Simulation-based Real-time Production Control with Different Classes of Residence Time Constraints

Residence time constraints are widely observed in production systems, such as semiconductor manufacturing, food industry and battery production, where the time that a part spends in one or several consecutive buffers is restricted. When the residence time of a part is beyond a certain level, the part might face quality problems and need to be further treated or scrapped immediately. To optimize the production performance such as production rate and scrap rate, one needs to properly manage all machines' behavior according to real-time system states to prevent from producing too many inter-mediate parts with high risk of scrap. To solve this problem, a simulation-based real-time control method is proposed to perform production control in face of four basic classes of residence time constraints that are widely seen in semiconductor manufacturing. A Markov Decision Process (MDP) model is first built, and a feature extraction method and a feature-based approximate architecture are proposed to deal with the curse of dimensionality. Simulation is applied in the training to estimate parameters of the feature-based approximate architecture, so the lookahead function in the MDP model can be approximately obtained. Simulation experiments suggest that such a method leads to significant system performance improvement with low computation overhead, which makes real-time production control feasible for longer serial lines with different classes of residence time constraints.