Research On Hyper-heuristic Approach For Molds-Constrained Flexible Job-shop Problem

Flexible job-shop scheduling problem(FJSP)is an important branch of current production scheduling problem,which inspired by production and manufacturing scenarios in reality.The goal of the research on FJSP is improving the efficiency of production management,allocating all kinds of resources scientifically and creating production lines of high efficiency in the actual production management.With the development of various algorithms,the research on flexible job shop problem has reached a new height.However,in the customized production environment,the mold is a very important production resource for the production of different types of ordered products.The restriction of the mold resource and the sequence-dependent set-up times(SDST)generated by the replacement of the mold are two problems that cannot be ignored in the production scheduling of some fields.In recent years,the hyper-heuristic algorithm provides a new way of thinking of solution for complex operations optimization problems.This algorithm is based on two main points: a method pool composed of several low-level heuristics(LLH)and a highlevel heuristic(HLH).The algorithm can manipulate LLHs to generate a new algorithm which can optimize the solution of actual problem.This method has the characteristics of less super parameters and high generality.In order to improve the problem solving in the actual production process,there are two points must be considered which are the scheduling model which is closer to the actual situation and the solution algorithm with stronger generality.The main research contents of this paper are as follows:(1)A high-efficiency sequence selection hyper-heuristic(SSHH)algorithm is proposed,which constructs the LHHs pool for double-layer coding,and employs a more reasonable exponential function selection strategy.We study the FJSP with SDST model,and use SSHH to solve FJSP and FJSP with SDST.It is verified that this algorithm has good ability of solving problems and strong robustness.(2)Inspired by a series of actual customized production such as the cable production workshop,A model of molds-constrained FJSP with SDST(MCFJSPSDST)with the goal of minimizing the maximum completion time was studied.The model not only considers the problem of machine assignment and process ordering of traditional FJSP,but also takes into account the consideration of mold group.For each process to be produced,a specific set of molds is required on each machine to assist production.And the replacement time of the mold will also affect the total time consumed by the production plan.In order to study the model better,118 experimental cases were constructed based on the standard FJSP cases.And an improved SSHH algorithm was used to solve the problem.It is proved that the problem can be solved better when the decoding strategy is well designed.