Applying Virtual Resequencing To Mixed-model Assembly Lines

Mixed-model assembly line is generally used by automotive manufacturers in order to meet the need for multi-varieties and small-batch product,therefore multiple products with similar structure and process but different models can be processed in one assembly line.While different workshops have different preferences regarding the feeding sequence,a resequencing operation is usually added between two consecutive workshops by adjusting the upstream sequence to meet the sequencing requirement of the downstream workshop.Physical and virtual resequencing are two basic forms of resequencing.Compared to physical resequencing,virtual resequencing has many advantages,such as it does not need expensive equipment investment and could save space.In this thesis,a series of studies are carried out on how to effectively apply virtual resequencing to achieve sequence adjustment in different workshops according to their resequencing requirements in a mixed-model assembly line context.Mathematical optimization models for virtual resequencing problems under different environments and requirements are established.Various heuristic algorithms aiming at quickly solving the problems are proposed,and their performances are tested by designed instances.The main research work and results are as follows:(1)The optimization problem of applying virtual resequencing in paint shop to achieve color-batching is further studied,and an effective ant colony optimization algorithm is proposed.It is shown that,for small and medium size instances,the proposed algorithm can quickly achieve comparable resequencing performance as commercial software CPLEX.For large size instances,the algorithm produces favorable results in comparison with physical resequencing when applied in a rolling horizon environment.Additionally,considerably better performance is obtained by integrated virtual and physical resequencing than using them individually.(2)To solve the virtual resequencing problem based on car sequencing in final assembly,a mathematical optimization model is first presented to describe the problem,follows by an analysis of its computational complexity.A local search algorithm and two different acceptance criteria are then proposed in order to quickly solve the problem.Results of numerical experiments have shown that the algorithm can produce comparable solutions in comparison with CPLEX in very short time durations,and its superiority grows as the length and options of the sequence increase.Besides,local search with threshold acceptance criterion performs slightly better than that with greedy acceptance criterion.(3)For the integrated resequencing problem based on car sequencing in final assembly,a mathematical optimization model of integrating mix bank and virtual resequencing is originally proposed.Heuristic resequencing algorithms with five bank filling methods and one car releasing method are developed,then these algorithms are combined with virtual resequencing to form integrated resequencing algorithms.The results of numerical experiments have shown that introducing virtual resequencing in releasing stage nearly always improves the performance of physical resequencing significantly.Besides,the genetic algorithm with integrated releasing method always outperforms the other four integrated algorithms,and is more efficient and effective than CPLEX.This thesis studies how to effectively apply virtual resequencing to adjust sequence in a mixed-model assembly line environment.It is found that virtual resequencing can achieve relatively good resequencing performance in some cases.Moreover,integrated resequencing which combines the advantages of physical and virtual resequencing performs better than any single resequencing method.The resequencing optimization models and algorithms proposed in this thesis not only have certain theoretical significance,but could also provide the actual resequencing problems in similar production environment with decision support.