AS2M department
Data science and System Health Assessment

You are here


DATA-PHM Team (Prognostics and Health Management)


The increased complexity of systems integrating multiphysics components of highly diverse natures drastically changes the methods used to study how these systems age.

Indeed, historically these systems possessed few failure modes and even fewer embedded sensors enabling surveillance of progressive degradation; prediction of lifetime was thus essentially based on pre-established models of deficiency.

Henceforth, the modeling of systems behaviors will very likely be more complex, but a great number of sensors will be embedded so as to render their real state easier to observe. For example, the automobile of the future will probably need tune-ups based on a prognosis of the aging of its components as calculated in real time, rather than on a pre-defined calendar.

Goals and Research Areas

The scientific corpus of our  DATA-PHM team is the development of advanced algorithms for classification, prediction and decision.
The system’s observation phase includes an indispensable and determining step, often neglected as it is associated with an engineering approach.

The team’s three themes rest upon what are, in fact, the three pillars of the DATA-PHM domain:

  •  Pre-processing of data : The performances of data-oriented approaches to prognosis depend upon, among other features, the form of the descriptors (characteristics of the signal) extracted from raw data. These descriptors, which clearly reflect degradation over time, enable greater precision in predictions of the evolution of the state of health of the system under consideration. The PHM team is thus working on a new approach for the extraction and selection of descriptors of vibratory data.
  • Prognosis guided by data : The data-guided prognostic approach targets the transformation of monitoring signals and operational data into actionable information as to the state of degradation in a system and its state of health.
  • The decision process : It makes sense to observe a complex system or critical component, to determine their state of health and to plan for their remaining duration before failure only if this can lead to the development of a decision-making process. We must be able to decide—prior to failure—whether or not to discontinue the use of any given equipment.


The team’s strength is its capacity to incorporate all PHM skills within its organization, from the observation all the way to the decision phase, including analysis.

Implementing our work

Closely associated with systems surveillance, the DATA-PHM domain is inherently transversal in nature. Much of the team’s research is therefore undertaken in partnership with three of the FEMTO-ST Institute’s other departments—ENERGY, APPLIED MECHANICS and DISC—on problems concerning respectively the aging of fuel cells, composite materials and observations from sensor networks.

Last modified: