Adaptive filtered high-gain observer for PEMFC systems in electric vehicles

This paper addresses the problem of estimating oxygen excess in proton exchange membrane fuel cells (PEMFC) systems. Indeed, poor management of the air supply system usually leads to varying degrees of oxygen starvation, which affects the durability and reliability of the PEMFC. Thus, real-time monitoring of cathode internal states is crucial for enhancing air supply system performance and fuel cell net power. This study estimates cell oxygen and nitrogen partial pressures using measurable variables, such as compressor speed and supply manifold pressure. To this end, we designed filtered high-gain observer that provides online estimates of variables of interest based on online measurements of compressor speed and supply manifold pressure. The filtered nature of the observer makes it less sensitive to measurement noise. Moreover, an extremum-seeking-based optimizer is developed to tune the observer gain in real time, aiming to meet robustness requirements against measurement noise. The stability of the observer is performed using Lyapunov stability tools. Unlike the standard high-gain observer, the new observer incorporates an additional filtering function to reduce its sensitivity to output measurement noise. In contrast to existing filtered high-gain observers, the new observer features real-time self-tuning of its parameters. The superiority of the new observer over both a standard high-gain observer and a filtered high-gain observer with fixed parameters is underscored by simulation results.