Reducing the Learning Time of Reinforcement Learning for the Supervisory Control of Discrete Event Systems

Reinforcement learning (RL) can obtain the supervisory controller for discrete-event systems modeled by finite automata and temporal logic. The published methods often have two limitations. First, a large number of training data are required to learn RL controller. Second, algorithms do not consider uncontrollable events, which essential control theory (SCT). To address limitations, we first apply SCT find supervisors specifications automata. These remove illegal violating these hence reduce exploration space algorithm. For remaining logic, algorithm is applied search optimal decision within confined space. Uncontrollable events considered as uncertainties in plant model. proposed method nonblocking supervisor all with less time than methods.