Chattering free sliding mode observer estimation of liquid water fraction in proton exchange membrane fuel cells

Abstract

The aim of this work is to deploy and experimentally validate a model-based strategy to estimate unmeasurable variables in a proton exchange membrane (PEM) fuel cell. First, a nonlinear PEM fuel cell dynamical model is implemented and calibrated using an optimisation approach that takes real measurement data as input. Then, an advanced observation approach is developed to retrieve non-measurable data from the fuel cell. Two states are estimated in this work: the fuel cell temperature and the internal liquid water fraction. To achieve this, a model-based high-order sliding mode observer (HOSM) with chattering-free capabilities is deployed. The fuel cell temperature is measured in real time to drive the estimation error to zero in a finite amount of time. Finally, the methodology is validated using experimental data stemming from a laboratory test station, comparing the the HOSM observer with and without chattering-free gain.