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

  • Virtual Sensing & Intelligent Engineering
  • Energy & Multiphysics Systems

Back to selected work