Deadlock avoidance in flexible manufacturing systems with concurrently competing process flows

TLDR: The authors develop a Petri net model of concurrent job flow and dynamic resource allocation in an FMS and define deadlock in terms of transition enabling in the PN model to prove that restricted deadlock can never occur for any resource allocation policy implemented under the DAA restriction policy.

Abstract: The concurrent flow of multiple jobs through a FMS can lead to deadlock conditions due to competition for limited resources in the system. The authors develop a Petri net (PN) model of concurrent job flow and dynamic resource allocation in an FMS and define deadlock in terms of transition enabling in the PN model. The problem of deadlock avoidance is addressed by introducing the notion of a restriction policy, which is a feedback policy for excluding some enabled transitions from the current resource allocation alternatives. The authors then present their deadlock avoidance algorithm (DAA) and prove that restricted deadlock can never occur for any resource allocation policy implemented under the DAA restriction policy. The DAA can be implemented easily in real time and is much less restrictive for FMS applications than existing algorithms for deadlock avoidance in computer systems. Application of the DAA is illustrated for three FMS examples: allocation of finite buffer space in a multicell machining facility, collision avoidance in a multirobot assembly cell, and coordination of multiple AGVs on a shop floor. >