Research On Optimization Method Of Dynamic Job Shop Scheduling Rules Based On GEP

With the increasing rapid development of science and technology,all industries are facing technological innovation,and the manufacturing sector is also faced with opportunities and challenges.Scheduling is the key of implementing high productivity,high reliability and flexibility for manufacturing enterprise.In order to improve their own comprehensive competitiveness,the manufacture enterprises pay more and more attention to how to improve the scheduling order in complex and changeable production activity,and how to save cost and increase efficiency so as to meet the needs of the diversification of customers.In summary,the dynamic job shop scheduling problem becomes one of research hotspots in the field of manufacturing.1.At first,we study the basic principle of GEP,by combining with the characteristics of the dynamic job shop,and a research framework of dynamic job shop scheduling problem based on GEP algorithm is set up.This framework includes two stages,namely learning stage and scheduling test stage.GEP in learning stage is used for constructing efficient scheduling rules;then,dynamic job shop simulation model is used evaluate the performance of scheduling rules in the test stage.2.We mainly focuses on the dynamic job shop scheduling framework mentioned above,and constructs a dynamic job shop simulation model with the job arriving at random over time.We propose a new GEP algorithm based double population for single objective dynamic job shop scheduling problem.This algorithm uses multiple gene chromosomes to code,and the evolutionary process consists of two populations: convergence population(CP)and diversity population(DP).The CP mainly maintains the convergence of the algorithm while the DP increases the diversity.The simulation experiments verify the performance of the algorithm and by comparing with other algorithms,our algorithm obtains the better results.3.In the actual job shop,the objective function may be more than one,such as the maximum completion time,delay time and so on.So,the study of multi-objective dynamic job shop scheduling problem is necessary.Based on the single objective dynamic job shop scheduling method mentioned above,we propose a multi-GEP based double population algorithm for multi-objective dynamic job shop scheduling problem.This method adopts the coding method of GEP,two populations co-evolve,and a unique update mechanism is set for specific populations.We test the performance of the proposed algorithm using constructed dynamic job shop simulation model,and our algorithm works better than other methods.