Predictive Maintenance of transportation systems under incomplete and imprecise data – MAPSYD
Predictive Maintenance of transportation systems under incomplete and imprecise data
Maintenance is an important factor in quality, dependability, and productivity in complex industrial installations. However, the development of a predictive maintenance system that is truly efficient in relation to the objectives of maintaining operational conditions currently requires defining the scientific foundations of predictive maintenance as well as the appropriate methods and technologies.
Scientific objectives and hypotheses
The original contributions of the MAPSYD project can be broken down according to the themes below:<br /><br />- Dependability : The first innovative element is the integration of the dependability in the process of predictive identification of system failures.<br /><br />- Predictive modeling in the presence of uncertainties: The second innovative element is formulated by the predictive model necessary to transform the data collected into a reliable instruction modifying the organization of maintenance actions.<br /><br />- Economic decision support: The third innovative element of the project is the integration of an economic decision support tool to complete the purely technical analysis of maintenance needs.
- Hidden Markov models and imprecise probabilities for the predictive part of the future state of systems.
- Prospect theory for the economic maintenance part
- Development of a methodology and a Toolbox under Matlab for the PHM (prognosis) by Hidden Markov methods (HMM).
- The validation of the Toolbox was carried out through tests on data from international challenges.
Application of our solution in other fields: health, military field, ...
R. Louhichi, M. Sallak, J. Pelletan, A Maintenance Cost Optimization Approach: Application on a Mechanical Bearing System, International Journal of Mechanical Engineering and Robotics Research, DOI 10.18178/ijmerr.9.5.658-664, 2020.
• R. Louhichi, M. Sallak, J. Pelletan. « A review on predictive maintenance purposes in terms of risk minimization, 30th European Safety and Reliability Conference and 15th Probabilistic Safety Assessment and Management Conference, Venezia, Italy, 2020.
• R. Louhichi, M. Sallak, J. Pelletan. « A cost model for predictive maintenance based on risk-assessment », 13ème Conférence Internationale CIGI QUALITA, Montreal, Canada, 2019.
• R. Louhichi, M. Sallak, J. Pelletan, « Elaboration of an economic model for decision aid optimizing the maintenance strategy of transport system », 6éme Journées Régionales des Doctorants en Automatique (JRDA), 2019, Lille, France, 2019.
• A. Delmas, M. Sallak, W. Schön, L. Zhao. « Méthodes de prédiction de durée de vie en vue de modèles de maintenance prévisionnelle : calcul d'intervalles et stratégies en présence de données incertaines », Congrès International de maîtrise des risques et de sûreté de fonctionnement « Lambda Mu 21 », Reims, France, 2018
• A. Delmas, M. Sallak, W. Schön, L. Zhao, « Remaining useful life estimation methods for predictive maintenance models: defining intervals and strategies for incomplete data », 10th IMA International Conference on Modelling in Industrial Maintenance and Reliability, Liverpool, United Kingdom, 2018.
• R. Louhichi, M. Sallak. « Maintenance prévisionnelle des systèmes de transport en présence de données incomplètes et imprécises (MAPSYD) », 5ème Journée Régionale des Doctorants en Automatique (JRDA), Amiens, France, 2018.
Maintenance is an important part of the life-cycle of embedded systems and considered as an important factor in quality, dependability, and productivity of complex industrial systems. The MAPSYD project proposes an original methodology for predictive maintenance when considering both incomplete and imprecise data collected by the sensors. First, we propose a methodology based on the combination of Hidden Markov Chains (MMC) and imprecise probability theory. Then, we will define a policy decision which aims to build an economic model based on the maintenance policy and the imprecise MMC model. Finally, we propose to develop a sensor system embedded that implements the algorithms developed in the proposed methodology in a tram or a bus, and located on critical components.
Project coordination
Mohamed Sallak (Heudiasyc UMR CNRS 7253)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partner
ILB INSTITUT LOUIS BACHELIER
SYNOX
SECTOR STE ETUDE CONSEIL TECHN ORG
Heudiasyc Heudiasyc UMR CNRS 7253
Help of the ANR 872,120 euros
Beginning and duration of the scientific project:
- 36 Months
