Research On Solving Algorithm Of Mixed No-idle Flowshop Scheduling Problem

With the emergence of various new manufacturing modes,the Mixed no-idle permutation flowshop scheduling problem(MNPFSP)widely existing in chip,ceramic,casting,textile and other industries is evolved into various extended scheduling problems to meet specific production requirements.However,there is a relative lack of theoretical research in this area,and only more than 10 relevant literatures can be retrieved in web of science,which seriously reduces the feasibility of using relevant theories to allocate resources reasonably in the actual production of such workshops.In view of this,this paper takes MNPFSP as the research object,considers sequence-based setup time(SDST),and uses Artificial Bee Colony algorithm(ABC)and iterated greedy(IG)algorithm are the main methods.The MNPFSP-SDST with single factory and single objective,the MNPFSP-SDST with single factory and Multi-objective(MMNPFSPSDST),and distributed MNPFSP-SDST(DMNPFSP-SDST)are studied.Combined with the intelligent production line platform,we use a small-scale case to verify the proposed algorithm.The main research contents of this paper are as follows:(1)Aiming at MNPFSP-SDST,a constraint programming(CP)model is established to minimize the makespan,and a calculation method of makespan is given.An Adaptive IG(AIG)algorithm was proposed.In AIG,in order to improve the initial search platform,a heuristic algorithm BS＿NEH is designed to initialize the generation of individual solutions.In order to expand the lateral search range of the algorithm,the destruction and search method based on time node change is used.In the early stage of the algorithm,the SRLS search method is used with large damage values for horizontal expansion,and in the later stage of the algorithm,the RLS search method is used with small damage values for vertical depth exploration.Finally,the proposed CP model and AIG algorithm are compared with other models and algorithms on benchmark examples to verify the effectiveness of the proposed CP model and AIG algorithm.(2)For MMNPFSP-SDST,on the basis of MNPFSP-SDST,the maximum tardiness index is added.This paper establishes a mixed integer linear programming(MILP)model with the goal of makespan and maximum tardiness,and a multi-objective discrete artificial bee colony algorithm(MODABC)was proposed.In MODABC,in order to ensure the population diversity,the initial population is created by NEH,EDD plus random generation,and the new non-dominated solution set ES is established.In the employed bees stage,a variable neighborhood search algorithm composed of two neighborhood structures was used to optimize the population individuals.In the onlooker bees stage,the crossover mode of LOX and PBX was used to enhance the diversity of the population.In the scout bees stage,Multi-objective RLS(MORLS)is proposed to explore the neighborhood solutions of the external archive set ES,and further promote the Pareto front solutions of the population.Finally,compared with other multi-objective algorithms,the effectiveness of MODABC is verified on multiple evaluation indicators.(3)For DMNPFSP-SDST,considering the different SDST of each factory,the MILP model is established with the makespan as the optimization goal,and the Twostage IG algorithm(TIG)is proposed.In TIG,the initialization phase was composed of NEH＿PQ algorithm and job block insertion search.A variable neighborhood descent search strategy consisting of two neighborhood operations is designed to emphasize the global exploration ability of the solution space,and the second stage IG for some key factory is opened at a scheduled time to emphasize the detailed and in-depth local development ability of the solution space.The compared algorithms are tested under different sizes of examples,and the results show that the proposed algorithm can provide better solutions in solving the problem.(4)Finally,with the help of the intelligent production line platform,the three metaheuristic algorithms proposed in this paper are used to solve small cases under different circumstances,and the final results verify the correctness of the proposed algorithm.