Partner 1 (Coordinator): Laboratoire Jean Kuntzmann (LJK)

The LJK laboratory is an Applied Mathematics and Computer Science laboratory created in January 2007 in Grenoble, France. A joint research unit of the Joseph Fourier University (UJF), the Grenoble Institute of Technology (Grenoble INP), the Pierre Mendes France University (UPMF), the CNRS and INRIA, it combines the forces of applied mathematicians and statisticians with graphics and computer vision. Its expertise centers on the computational and statistical sciences and their uses in analyzing natural phenomena, with applications ranging from environmental modelling through life sciences, nanosciences, visualization and signal processing to mathematical finance.

The laboratory is structured into three scientific departments, each containing a number of research teams. Geometry and Images focuses on geometric modelling and image processing including graphics and computer vision. Deterministic Models and Algorithms develops tools for numerical and symbolic computation. Statistics gathers theoretical and applied statisticians and specialists in data analysis and signal processing.

Three participants of the project (O. Gaudoin, F. Corset, L. Doyen) belong to the FIGAL team of the Statistics department. The last one (S. Despréaux) is member of the scientific computing division of the laboratory. The LJK has a long standing activity in stochastic modelling and statistical inference for reliability, with regular industrial cooperations, among which partner 5 (EDF).

Web site

Partner 2: Laboratoire de Mathématiques et de leurs Applications (LMAP)

The members of partner 2 belong to the LMAP "Probabilités et Statistique" (P&S) team. The LMAP is a laboratory affiliated to CNRS and associated to INRIA, with 64 permanent members and research topics mainly oriented towards applied mathematics. Safety and dependability is one of the two applied research themes of the P&S team. The main P&S team contributors to this topic are involved in this project. The partner members published 28 peer-reviewed articles in international journals since 2006.

Stochastic modelling, applied probabilistic study, theoretical and numerical computation of reliability indicators and mathematical studies of (semi-/non-) parametric inference methods including covariates and missing data problem are core competences of partner 2. This partner has ever experienced collaborations:

  • with partner 4 (ICD) through several joint works and the joint supervision of a PhD thesis on maintenance policy for deteriorating systems.
  • with partner 3 (LMB) through joint works on missing data information on competing risk models and first-passage time for Lévy type processes.
  • with partners 5 (EDF) and 6 (SNCF) on special applications through research contracts and a CIFRE PhD thesis.

Web site

Partner 3: Laboratoire de Mathématiques de Besançon (LMB)

The Laboratory of Mathematics of Besançon includes 53 researchers spread out in 5 teams of different fields of research: Number Theory, Functional Analysis, Partial Differential Equations, Numerical Analysis and Probability and Statistics. The two members of partner 3 involved in this ANR project lie in this latter team. The subjects of research developed by the Probability and Statistics team are, among others: mathematical finance, insurance, stochastic differential equations, empirical processes, nonparametric statistics, survival analysis or reliability based on counting processes and martingales, sparse regression and selection of variables methods.

There are already some links between this partner and some other partners of this project. In addition to the works with partner 2 cited above, a work with partner 4 has been done on the construction of test statistics for some reliability models.

Web site

Partner 4: Institut Charles Delaunay - Troyes (ICD)

The Charles Delaunay Institute (ICD), created in January 2006, is a grouping of all the research teams of the Troyes University of Technology (UTT). It involves 100 permanent members and 152 PhD in 8 project teams mainly in Information and Communication Sciences and Technologies, and in Engineering. The half of the research activities in the ICD is dedicated to the common pluridisciplinary topic "Sciences and Technologies for Risk Management" (Sciences et Technologies pour la Maîtrise des Risques, STMR) structured as a CNRS "Unité Mixte de Recherche" (UMR 6279) from January 2010.

The participants of the project belong to the Systems Modelling and Dependability project team (LM2S) and to the STMR CNRS UMR 6279. The LM2S develops research in systems safety and dependability, following probabilistic and statistical approaches. Two main axes structure the research developments: systems monitoring (probabilistic approaches of monitoring and diagnosis; machine learning and pattern recognition; distributed monitoring and sensors networks) and reliability and maintenance (deterioration stochastic modelling; maintenance policies performance assessment and optimization; methods for reliability and probabilistic models). Both theoretical and methodological researches, and strong collaborative researches with industry are investigated. Within this project, the LM2S intends to contribute with its skills and to further develop a modelling framework for "dynamic" decision making in reliability and maintenance with a special emphasis on applications in the fields of energy and transport. Collaborations already exist between ICD and partners 2 (LMAP), 3 (LMB) and 5 (EDF).

Web site

Partner 5: Electricité de France (EDF)

The EDF Group is a leading player in the European energy industry, active in all areas of the electricity value chain, from generation to trading, and increasingly active in the gas chain in Europe. The vocation of EDF R&D is to contribute to improving performance among EDF Group operating units and to identify and prepare new growth drivers for the medium and long terms. EDF R&D has a committed policy of working with partners in France and Europe. A growing share of research is being conducted via partnerships. It is a way of drawing on outside talent and ideas, stepping up research and sharing costs and risks.

The "Industrial Risk Management" Department of EDF R&D devotes 80 engineers developing methods and tools to improve risk management, considering safety, performance (availability, costs...) and lifetime objectives. The study object is a socio-technical system subjected to risks, operated within the EDF Group, as nuclear, hydro or fossil fired power plants, including the following dimensions: component, technical system, organizational and human factors and environment.

Within the "Industrial Risk Management" department, the "Component Reliability and Uncertainty Modelling" Group develops innovative probabilistic and statistical methods and tools to carry out reliability studies in order to support the operation division strategy (operation and maintenance optimization, argument towards the regulators...). It is involved in many national working groups, such as the National Risk Management Institute (IMdR) or the French Statistical Society (SFdS), and its works are presented in the main international conferences related to reliability (lambda-mu, MMR, ESREL, ARS...). It has yet collaborated with partners 1 (LJK), 2 (LMPA) and 4 (ICD).

Web site

Partner 6: Société Nationale des Chemins de Fer Français (SNCF)

SNCF INFRA, part of the historical French National Railway Operator SNCF Group, is responsible for maintenance of the French railway network 24 hours a day, 7 days a week, from the high-tech systems of the high-speed lines to the oldest equipment.

SNCF INFRA manages the French railway network on behalf of RFF, who sets requirements: guarantee the reliability of facilities at the best price, increase the commercial availability of the infrastructure and collect data on the network at the same time. This is a role which requires a wide range of skills and equipment (operations involve track, signals, overhead lines, telecommunications, overground and underground structures, etc.)

SNCF INFRA optimises maintenance policies for individual lines. On lines at maximum capacity, particularly in suburban areas, availability is the key priority and infrastructure must be replaced before problems arise. However, on small lines, maintenance teams schedule replacements over time. The industrialisation of maintenance work also makes a difference. Operations are planned two years in advance on a national scale, and working on large sections at the same time frees up more time slots for trains.

Each monitoring and maintenance operation improves our understanding of the national railway infrastructure. Data is entered into IT systems which give an overall view of the condition of the network.

Web site