Research On Some Critical Issues About Job Shop Real Time Scheduling

One of the most important subsystem in the advance manufacturing system is theJob-Shop Scheduling system. In manufacturing enterprise, it plays an important role ofoptimizing the utilization of resources, emproving work efficiency and saving costs.Due to the rapid development of global economy, competition between manufacturingenterprises is becoming more and more intense. In order to enhance theircompetitiveness, these enterprises increasingly concern on the production schedulingsystem.The classical Job-Shop Scheduling problem has been extensively studied in thepast years, meanwhile, abundant theories about it have been put forward by manyresearchers. However, the frustrating thing is that these theories are hard to be used inthe actual production, because of the unrealistic assumptions in setting up the Job-ShopScheduling model.Now the Job-Shop Online(Real-Time) Scheduling research has become to be a hottopic in advance manufacturing field. Because of its importance, some critical issuesabout Job-Shop Online Scheduling are discussed in this paper. The main contributionsof this paper are shown as follows.①Job-Shop Online Scheduling system cann’t run well in actual production, themain problem is the lack of an effective platform used to simulate and verify thescheduling system or the scheduling optimization algorithms. To solve this problem, anonline scheduling simulation platform based on RFID is discussed, with the hope ofproviding a simulation and verification environment for the theoretical research onJob-Shop online scheduling problem. The components of the platform, as workpiece,manufacturing equipment, logistics system, storage system, communication networkarchitecture and data format are also be introduced.②In a Job-Shop Online Scheduling system, how to improve the flexibility ofproduction scheduling system is an important topic. To it, this paper presents a newmethod of integrating process planning and scheduling system, which is based on someconcepts,. such as, process tree, the set of leaves, the set of idle equipments andoperation process table. With being improved and perfected, the program designed forexperiment and verification can run well on the simulation platform which is discussedearlier. Furthermore, the scheduling module of the program can be easily embedded in a variety of scheduling optimization algorithms.③The basic topic in the research of Job-Shop Scheduling problem is to find abetter optimal scheduling algorithm. To it, this paper presents two methods which areshown as follows:1）Presents a combining rule scheduling optimization method based on doubledeck dispatching rule library. To the combining rule scheduling, the difficulty lies inhow to combine different rules to achieve the best scheduling result. The solution is asfollows: first, build the job selection rule base and the equipment selection rule base.;secondly, based on the objective of the scheduling requirements to sort the rules in therule library using AHP method; finaly, depend on the sequencing results, combine therules from the two rule bases. The results of the experiment verified that the method canrun well in Job-Shop online scheduling optimization.2）Presents an adaptive rule scheduling optimization method based onreinforcement learning theory. Because of the scheduling optimization method based onfixed rules can not adapt to changed environments, the effect of this method is poor. So,the reinforcement learning algorithm is used to train dispatching rules to achieveadaptive rule scheduling; in which, the environment was mapped into a certain patternor state, and the scheduling algorithm intelligently choose the most appropriatedispatching rule, according to the different environments. The results of the experimentverified that the method can run well in Job-Shop online scheduling optimization.