Modern manufacturing systems require a high degree of production flexibility to adapt to more personalized market demands in a timely and cost-effective manner, which the Industry 4.0 paradigm's technologies enable. As a result, in today's manufacturing environment, it is critical to optimize the use of these technologies. Simultaneously, to remain competitive, firms must commit to addressing external and/or internal restrictions in the manufacturing system. As a result, considering t he growing interest in artificial intelligence (AI) and the promising results of its industrial application, this paper proposes a novel approach to production control based on Reinforcement Learning (RL) for resolving production scheduling difficulties of varying complexity. In this way, human intervention in production scheduling can be reduced, while planning and decision-making capabilities are improved at the same time. To support this claim, a simulation study was conducted that aims to assess the behaviour of an automated factory regarding various external and internal operational constraints. Consider a Flow Shop production line working in an Industry 4.0 environment capable of adopting Cyber-Physical Systems (CPS) and the Internet of Things (IoT); this study provides a novel flexible dispatching rule based on production line performance monitoring. The performances of the proposed new approach are compared to that of previously suggested dispatching regulations in the scientific literature. The simulation results revealed several intriguing conclusions, emphasizing the rule's flexibility and practical use is given certain practical assumptions.
A Reinforcement Learning approach in Industry 4.0 enabled production system / Marchesano, M. G.; Salatiello, E.; Guizzi, G.; Santillo, L. C.. - In: ...SUMMER SCHOOL FRANCESCO TURCO. PROCEEDINGS. - ISSN 2283-8996. - (2022). (Intervento presentato al convegno 27th Summer School Francesco Turco, 2022 tenutosi a ita).
A Reinforcement Learning approach in Industry 4.0 enabled production system
Marchesano M. G.
;Guizzi G.;Santillo L. C.
2022
Abstract
Modern manufacturing systems require a high degree of production flexibility to adapt to more personalized market demands in a timely and cost-effective manner, which the Industry 4.0 paradigm's technologies enable. As a result, in today's manufacturing environment, it is critical to optimize the use of these technologies. Simultaneously, to remain competitive, firms must commit to addressing external and/or internal restrictions in the manufacturing system. As a result, considering t he growing interest in artificial intelligence (AI) and the promising results of its industrial application, this paper proposes a novel approach to production control based on Reinforcement Learning (RL) for resolving production scheduling difficulties of varying complexity. In this way, human intervention in production scheduling can be reduced, while planning and decision-making capabilities are improved at the same time. To support this claim, a simulation study was conducted that aims to assess the behaviour of an automated factory regarding various external and internal operational constraints. Consider a Flow Shop production line working in an Industry 4.0 environment capable of adopting Cyber-Physical Systems (CPS) and the Internet of Things (IoT); this study provides a novel flexible dispatching rule based on production line performance monitoring. The performances of the proposed new approach are compared to that of previously suggested dispatching regulations in the scientific literature. The simulation results revealed several intriguing conclusions, emphasizing the rule's flexibility and practical use is given certain practical assumptions.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
