MA63562B1 - Device and method for intelligent diagnosis and robust prognosis in predictive maintenance - Google Patents

Abstract

The present invention solves the problems of rotating machine failure by offering an intelligent system for robust diagnosis and prediction of interventions, which can be customised as required. This system comprises two complementary subsystems, the first of which predicts anomalies via machine learning, in this case the FPT (First Prediction Time) of each component affected by the anomaly. The second predicts the remaining useful life (RUL) of these components before end of life (EoL). The robustness of our system's predictions is justified by the association of each point of the RUL with a reliability threshold. The system is therefore capable of providing not only an early warning of required maintenance and detailed information on the necessary intervention, such as the type of intervention to be carried out and the time before the intervention, but also a health status of the components over the RUL horizon. To this end, our system allows users to easily prepare and plan their interventions, monitor the health status of their equipment, while ensuring high machine availability and the expected reliability of rotating systems.