汽车涂装车间的重排序问题研究

论文总字数：39798字

摘 要

随着客户定制化需求的提升，部分汽车生产商将生产模式改变为混合型装配生产。面向多品种小批量的订单，汽车生产商会根据下游车间的生产需求，在相邻两个车间内对车辆序列或订单进行调整，这个过程为重排序。

本文研究的是涂装车间前的虚拟重排序问题，目的在于通过对订单的重新分配，减少车辆序列的颜色改变次数，降低因清理喷嘴内所引起的加工时间和成本的浪费。本文首次采用蚁群算法对虚拟重排序问题进行求解，采用两种启发式因子为路径选择提供指导，用C 语言编程实现，并对其参数设置进行调试和选择。

为验证蚁群算法的有效性，将其结果与CPLEX在一定条件下求得的解相比较，分析启发式因子对不同算例的影响和蚁群算法的适用性。

对结果分析后得出结论：蚁群算法对涂装车间前的虚拟重排序有改善效果；具有单个启发式因子的蚁群算法对小规模的算例较为有效，50辆车的大规模算例采用引入“试排序”思想的两个启发式因子能有效提高算法的改善效果；且蚁群算法受序列不同的颜色分布的影响较大，对颜色分布均匀的序列改善效果更好，受车型分布的影响较小；蚁群算法的时间较为稳定，对大型规模算例能在较短时间内求解。

关键词：虚拟重排序；蚁群算法；CPLEX；混合型汽车装配

A STUDY ON THE RESEQUENCING BEFORE PAINT SHOP

IN AUTOMOBILE INDUSTY

Abstract

Nowadays, automobile manufactures have applied the idea of resequencing between successive shops by altering the car models or orders to meet the rise of customization in automobile industry.

In order to reduce the cost of cleaning paint nozzles caused by color change between two successive cars, this thesis treats the problem of virtual resequencing before paint shop with the Ant Colony Optimization(ACO). Basic rules and different heuristic factors are discussed, and formulas are given according to the specific problem context. Program is coded based on the platform of Microsoft Visual Studio 2016, using C . Parameters are explained and their combinations are tested on several instances. Then, more instances including different car numbers, color/type numbers, and their distributions are run on this program.

To assure the reliability of the conclusion, another program is coded on CPLEX, which aims at searching for the global best result. The test of this program shows that with the increase of car numbers, it takes much longer time to run, in which case, necessary time restriction is therefore set.

Comparing the results of these two programs, the conclusion is drawn that ACO can effectively reduce the color change of a given car sequence before paint shop, and as for instances with 50 cars, the ACO with two heuristic factors outperfoms ACO with one. And the results also show that ACO can do better with those instances whose colors are distributed uniformly, which suggests that different color distribution of car sequence may have influence on the performance of ACO. Futhermore, the running time ACO needs is more steady and ACO can output a solution efficiently within a short time for instances with larger scale.

Keywords: Virtual resequencing; Ant Colony Optimization; CPLEX; Mixed-model assembly line

目录

摘要 I

Abstract III

目录 V

第一章 绪论 1

1.1 研究背景与意义 1

1.2 国内外研究现状 2

1.3 论文研究内容及意义 4

1.4 论文组织结构 6

第二章 应用蚁群算法进行虚拟重排序 7

2.1 蚁群算法思想及内容 7

2.2 蚁群算法的编程应用 8

2.2.1 信息素与候选集 8

2.2.2 路径选择规则 9

2.2.3 局部信息素更新 11

2.2.4 全局信息素更新 11

2.2.5 C 程序流程 12

2.2.6 算例分类 13

2.3 参数选择与分析 14

2.3.1 蚁群参数 14

2.3.2 单个启发式因子 14

2.3.3 两个启发式因子 18

2.4 本章小结 18

第三章 CPLEX编程应用 19

3.1 ILOG CPLEX简介 19

3.2 使用CPLEX编程 19

3.2.1 数学模型 19

3.2.2 CPLEX模型及说明 20

3.2.3 数据文件 21

3.2.4 运行时间设置 21

3.3 运行结果 23

3.4 本章小结 23

第四章 结果分析 25

4.1 结果比较与分析 25

4.2 本章小结 27

第五章 总结与展望 29

5.1 总结 29

5.2 展望 29

致谢 31

参考文献 33

附录 35

附录A 蚁群算法的C 程序 35

附录B 蚁群算法参数调试 41

剩余内容已隐藏，请支付后下载全文，论文总字数：39798字