Research On Reentrant Hybrid Flow Shop Scheduling Using Hybrid Genetic Algorithm

Since entering the 21 st century,a new type of manufacturing workshop,the reentrant HFS(RHFS),has attracted domestic and foreign scholars,driven by the internal demand of manufacturing companies to increase production efficiency and increase production flexibility.The basic feature of RHFS is that a workpiece can access certain stations several times.It is a type of production scheduling problem represented by semiconductor manufacturing,steel production,and passenger car manufacturing.The RHFS problem is more complex than the traditional hybrid flow shop(HFS).The research on modeling and optimization theories and methods is a challenging issue.This paper mainly discusses a class of multi-stage RHFS scheduling problems considering the dynamic arrival of workpieces and the transportation time between two adjacent stages.In the past,many scholars discussed the issue of minimizing the maximum completion time(makespan).This goal refers to the time required for the enterprise to complete the processing of all workpieces as required.The shorter the maximum completion time of the workpiece,the higher the efficiency of the enterprise.On the contrary,the slow processing of the company affects the scheduling decision afterwards.The size of the maximum completion time is an important indicator to determine whether the production scheduling is good.Therefore,this paper first studied the issue of minimizing the maximum completion time in the RHFS environment.Considering that the total weighted completion time has also received increasing attention from many scholars in recent years,it is closely related to today's dynamic production environment because many customers want to deliver products at the fastest speed,ideally all Orders can be delivered on time,but it is difficult to achieve in reality.Therefore,how to scientifically and effectively determine the delivery order of different customers is a concern of manufacturing companies.Based on the customer's requirements,different weighting factors are assigned to the completion time of each customer's work piece,and the minimum total weighted completion time is solved to obtain a better scheduling decision.Therefore,based on the study of the makespan problem,this paper discusses the RHFS scheduling optimization with the total weighted completion time as the objective function.First of all,with the above-mentioned cost function related to the completion time as the goal,the process and resource constraints such as machine capacity constraints and priority constraints are extracted from the actual production,and a mathematical planning model is established.Secondly,the characteristics of the model are analyzed and the genetic algorithm is combined with the stereotyped heuristic algorithm.A hybrid genetic algorithm for RHFS problem is proposed.In the algorithm,in order to generate a high-quality initial scheduling solution group,a two-dimensional matrix coding method based on the initial layer-reentrant layer processing system is designed,and a NEH heuristic algorithm is used to perform the initial processing sequence of the workpiece;then,in order to prevent The algorithm converges in advance,and an adaptive dynamic adjustment strategy based on crossover and mutation probabilities is introduced to form the HGA algorithm.Finally,using the proposed algorithm to simulate and test different scale problems,the results show that the proposed algorithm can obtain better near-optimal solution in a shorter computing time,with good performance,and can provide reference for manufacturing enterprises scheduling.