Research On Distributed Flexible Job Shop Scheduling Considering Job Transfer And Worker Resource

Recently,the distributed flexible job shop scheduling problem(DFJSP)has gradually attracted the attention of researchers.The research on DFJSP can help companies to fully utilize distributed resources,quickly meet market demands,reduce the production cost,and enhance their competitiveness under economic globalization.The existing researches on DFJSP assume that all operations of a job can only be assigned to one production unit for processing,while a general situation that all operations of a job processed by multiple factories exist in the actual production,which makes the existing researches on DFJSP inconsistent with the reality.Also,in the real manufacturing environment,there are differences in the ability of different workers to operate machines.A reasonable allocation for worker resources to operate machines can stabilize production and reduce worker cost.However,the published researches on DFJSP only focus on the scheduling of factory and machine,and lack of consideration on worker resource,which may lead to unreasonable worker allocation,and ultimately extend the production cycle and increase the labor cost.Based on the above,this paper carried out the following researches:(1)Considering that jobs can be processed by multiple factories in the distributed manufacturing system,The DFJSP with job transfer(DFJSPT)was proposed.A multiobjective scheduling model of the DFJSPT was constructed to minimize the makespan,the maximum machine load,and total energy consumption simultaneously,in which each operation was allowed to be transferred once in the factory or between factories.In terms of how to efficiently assign workers in distributed factories,the DFJSP with worker resource(DFJSPW)was studied.A mathematical model of the DFJSPW was formulated to optimize the makespan,the maximum average worker workload,and processing energy consumption simultaneously,in which each worker can operate multiple machines,and different workers operate the same machine at different times.(2)An efficient memetic algorithm(EMA)was designed to solve the model of DFJSPT.In the EMA,an effective chromosome encoding method was proposed to express the DFJSPT.Based on the encoding method,several decoding,initialization,crossover,and mutation operators were presented.Besides,a local search operation based on a critical path was proposed to optimize the combination population obtained by crossover and mutation operators.To verify the effectiveness of the EMA,40 benchmarks of the DFJSPT were constructed,and the Taguchi method was used to determine the parameters of the EMA before the experiment.The effectiveness of the EMA was verified by comparing three well-known multi-objective optimization algorithms in solving the DFJSPT.(3)An improved memetic algorithm(IMA)was developed for the model of DFJSPW,in which the framework,decoding method,non-dominated sorting of the EMA were applied in the IMA.Also,several operators including encoding,crossover,and mutation were improved,and the effective initialization method and local search operation were designed to improve the performance of the IMA.50 benchmarks of the DFJSPW were constructed to test IMA’s performance.The effectiveness of the initialization method and local search operator was verified through two comparative experiments.The comparison experiment between the IMA and three popular multiobjective optimization algorithms showed that the IMA was efficient in solving the DFJSPW.