IEEE PHM 2014 Data Challenge

The challenge is now closed. Results will be disseminated during 2014 IEEE International Conference on Prognostics and Health Management.

The IEEE Reliability Society, FCLAB research federation, FEMTO-ST Institute, and the Laboratory of excellence ACTION were proud to launch the IEEE PHM 2014 Data Challenge. The challenge was focused on the estimation of the State of Helath (SoH) and on the prediction of the remaining useful life (RUL) of a Proton Exchange Membrane Fuel Cell (PEMFC).


1. Challenge Outline

Fuel Cell Systems (FCS) appear to be a promising alternative energy source to face some of the economic and environmental challenges of modern society. However, even if this technology is close to being competitive, it is not yet ready to be considered for large scale industrial deployment: FCS still must be optimized, particularly by increasing their limited lifespan. Indeed, Proton Exchange Membrane Fuel Cell systems (PEMFC) usually have a life span of around 2000 hours, whereas 6000 hours are required for some applications, including transportation… Enhancing FCS durability involves not only developing a better understanding of ageing phenomena but also requires the ability to emulate the behavior of the whole system to support the development of improvements to those systems. Prognostics and Health Management (PHM) of FCS is an emerging field of scientific and technological development that has the potential to provide and enable improvements in the life management, use and support of Fuel Cell Systems. This was the aim of our challenge: we expected participants to develop robust PHM methods to:

  • to assess the health state of a FC (State of Health – SoH);
  • to predict its remaining useful life (RUL).


2. Ageing tests performed for the challenge

Challenge datasets are provided by FCLAB Research Federation (FR CNRS 3539, France). Experiments were carried out on test facilities that enable normal or accelerated ageing of FCS stacks under constant and/or variable operating conditions, while controlling and gathering health monitoring data like power loads, temperatures, hydrogen and air stoichiometry rates, etc. The data sets provided for the challenge include both temporal and frequential data.

Regarding the PHM Challenge two kinds of experiments were considered.

  • A first stack (FC1) was operated in stationnary regime at roughly nominal conditions. This experiment served actually as a reference test.
  • A second stack (FC2) was operated under dynamic current testing conditions, i.e. with current ripples.

Experimental platform @ FCLAB


3. Challenge Data Sets and details (to download)

The following figures depicts an example of voltage drop (raw) gathered during one of the experiments, as well as a polarization curve and an Electrochemical Impedance Spectroscopy treatments. Participants were provided with this kind of data (completed with monitoring signals – temperature, pression, etc.).

  • Data from experiment on FC1 (Fuel Cell operated in stationary regime) were fully given. This set of data was refered as the “learning dataset”.
  • Data from experiment on FC2 (Fuel Cell operated under dynamic current) were partially given: only data until time t = 550 h were available. This set of data was refered as the “testing dataset”.

According to that, participants were expected to predict the behavior of FC2, in order to determine either the SoH of the stack, either the RUL of the stack.

  • More details on the challenge can be downloaded here.
  • Training data set can be downloaded here.
  • Testing data set can be downloaded here.


4. Actual estimates and top-scoring participants

The challenge was open to all potential conference attendees. Both Academic (from University) and Professional teams (from Industry) were encouraged to enter! The top scoring participants will be distinguished and invited to present at a special session of the 2014 IEEE International Conference on Prognostics and Health Management.

Actual estimates

  • The complete “testing dataset” (including hidden part) is now made publicly available.
  • Actual SoH estimates and RUL predictions can be dowloaded here.
  • More details can be found here.

SOH estimation – Top-scoring participants

  • Academic Winner (Top scoring participant)
    • Dep. of Mechan. & Aerospace Engin., Seoul National University, Korea
    • Score: 5.3177e^{-6}
  • Industrial Winner
    • EMBRAER, São José do Campos, Brazil
    • Score: 7.8932e^{-6}

RUL prediction – Top-scoring participants

  • Academic Winner (Top scoring participant)
    • Dep. of Mechatronic \& Dynamic, University of Paderborn, Germany
    • Score: 0.7760
  • Industrial Winner
    • IBM Almaden Research Center, San Jose, CA, USA
    • Score: 0.3592


5. Acknowledgment and local organization committee

Datasets are being made publicly available. Publications making use of these databases are requested to explicitly mention in an acknowledgment paragraph that data were provided by FCLAB Federation, FR CNRS 3539, France.

For any request please contact the local organizing committee (Rafael Gouriveau, Mickaël Hilairet, Daniel Hissel, Samir Jemeï, Marine Jouin, Elodie Lechartier, Simon Morando, Elodie Pahon, Marie-Cécile Péra, Noureddine Zerhouni) at:


FCLAB Research Federation, FR CNRS 3539 / FEMTO-ST
Labex ACTION program (contract ANR-11-LABX-01-01)
Rue Thierry Mieg, F-90010 Belfort – France ;

Powered by WordPress