Event-based optimization control of conveyor-serviced production station

This paper is concerned with event-based optimization control of conveyor-serviced production station (CSPS), in which two event-based methods are introduced. One is that a general event is composed of state and the meta-event. The other maps the vacancies of the buffer into some state-cluster events, which are fully-full event, semi-full event, semi-vacant event, and fully-vacant event. Then, Q-learning based on the two methods is proposed. Finally, the simulation results illustrate the effectiveness of the proposed two event-based optimization control methods for CSPS. It shows that the former event-based optimization control method can obtain a better policy, but spends much longer time on learning, since it has larger event space. The later event-based optimization control method spends much shorter time on learning, which needs smaller simulation data samples for training to get good results.