Research On Flexible Job Shop Scheduling Considering Outsourcing And Optimization Algorithm

With the rapid development of manufacturing and intelligent technology,market demand is increasingly diversified and personalized,so small and medium-sized batch,customized flexible production model emerges.Flexible job shop scheduling problem(FJSP)which meets the personalized requirements has gradually become a research hotspot in the field of job shop scheduling.Meanwhile,outsourcing strategy as another flexible way to respond to market demand is widely used in the actual scheduling process.But there is no research on FJSP considering outsourcing.Nowadays,most studies of scheduling problem involving outsourcing assume that the capacity of the machine in the outsourcing enterprise is unlimited.In the real manufacturing system,outsourced jobs can only be produced by outsourcing enterprises within available time defined as the service time window(STW).Although the manufacturing mode of multiworkshop unit joint production becomes popular,the distributed flexible job shop scheduling problem based on outsourcing is rarely studied.In conclusion,the main research contents of the thesis are indicated below:(1)Aiming at the outsourcing strategy and different shop types,two kinds of shop scheduling problems based on outsourcing are proposed: the flexible job shop scheduling problem based on outsourcing(FJSPO)and the distributed flexible job shop scheduling problem considering outsourcing(DFJSPO).Among them,FJSPO considers STW and operation transportation comprehensively,and the optimization objective is to minimize the maximum completion time and total cost.In addition,DFJSPO,which is based on FJSPO,further considers the outsourcing factories in different geographical locations,STW and operation transportation between different factories,coordinating production resources of internal and external plant to optimize the makespan and total cost of the internal plant.The above FJSPO and DFJSPO scheduling models are constructed,and the characteristics of the models are explained by related examples.(2)The improved memetic algorithm(IMA)is designed to solve FJSPO.Encoding/decoding method,crossover operator and mutation operator are put forward based on the characteristics of FJSPO.And the initialization method mixes three priority scheduling rules.Besides,critical operations,critical blocks and critical path are defined to design two kinds of neighborhood structure for local search.Firstly,twenty FJSPO benchmarks are constructed,and the performance of IMA algorithm is verified after the optimal parameter combination is obtained by DOE.The experimental results show that the solutions of IMA algorithm are more evenly in solving FJSPO model compared with NSGA-II,NNIA and MOPSO.(3)The hybrid memetic algorithm(HMA)is developed to solve DFJSPO,in which DFJSPO-related encoding/decoding methods,crossover operators and mutation operators are employed.The heuristic algorithm based on decoding method and random selection method are combined in the initialization stage.The critical path for the local search acquired to design three types of neighborhood structures.Thirty DFJSPO benchmarks of different sizes are constructed,and the best parameter combination of HMA is obtained.After verifying the validity of initialization method and local search method of HMA,it is proved that most of the solutions obtained by HMA algorithm in solving DFJSPO model belong to the global optimal Pareto solution compared with NSGA-II,NNIA and MOPSO.