Concentration Estimation for Fuel Cells: Design of Experiments, Nonlinear Identification, and Observer Design With Experimental Validation

Abstract

Fuel cells (FCs) are promising eco-friendly power sources. Nevertheless, there are challenges to overcome if they are to be widely deployed in areas such as degradation avoidance and control, where the knowledge of the unavailable concentrations is crucial. In this respect, observers can provide unavailable quantities based on an estimation algorithm and available measurements. This paper presents an FC concentration observer design workflow, covering the model-based design of experiments (DOE), their execution, systematic nonlinear identification, and measurement-based validation. The model-based DOE and the validation with a mass spectrometer, including dynamic operation, are unique for PEMFC observers. The workflow is demonstrated with a constrained extended Kalman filter observer on a $\mathrm {30 \text {k} \text {W} }$ polymer electrolyte membrane FC (PEMFC) test stand. A control-oriented model serves as the workflow basis, and the DOE is based on optimizing the parameter sensitivity. The test stand delivers the measurements, the parametrization comprises a sensitivity analysis, and the experimentally validated observer yields outstanding concentration estimation performance.

Keywords

• Design of experiments
• design workflow
• experimental validation
• fuel cells
• Kalman filter
• mass spectrometer