Predictive model for the degradation state of a hydraulic system with dimensionality reduction

In recent years, the optimization in the use of resources has a key role in achieving a bigger marginality, reducing the operative costs. Due to the advances in the data science field, even the maintenance context is living important changes. The predictive maintenance and the condition-based maintenance can overcome the classic traditional maintenance methods, like the time-based maintenance or the corrective maintenance, with respect to the first intervention, reducing the costs for unscheduled maintenance, manpower, or loss of production and extending the useful life of the components. Based on these presuppositions, the paper proposes the development of a predictive model for the degradation state of the components of a complex hydraulic system, with some tests and some suggestions about the dimensionality reduction. The system has four known types of breakdown, with different degrees of severity; moreover, a fifth parameter represents whether the cycle has reached stable conditions or not.