Design and Analysis of a Gearbox Predictive Maintenance System

In today's world of industry and manufacturing, the maintenance of machines and their components plays an important role. Each year companies invest large sums of money in routine and preventive maintenance, based on schedules defined without knowledge of the actual state of wear and tear of machines. Predictive maintenance, on the other hand, allows companies to perform restorations on machines only when necessary, avoiding needless interventions and cutting the cost of maintenance. In this work, we will present a predictive maintenance system applied to gearboxes. System designed not only to recognise and anticipate failures, but also to estimate the remaining useful life of the gearbox under analysis. We will present its structure and operation, and then analyse its performance through ad-hoc defined surrogate metrics, made necessary by the absence of ground truth on gearbox wear.