混流U型拆卸线平衡排序优化模型与算法研究

发布时间：2018-10-22 09:27

【摘要】：电子信息产业市场的驱动技术创新,导致电子产品的更新换代速度加快,许多电子产品在没达到使用寿命时就报废,产生了非常多的电子垃圾,如何回收并合理利用已经成了全球性的难题。因此,本文对面向拆卸多类型电子产品的混流U 型拆卸线平衡排序(Mixed-Model U-Shaped Disassembly Line Balancing and Sequencing Problem,MUDLB/S)问题进行了研究,主要包括以下两个部分:针对循环时间(Cycle Time,CT)为主要约束下的MUDLB/S问题,本文考虑了拆卸任务完成任务时间随机分布,建立了该问题最小拆卸线平均闲置率、尽早拆卸危害和高需求零部件、最小化平均方向改变次数的多目标优化模型;此外,还设计了基于分解和动态邻域搜索的混合多目标进化算法(Hybrid Multi-objective Evolutionary Algorithm Based on Decomposition,HMOEA/D)来同时优化以上多个目标,该算法通过采用弹性任务分配策略、动态邻域结构和动态调整权重以保证解的可行性并搜索得到分布较好的非劣解集。最后,通过正交实验设计了一组算例进行仿真,进而与并行邻域搜索算法和基于局部搜索的遗传算法进行了对比,仿真实验的结果表明了本文所提出算法的优越性。针对固定工作站数量为主要约束的MUDLB/S问题,考虑了拆卸任务完成时间服从随机分布和任务具有位置约束的情况,并建立了以最小化循环时间CT和最小化工作站平均空闲时间为目标的数学模型;此外,设计了改进的并行邻域搜索算法(Improved Parallel Neighborhood Search,IPNS)来依次优化以上目标,该算法定义两类不同的邻域结构,采用动态搜索策略,通过独立搜索以及直接交换邻域的方式以最大限度寻找最优解。最后,通过仿真实验表明,本文所提算法不论在搜索效率还是解的质量方面都优于并行邻域搜索算法。
[Abstract]:The driving technology innovation in the electronic information industry market has led to the acceleration of the upgrading of electronic products. Many electronic products are scrapped before they reach their useful life, resulting in a lot of electronic waste. How to recycle and make rational use of the problem has become a global problem. Therefore, in this paper, the hybrid U-type disassembly line balancing (Mixed-Model U-Shaped Disassembly Line Balancing and Sequencing Problem,MUDLB/S) problem for disassembly multi-type electronic products is studied, which includes the following two parts: for the MUDLB/S problem with cycle time (Cycle Time,CT) as the main constraint, In this paper, the random distribution of disassembly task completion time is considered, and a multi-objective optimization model of minimum disassembly line average idle rate, early disassembly hazard and high demand parts, minimizing the number of average direction changes is established. A hybrid multiobjective evolutionary algorithm (Hybrid Multi-objective Evolutionary Algorithm Based on Decomposition,HMOEA/D) based on decomposition and dynamic neighborhood search is designed to optimize the above targets simultaneously. Dynamic neighborhood structure and dynamic adjustment of weights to ensure the feasibility of the solution and search for a well-distributed set of non-inferior solutions. At last, a set of examples are designed and simulated by orthogonal experiment, and then compared with the parallel neighborhood search algorithm and the genetic algorithm based on local search. The simulation results show the superiority of the proposed algorithm. For the MUDLB/S problem with fixed number of workstations as the main constraint, the random distribution of the completion time of the disassembly task and the position constraint of the task are considered. A mathematical model aiming at minimizing the cycle time (CT) and minimizing the average idle time of a workstation is established. In addition, an improved parallel neighborhood search algorithm (Improved Parallel Neighborhood Search,IPNS) is designed to optimize the above objectives in turn. The algorithm defines two different neighborhood structures and uses dynamic search strategy to find the optimal solution to maximum extent by independent search and direct exchange of neighborhood. Finally, the simulation results show that the proposed algorithm is superior to the parallel neighborhood search algorithm in terms of search efficiency and solution quality.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【相似文献】