Virtual sensing · Hydrogen systems
Fuel-Cell Virtual Sensor Platform
Development of physics-informed and data-assisted estimators for hydrogen mixture properties, density, mass flow and recirculation states in fuel-cell subsystems.
01 / Problem
Problem
Important subsystem states can be difficult or impractical to measure directly across the complete operating envelope.
02 / Context
Why it matters
Reliable state estimates can strengthen observation, calibration, diagnostics, and control development when physical instrumentation is constrained.
03 / Boundary
System boundary
A public-safe platform view spanning physical relationships, available signals, estimation methods, validation logic, and deployment constraints.
04 / Contribution
Contribution
Initiated the technical direction, secured internal innovation funding, formed an engineering team, and guided physics-informed and data-assisted estimator development.
05 / Method
Methods and tools
- Virtual sensors
- State estimation
- Physics-informed models
- MATLAB ML/DL workflows
- Validation design
06 / Decisions
Architecture decisions
- Use physical relationships to constrain learning and improve interpretability.
- Design algorithms around available signals and deployment realities.
- Keep estimator evaluation separate from confidential system calibration.
07 / Insight
Outcome or insight
A deployment-oriented platform direction and reusable engineering workflow, described without internal funding amounts, customer details, or proprietary performance claims.
08 / Confidentiality
Confidentiality note
This case study is intentionally limited to public-safe methods and architectural patterns. It excludes customer names, proprietary diagrams, internal identifiers, supplier information, funding amounts, and confidential performance data.
Related capabilities