Research On Production Scheduling Optimization Of Mixed-model Assembly Job Shop With Batch Transferring And Lot Streaming

The manufacturing process of products such as high-speed railway bogies and missile structural parts can be modeled as a kind of assembly job shop scheduling problem.With the diversification of product types and the standardization of the assembly process,the mixed-model assembly line is often used in the assembly stage to carry out the collinear assembly of multiple products.However,the existing research on assembly job-shop scheduling focuses on the single machine or parallel machine at the assembly stage,which cannot meet the requirements of actual production scheduling.Therefore,It is of great engineering and academic value to study the mixed-model assembly job-shop scheduling problem(MAJSP)to reduce setups,work-in-process inventory,and shorten the manufacturing cycle.The specific research contents are listed as follows:Due to the mixed-model assembly of multiple types of products,reducing workin-process inventory from processing to assembly has become an important objective of the MAJSP.Therefore,a lexicographical optimization model and an enhanced simulated annealing algorithm(SA)were proposed to minimize the primary objective,total assembly completion time and the secondary objective,total inventory time.Based on pull production,the lexicographical optimization model is formulated to try to ensure that jobs are processed and transferred when they are needed at each assembly station.For the enhanced SA,a semi-active right-shift two-stage decoding is designed to deal with the primary and secondary objectives.Besides,the algorithm is embedded with an assembly-driven initialization,an improved neighborhood search with a critical block-oriented selection and a filter selection,and a restart mechanism.The experimental results show that the proposed method can obtain the near-optimal solution and even the optimal solution of the scheduling scheme,and can greatly reduce the work-in-process inventory and the cost of production operation.By adopting batch transferring to move the processed jobs to the assembly stations,the transfer frequency and logistics costs can be greatly reduced.Therefore,the MAJSP considering batch transferring is considered.In the constructed mixedinteger linear programming(MILP)model,load constraints of transport devices,batch allocation constraints of jobs,and operation logic constraints among the processing-batch transferring-assembly are added.In the proposed batch transferring knowledge-based SA,a sequence rule considering the processing and assembly features,a batch transferring allocation scheme by a clustering method,neighborhood structures considering the critical transfer batch,a repair mechanism for infeasible transfer batches,a diversity maintenance mechanism based on crossover operator and a restart mechanism considering tempering are proposed.The experimental results show that by transferring jobs with a close processing finish time and assembly start time together,the number of transfer batches can be greatly reduced,which is conducive to the rapid delivery of a large number of jobs to be assembled.The assembly stage may require a large number of certain jobs.If single-piece production is adopted to process these jobs,it will lead to excessive production switching;if batch production is adopted,the flow of production is slow and the manufacturing cycle is long.Thus,lot streaming is introduced into MAJSP.For the MAJSP with lot streaming in the processing stage,new constraints including lot split,sub-lot processing sequence and processing sequence of jobs within a sub-lot are added to construct a MILP model.In the developed adaptive enhanced SA,sub-lot sequence rule considering lot feature,adaptive adjustment mechanism of neighborhood size,adaptive temperature control mechanism,and tabu acceptance mechanism are designed.The experimental results show that by determining an appropriate lot size,the lot difference between the processing and assembly stages can be reduced,so as to achieve the balance between the manufacturing cycle and production setups.Based on the above works,the cooperative effects of mixed-model assembly,batch transferring and lot streaming on MAJSP are considered to solve the joint optimization problem of three-stage production scheduling of processingtransportation-assembly.An improved cooperative SA was designed to respectively construct sub-problems based on three-stage of processing-transportation-assembly,to promote the independent evolution of each sub-problem and the co-evolution of the three sub-problems.In addition,multi-rule-based initialization and reinforcement learning-based dynamic execution mechanism of subproblems are proposed.The experimental results show that the cooperative optimization of lot split,sub-lot sequence,number of transfer batches,transfer batch allocation and product assembly sequencing can shorten the manufacturing cycle,and reduce production setups,workin-process inventory and job transfer frequency.Finally,key operations of the high-speed rail bogie assembly manufacturing process are extracted and modeled,and the theory and method proposed in this paper are verified.The experimental results show that: compared with the two production modes of "processing without lot streaming-batch transferring-mixed model assembly" and "processing with lot streaming-whole set transport-mixed model assembly",the developed production mode of "processing with lot streaming-batch transferring-mixed model assembly" requires shorter manufacturing cycle,fewer setups,fewer transfer times and shorter inventory time,and the comprehensive performance of the scheduling scheme is better,and hence it is more conducive to the fine management of production and logistics in the mixed-model assembly workshop.