Panel sessions

Moderators: Mayank Jha (U.Lorraine) and Olga Fink (EPFL)

• Christophe Bérenguer, Université Grenoble-Alps (GIPSA)
• Didier Theillol, Université de Lorraine
• Zhiguo Zeng, Centrale-Supélec (Saclay)
• Peter Fogh Odgard, Goldwin Energy

Abstract:

Nearly all mission- and safety-critical systems—energy and water infrastructure, process plants, transportation and autonomous vehicles, aerospace platforms—operate in closed loop under uncertainty and progressive wear. Controllers must remain resilient under abrupt faults and gradual degradations, often with incomplete physics and imperfect degradation models.

In this context, Health-Aware Control (HAC) is an emerging paradigm that explicitly or implicitly embeds prognostic information—such as state of health (SoH), reliability estimates, and remaining useful life (RUL)—into feedback control synthesis and reconfiguration.

Unlike traditional Prognostics and Health Management (PHM), which typically reasons in open loop, HAC reasons in closed loop: degradation dynamics are modeled (or learned) as part of the plant, and control laws are designed to optimize not only short-horizon performance but also long-horizon availability, safety, and asset lifespan

• Map the state-of-the-art vs. state-of-practice in HAC across sectors.
• Clarify interfaces among diagnosis, prognosis (RUL),and closed-loop control.
• Contrast HAC with Fault-Tolerant Control (FTC) and define complementarity with FDI.
• Examine roles of AI/ML (incl. Safe RL), digital twins, and physics-informed models.

Moderator: Nenad Mijatovic (Alstom)

• Jonathan Sprauel (Thales-Alenia Space)
• Ayoub Drissi (SNCF -Réseau)
• Diego Galar (Lulea University of Technology)

Abstract: Prognostics and Health Management (PHM) is experiencing a transformative shift toward more reliable, explainable, and trustworthy data-driven approaches, such as Data Science and Artificial Intelligence (AI). This panel brings together industry leaders from transportation, railway, healthcare, finance, and other critical sectors, as well as academic and research institutions, to explore cutting-edge methodologies that ensure technology and operational excellence in prognostics and health management systems.

• Building and Establishing Trust in AI-Driven Maintenance Decisions Through Reliability and Explainability
• Real-World Implementation Challenges
• EU AI Act Risk Assessment and Classification
• Future of PHM using AI

Moderator: Dave Larsen (Collins)

Abstract: To be added soon.

Moderator: Knut Knutsen, DNV

• Vilmar Æsøy, NTNU
• Børre Pedersen, DNV

Abstract: The maritime sector faces increasing complexity with the rise of automation, digitalization, and sustainability demands. Prognostics and Health Management (PHM) offers a transformative approach to enhance safety by enabling predictive insights into equipment health and operational risks. Unlike traditional reactive or time-based maintenance, PHM integrates condition monitoring, fault diagnosis, and remaining useful life prediction to prevent failures before they occur. This is critical in maritime operations where system downtime or catastrophic failures can lead to severe safety, environmental, and economic consequences. While traditional safety frameworks emphasize preventing consequences of failures through design redundancy, fail-safes, and proven components, PHM focus on predicting failures during operation. The panel will explore how PHM can be applied across ship systems to mitigate hazards and support compliance with safety standards. Panelists from industry and academia will discuss implementation challenges, data-driven strategies, and the role of PHM in enabling autonomous and conventional vessels to operate safely under demanding condition

• What barriers do you see to uptake of PHM in maritime?
• How can PHM model failures including uncertainty to improve safety beyond what statistical failure-rate approaches can capture?
• Have you seen success stories where PHM or condition based maintenance has been well implemented in maritime?
• Have you seen cases that were not successful and can you provide learnings from this experience?