Coordinated Scheduling Of Production And Transportation In A Two-stage Assembly Flowshop

The two-stage assembly flowshop scheduling problem(TSAFSP)is a classical combinatorial optimization problem,which is widely existed in manufacturing enterprises.With increased global competition,the manufacturing industry has been focusing on activities that transport the products at right time to meet the customer's demand instead of their own.Therefore,this study focuses on the two-stage assembly flowshop scheduling problem with batch delivery(TSAFSP-BD).To enhance the overall performance of production efficiency and decrease the sum of holding costs and earliness?tardiness costs etc.,this problem considers the coordination of production and transportation in a two-stage assembly flowshop environment.The TSAFSP-BD can only use a near-optimal algorithm to obtain the solution because of the strong NP-hard nature.A new hybrid meta-heuristic(GA-OVNS)is presented,which distinguishes itself from other existing meta-heuristics by the hybridization of Genetic Algorithm(GA),Opposition-Based Variable Neighborhood Search(OVNS).GA is a well-known and widely used meta-heuristic for global search with crossover and mutation operation,while lacks the capability of linkage learning and gets bad performance in local search,which leads to premature convergence.As an effective local search procedure,VNS(Variable Neighborhood Search)obtain the optimal solution by searching the neighborhood structure.But the performance of VNS greatly depends on the initial solution,so it always integrated with other evolutionary algorithms.To enhance the search efficiency of VNS,the idea of OBL(Opposition-Based Learning)is adopted to establish some novel opposite neighborhood structures for local search.To the best of our knowledge,GA-OVNS hybridizes GA,OVNS to address TSAFSP-BD,can improve the quality of solution.Research progress trends are as follows.To tackle the single customer TSAFSP-BD,the population initialization of GA-OVNS initial solution involves two decision problem,namely TSAFSP and job allocation to batches.Accordingly,a matrix of two rows and n columns is applied to represent an individual.To improve the performance of GA-OVNS,Shortest Processing Time(SPT)based heuristic are employed to construct some initial individuals.The two-point crossover and modified mutation are developed to GA-OVNS,and the optimal solution is taken as the initial solution of OVNS to guide the search for more promising regions of the solution space.Moreover,motivated by the opposition concept,OBL is integrated with VNS to establish three novel opposite neighborhood structures form Insert,Swap,Inverse for local search to enhance the performance.For the TSAFSP-BD with multiple customers,we need to decide the jobs of every one.Therefore,the encoding,decoding,crossover,mutation and other operations of GA-OVNS must be adjusted,to avoid confusion.In the real production,we usually need to optimize multiple objectives at the same time.So the objective of this essay is to minimize the sum of holding costs,earliness?tardiness costs,delivery cost.To validate the effectiveness of the proposed algorithm,GA-OVNS is compared with some potential competitive algorithms,including GA,GA-VNS,and two fast heuristics namely EDD(Earliest Due Data Rule),SLACK(Slack Time Rule).The based parameter values are obtained by the Taguchi experimental.They are validated n a set of randomly generated instances,and the computation results indicated the superiority of GA-OVNS in quality of solutions.