Research Of Flexible Job Shop Batch Rescheduling Methods

With the development of artificial intelligence,the earth-shaking changes take place in manufacturing.Many new manufacturing conceptions appear,such as Intelligent Manufacturing,Intelligent Factory and so on.If enterprise do not want to merge in the stream of competition,they must improve the ability to respond to the market.What's more,they have to be able to adapt many varieties,small batch production mode,and deal with dynamic events such as order inserting,cancellations,machine breakdown and lack of raw materials,which appear in a production environment,and has rapid response and handling mechanism.The traditional static scheduling already unable to apply to the dynamic change of production environment,so the dynamic scheduling problem also more and more get the attention of many scholars.Because in the actual production process,most of the enterprises to adopt batch production,so the study of the dynamic scheduling problem for batch has very important research value and practical significance.Firstly the paper describes the dynamic scheduling problem of flexible job shop,and summarizes the current measures of dynamic scheduling.Then through defines the variables and the model hypothesis,base the three optimization objectives: makespan,tardiness penalties,tardiness number,the paper established the mathematical model.And discusses the problems of dynamic scheduling framework,according to the demand of the actual production,puts forward the concept of tardiness tolerance,and for machine breakdown,two dynamic events are given to improve scheduling framework based on hybrid driven mechanism.Secondly,given the scheduled results,the paper done lots of research about how to rescheduling when the order inserting and machine breakdown events happened.Based on swarm intelligence and GA,proposed scheduling method for order inserting problem,from the choice of batch,batch,and the quantity distribution,put forward a set of feasible solutions.According to machine breakdown problem,firstly expounds the process and the mechanism of reinforcement learning algorithm.Then built the mathematic model for machine breakdown,and design environment,division of state and return function and action set.Through the Python implementation algorithm and using two kinds of working condition: the single operation with multiple operation to training.And by comparing with separate rules,verify the effectiveness of the algorithm.Thirdly based on theory and algorithm research,to facilitate the practical application, developed an integrated scheduling prototype system,which set task management,resource management,feedback information as a whole,and puts forward the algorithm,human-computer interaction and rule as the core of the scheduling model.In the end,through the instance of the system was tested.At last,the main work of the thesis is summarized,and the current research deficiencies and future research methods are put forward.