H2DECODE: Cross-Domain Degradation Intelligence for Hydrogen Fuel-Cell Vehicles

H2DECODE is developing a vehicle-level Degradation Intelligence System that combines a physics-informed Digital Twin, a Cross-Domain Degradation Attribution Engine and Counterfactual Decision Intelligence.

The system brings together four interacting domains: vehicle and fuel-cell state, driving and operating conditions, environmental conditions, and hydrogen exposure. Rather than producing only a health score or degradation prediction, H2DECODE reconstructs the degradation trajectory and quantifies the contribution of different degradation drivers and their interactions, together with confidence and uncertainty.

Its counterfactual layer then goes one step further: What would happen if the vehicle had been operated differently? Which conditions should be avoided? Which intervention could slow future degradation?

The result is an explainable decision-support technology designed to help turn fuel-cell degradation from an observed outcome into an actionable engineering variable.

H2DECODE builds on SMITH₂'s existing capabilities in hydrogen-system digital modelling, virtual sensing, AI/ML, optimisation and intelligent control. The project is designed to mature the technology toward a TRL 8 integrated and qualified system, preparing the pathway for subsequent operational deployment.

What H2DECODE offers?

• Physics-informed Digital Twin: reconstruction of relevant internal fuel-cell and vehicle states from measured and estimated operational data.

• Cross-Domain Degradation Attribution: identification and quantification of degradation contributions from vehicle state, operation, environment and hydrogen exposure, including interaction effects.

• Confidence-scored intelligence: explicit treatment of uncertainty, data availability and attribution confidence rather than opaque AI outputs.

• Counterfactual Decision Intelligence: evaluation of alternative operating conditions and “what-if” scenarios to identify opportunities to reduce degradation.

• Deployment-oriented architecture: designed for future integration with OEM engineering environments, fuel-cell systems, vehicle-data platforms and fleet-management solutions.

What we're looking for?

We would like to connect with hydrogen fleet operators, vehicle OEMs, fuel-cell system developers and technology partners interested in fuel-cell durability, degradation diagnostics and lifecycle management.

We are particularly interested in partners able to contribute representative or real fuel-cell vehicle operational data, support validation under operationally relevant conditions, or explore a future pilot deployment of H2DECODE.

We are also open to collaboration with Digital Twin, vehicle-data and fleet-management providers interested in integrating degradation intelligence into their existing platforms.

Our goal is simple: not only to detect that a fuel cell is degrading, but to understand why, and determine what can be done about it.