Proactive And Reactive Resequencing Of Mixed-model Assembly Lines

Mixed-model assembly line(MMAL)is usually able to produce multiple products with similar structure and process but different models on the same assembly line,which can meet the needs of consumers for product diversity.In fact,MMALs have been widely utilized in the automotive industry.Due to different departments on the MMAL have different preferences for the production sequence,the upstream product sequence need to be adjusted by proactive resequencing to meet the production sequence of the downstream workshop.In addition,unexpected disturbances during the production process will also cause the product to deviate from its position in the planned sequence,so it is necessary to restore the disrupted sequence through reactive sequence adjustment to reduce the fluctuation of the initial production plan.The so-called product resequencing is an important part of production planning and control for MMALs,and the basic form of resequencing includes the physical and virtual resequencing.The physical resequencing requires buffer facilities and generally has better resequencing capabilities,while the virtual resequencing does not need to set up buffers,which can save equipment investment and production area.This thesis focuses on the proactive and reactive resequencing problem before automotive final assembly shops.First,mathematical optimization models are established for different resequencing objectives,and a variety of heuristic algorithms aiming at quickly solving the problems are proposed.Then,the performances of these algorithms are tested by designed instances,and the resequencing ability of selectivity bank is also studied.Finally,the mathematical models and resequencing algorithms described provide a theoreticaland practical guidance for sequence adjustment in the environment of MMALs.The main works and results are as follows:(1)For the proactive resequencing problem based on virtual resequencing before final assembly shops,a mathematical programming model is constructed to describe the problem with the objective of leveling part usage,and three beam search algorithms are proposed.Results of the computational experiments show that the proposed algorithms can solve design instances of three different scales quickly and efficiently.The performance of the iterated beam search algorithm is as good as the business software CPLEX,and the solution time required is much less than CPLEX.(2)Aiming at the problem of reactive resequencing using selectivity banks before assembly shops,a 0-1integer programming model is established to describe the problem with the goal of restoring the initial production sequence.A heuristic rule and three local search algorithms for fast solution are proposed.Results of the Computational experiments show that the four proposed algorithms can adjust the disturbed sequences effectively,and the performance of local search algorithm combined with ant colony algorithm is superior to the other three algorithms.(3)In order to solve the reactive resequencing problem before assembly shops,a series selectivity bank is proposed to restore the upstream sequence.Two resequencing strategies and six heuristic resequencing methods are proposed.Results of the computational experiments show that the proposed methods all have significant resequencing effect and one of the rule-based methods performs best overall.In addition,the resequencing effect of series selectivity banks is better than that of independent selectivity banks.