Decentralized line equipment detection and production control by multi-agent technology

The complexity of modern industrial production systems and the persistent connectivity at the local level of manufacturing resources by means of IoT technologies are encouraging the study of increasingly flexible and reconfigurable control methods. Currently industrial software agents are mainly used for controlling job-shop organized systems, relying on negotiation mechanisms between Product agents and Resource agents. In this paper instead, we aim to develop a distributed production line controller, using agent-based technology for designing decentralized and communicating local state machines. The objective of the study is to obtain a global control system deriving from the interaction and exchange of information between software agents associated with each machine on the production line. The controller implemented has the purpose of eliminating the central infrastructure, generally present in industrial automation systems: the goal is to move the decision-making process locally by associating intelligent entities to each machine. The advantage of decentralization lies in the possibility of having a modular, flexible and reconfigurable line controller concept: indeed, there is no supervisor aware of the entire system to control. The agents can also collaborate for a line detection in case of very complex line without there being statistical evidence of performance deterioration. To validate the functionality of this approach, experimental tests were conducted in a virtual environment.