论文部分内容阅读
Simulation framework of re-entrant lines with Neuro-Dynamic Programming is designed in this paper. State space is constructed based on the characteristics of re-entrant lines and scheduling problem is described as Markov decision process and simulation offers an efficient way to explore the state space. Different schemes of simulation and parameters adjustment are presented based on different performance indexes, an example are given to illustrate the application of the simulation framework, which also shows the validity of the frame work. The comparison of four scheduling policies indicates the superiority of NDP in scheduling re-entrant lines. The simulation framework can deal with various re-entrant lines and can be extended to other kinds of production system.
Simulation framework of re-entrant lines with Neuro-Dynamic Programming is designed in this paper. State space is constructed based on the characteristics of re-entrant lines and scheduling problem is described as Markov decision process and simulation offers an efficient way to explore the state space. Different schemes of simulation and parameters adjustment are presented based on different performance indexes, an example is given to illustrate the application of the simulation framework, which also shows the validity of the frame work. The comparison of four scheduling policies indicates the superiority of NDP in scheduling re-entrant lines. The simulation framework can deal with various re-entrant lines and can be extended to other kinds of production systems.