基于最优样本和最优属性组合的作业车间调度规则挖掘  

Job Shop Scheduling Rule Mining Based on Optimal Sample and Optimal Attribute Combination

作　　者：张鑫[1] 吕海利[1] ZHANG Xin;LYU Haili(School of Transportation and Logistics Engineering,Wuhan University of Technology,Wuhan 430063,China;不详)

机构地区：[1]武汉理工大学交通与物流工程学院,湖北武汉430063

出　　处：《武汉理工大学学报（信息与管理工程版）》2024年第4期631-636,共6页Journal of Wuhan University of Technology：Information & Management Engineering

摘　　要：作业车间调度问题可使用调度规则解决。为挖掘到高效、准确的调度规则,基于训练样本最优和属性组合最优的核心思想,提出一种基于最优样本与最优属性组合的决策树-遗传算法框架(NDTGA)。该框架在构造训练数据时采用成对比较的方式,在构造属性组合时使用属性原值、差值、对比值等多种组合;在遗传算法的每次寻优过程中,调用决策树挖掘全新的调度规则;最终得到最优训练样本和最优属性组合,进而得到最优的调度规则。通过与经典调度规则和其他机器学习算法的对比实验论证了NDTGA框架挖掘所得调度规则的优越性。The job shop scheduling problem can be solved using scheduling rules.To discover efficient and accurate dispatching rules,a near-optimal scheduling data and attributes based Decision Tree-Genetic Algorithm(NDTGA)framework was proposed based on the core idea of optimal training samples and attribute combinations.This framework used pair-wise comparison when constructing training data.Multiple combinations of attribute original values,differences,and comparison values were used when constructing attribute combinations.Decision tree was called to mine new scheduling rules in each optimization process of the genetic algorithm;Finally,the optimal training sample and optimal attribute combination were obtained,and based on this,the optimal scheduling rules were obtained.The superiority of the NDTGA framework in mining scheduling rules was demonstrated through comparative experiments with classical dispatching rules and other machine learning algorithms.

关 键 词：调度规则 作业车间调度 最优样本 属性组合 决策树-遗传算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]