Key points are not available for this paper at this time.
In this paper, a real-time discrete-event (DE)-based monitoring system is developed for radio-frequency identification (RFID)-enabled shop-floor monitoring in manufacturing industries. The monitoring system uses rigorous mathematical techniques for event construction, state prediction, and disturbance detection that are suitable for big-data environments of modern complex manufacturing systems. The biggest challenge is to design an efficient scheme for computers to process the event data fast and for engineers to modify the monitoring rules conveniently. First, a DE observer is designed to construct complex events from the simple events extracted from the raw RFID data. The DE observer is based on matrices with binary entries, and thus is easy for multiple users to interpret and modify to define new events or delete event definitions. Temporal relations between time-related events are also included. Second, a hidden Markov model, which considers the impact of user actions and disturbance events, is developed to predict the belief state of manufacturing systems and detect disturbances. Finally, an application case study of the developed system in the shop-floor monitoring of a precision machining parts manufacturing process is provided to show how it can help engineers/managers monitor the events and states efficiently.
Hu et al. (Thu,) studied this question.