Scheduling Problem Research Of Re-entrant Manufacturing System Based On Petri Nets

In recent years,the fast development of semiconductor manufacturing brings a greatchallenging in semiconductor companies.In order to effectively improve the productivity,reduce the cost,and make full use of the existing resources to improve competitive powerof enterprises,as a representative of semiconductor manufacturing,the scheduling problemof re-entrant manufacturing is always researched broadly.Re-entrant manufacturing systemhas important characteristics such as re-entrant process flows,complexity,uncertainty.So,itmakes the conventional technologies difficult to get satisfied optimal scheduling policy.Anew efficient methods are needed to improve the performance.Because of these above characteristics of re-entrant manufacturing system,anobject-oriented technique and hierarchical Petri nets modeling are proposed,whichcombined with genetic algorithm to solve the problem of scheduling.The main work ofthis paper includes:（1）The research progress of modeling methods and scheduling strategies forre-entrant manufacturing are systemically reviewed according to the main researchachievements in recent years,and their primary virtues and deficiencies are alsodiscussed.The significance of optimal scheduling both in theory and practice of re-entrantmanufacturing is also set forth.（2）An object-oriented and hierarchical Petri nets are used in semiconductorsystems,which include timed and colored variables,and it was analysed from pathscheduling and workpiece scheduling.In path scheduling,it simulate the transportation ofthe system resources with Petri nets.In workpiece scheduling,according to the process ofsystem,the first work is to build the macroscopic model,divede the process into severalmodules,package up each of them,and then expound the informations among them;thesecond work is to build microscopic model of each module,according to the specificfunction.This modeling method can effectively reduce the complexity and enhance theflexibility of the model.Forthermore,it analyzed the results and proved the correctness and effectiveness of the model.（3）According to the model,a genetic algrorithm is proposed to solve the problem ofscheduling.A binary coding method and C++language was adopted to program the choiceof population,crossover and mutation in it’s specific situation,then the paper compared thediffrernces of the system performance between conventional method and the geneticalgorithm method through the simulation,and prove the genetic algorithm is the effectivemethod to solve the problem of scheduling.