China Mechanical Engineering ›› 2021, Vol. 32 ›› Issue (09): 1126-1133.DOI: 10.3969/j.issn.1004-132X.2021.09.014

Previous Articles     Next Articles

An Improved ACO Algorithm for Automobile Mixed-flow Assembly Scheduling Problems#br#

LI Yi1;TANG  Qian1;LIULianchao1;PENG Xiaogang2;YAN Xianhong2   

  1. 1.State Key Laboratory of Mechanical Transmissions,Chongqing University,Chongqing,400044
    2.Chongqing Changan Automobile Co.,Ltd.,Chongqing,400023
  • Online:2021-05-10 Published:2021-05-28

基于改进蚁群算法的汽车混流装配调度模型求解

李燚1;唐倩1;刘联超1;彭小刚2;颜先洪2   

  1. 1.重庆大学机械传动国家重点实验室,重庆,400044
    2.重庆长安汽车股份有限公司,重庆,400023
  • 通讯作者: 唐倩(通信作者),女,1969年生,教授、博士研究生导师。研究方向为智能制造技术及智能算法的应用。发表论文40余篇。E-mail:tqcqu@cqu.edu.cn。
  • 作者简介:李燚,男,1996年生,硕士研究生。研究方向为智能制造、车间调度、生产线仿真。
  • 基金资助:
    国家重点研发计划(2018YFB1701203);
    汽车整车生产线自适应智能管控系统研制及应用项目(cstc2019jscx-mbdxX0014)

Abstract: Since overload work stations were common and the vehicle assembly quality could not be guaranteed in mixed-model lines which involved a great deal of manual and human-machine operations, a bi-objective car sequencing model was proposed with load balancing at the bottleneck selection as a primary criterion, and minimizing the number of processing delay when considering the replacement and advanced operation time as a secondary criterion. Besides, an improved ACO was designed which used a specific heuristic function in the processes of pheromone global update and probability transfer rules, and changed the evaluation method of the optimal solution in the iterative processes. Compared with traditional ACO and genetic algorithm, computational experiments illustrate that both objectives may be effectively solved using the presented ACO algorithm. In addition, the algorithm may also reversely compute the planned production cycle when the second objective value is zero, which has certain guiding significance in vehicle production.

Key words: mixed-flow assembly, car sequencing, flow shop scheduling, multi-objective optimization, ant colony optimization(ACO) algorithm

摘要: 针对某汽车总装车间混流装配过程涉及大量人工以及人机协同操作而导致工位过载、整车装配质量无法得到保证的问题,建立了瓶颈选装工位负载平衡化、考虑换装与提前作业时间的加工滞后次数最小化的分层序列双目标优化模型,同时设计了一种改进蚁群算法。该算法在信息素全局更新以及概率转移规则过程中,使用一种特定启发式函数,并更改迭代过程中最优解的评价方法。仿真对比实验结果表明,该算法在优化目标函数过程中的收敛速度、收敛精度、最优解质量等方面均优于传统蚁群算法和对比遗传算法,验证了模型和算法的有效性。此外,该算法还可反向求解加工滞后次数为零时的计划生产节拍,具有一定的生产指导意义。

关键词: 混流装配, 车辆排序, 流水车间调度, 多目标优化, 蚁群算法

CLC Number: