Predictive maintenance based on decentralized CPS and convolution’s neural network
Résumé
The massive ingest of non-intrusive sensors on manufacturing processes lifecycle allows to collect and monitor the health state of machines and serve maintenance programmes. In this research paper, we use machines’ monitoring data in a decentralised Cyber Physical System (CPS) architecture implementing Convolution’s Neural Network algorithms to influence the future behaviour of manufacturing equipment in a predictive maintenance perspective. This research contributes to extend the remaining useful life (RUL) of equipment and was implemented in an industrial use case supported by TARDY partner for the predictive maintenance of metal transformation machines.