Nonlinear observation in fuel cell systems: A comparison between disturbance estimation and High-Order Sliding-Mode techniques

Abstract

This paper compares two Nonlinear Distributed Parameter Observers (NDPO) for the observation of a Proton Exchange Membrane Fuel Cell (PEMFC). Both NDPOs are based on the discretisation of distributed parameters models and they are used to estimate the state profile of gas concentrations in the anode and cathode gas channels of the PEMFC, giving detailed information about the internal conditions of the system. The reaction and water transport flow rates from the membrane to the channels are uncertainties of the observation problem and they are estimated throughout all the length of the PEMFC without the use of additional sensors. The first observation approach is a Nonlinear Disturbance Observer (NDOB) for the estimation of the disturbances in the NDPO. In the second approach, a novel implementation of a High-Order Sliding-Mode (HOSM) observer is developed to estimate the true value of the states as well as the reaction terms. The proposed observers are tested and compared through a simulation example at different operating points and their performance and robustness is analysed over a given case study, the New European Driving Cycle.

Nonlinear predictive control for durability enhancement and efficiency improvement in a fuel cell power system

“…(9) and (12) are discretised using a forward-backward discretisation procedure [10]. While the discretisation of the partial differential equations allows to make use of lumped systems theory which takes advantage of the known boundary conditions, it introduces a high computational effort.…”

“…For this reason, a state observer is implemented using the measurement of some PEMFC outputs. In this paper a nonlinear distributed parameters observer (NDPO) [9,10] is implemented. The NDPO approach will allow to consider the internal distribution of the reactants concentrations and the water content when designing the control strategy in Section 4.…”

“…The discretisation of Eq. (9) over n finite volumes [9,10] along the gas channels direction for each one of the i-th gas species is: …”

“…Eq. (22) has to be expressed in block controllable structure in order to apply the observation strategy [10,36] proposed in Section 3.3:…”

“…However, while some of the internal variables can be measured using the existing sensor technology, there are parts of the system that are inaccessible. Henceforth, dynamic fuel cell modelling [1,2,3,4], fuel cell model-based control [5,6,7] and model-based observation and identification of parameters [8,9,10,11] are compelling research topics in the field.…”