshop Scheduling Problem

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Vigneshwar Pesaru,
  • Venkataramanaiah Saddikuti,
  • Mukund Janardhanan

摘要

Combinatorial optimization problems (COP) constitute a complex set of problems characterized by discrete decision variables and a well-defined exploration space. Achieving optimality for combinatorial optimization problems typically demands exponential time, leading to their classification as NP-hard optimization problems. With recent advances in computational intelligence, metaheuristic (MH) algorithms have become widely employed for solving COPs. In this research we sought to adopt Machine Learning (ML) algorithm (k-means clustering) at the population initialization stage for multi-objective problems. The multi-objectives considered in this paper aims to minimize flowtime (FT) and energy consumption (EC). Further within each cluster, a Taguchi based parameter selection is being considered to achieve best parameters with in each cluster to enhance MH performance in solution identification with a reduced number of iterations. With the proposed methodology we achieved a performance improvement of 23.5% in flowtime minimization and 28.7% in the minimization of energy consumption. The knowledge gathered from this research provides a useful foundation for investigating novel avenues and refining the methodology for wider implementation and influence.