Fuel cell remaining useful life prediction and uncertainty quantification under an automotive profile

Abstract

Proton exchange membrane fuel cell is a clean and efficient energy converter that can be use to power an electrical vehicle efficiently. Nevertheless, degradation mechanisms affect the lifespan of this electrochemical converter. Consequently, the estimation of the State of Health and Remaining Useful Life have been the subject of numerous researches in the past years. However, most of the methods available in the literature dealing with fuel cell prognostic do not allow the uncertainty quantification of the estimation that can be implemented online due to the computational cost. As a novelty, this paper presents a prognostic algorithm based on an Extended Kalman Filter. This observer estimates the State of Health, the speed of the degradation and also provides the estimation uncertainty. Then, an Inverse First Order Reliability Method computes the Remaining Useful Life with a 90% confidence interval based on the estimation of the observer. This method is applied on a 175 hours dataset coming from a experimental test on a 8-cells fuel cell stack subjected to an automotive power profile.

Prognostic methods for proton exchange membrane fuel cell under automotive load cycling: a review

“…Bressel et al [39] presented the first work on PEMFC RUL estimation using BFA under an automotive data set of 175 h obtained in the laboratory. The authors propose an empirical degradation model based on the EKF algorithm to estimate the SoH and the RUL of an eight‐cells PEMFC stack of CEA Liten.…”

“…The degradation observed in realistic environmental conditions is different from those observed in the laboratory. As there is a lack of PEMFC ageing data under automobile load cycling, some authors validate their prognostic results assuming a linear degradation pattern. However, the decreasing tendency of the PEMFC voltage is not a monotonic function [39, 46]. As it is mentioned in the previous section about the needed of historical system information for data‐driven approach, some authors decide to use simulated data in order to implement and validate their methods. Data‐driven and hybrid methods show a high performance in terms of precision during the RUL prediction process as it is presented in [50].…”

Advanced passivity-based, aging-tolerant control for a fuel cell/super-capacitor hybrid system

“…In order to study the effect of aging on the parameter values of (1), a 175-hour continuous experimental test was performed on an 8-cell FC stack under an automotive load profile with a surface of 220 cm 2 , provided by the French Atomic Energy and Alternative Energies Commission (CEA) (see [35] for more details about the FC operating conditions). Using a fitting technique, a set of model parameters were extracted from these curves [33].…”

“…SoH α(t) is set equal to zero at the beginning (i.e., no degradation), and increases to α max = 0.3 at the considered End-of-Life (EoL) of the FC. A value of α max equal to 0.3 represents a reduction of around 10% of the nominal power of the FC at 0.5 A/cm 2 (see [33,34,35] for more details). , it is apparent that the controller needs to know the maximum current that the FC can deliver over time.…”

Proton Exchange Membrane Fuel Cell Prognosis Based on Frequency-Domain Kalman Filter

“…The degradation models can be divided into three categories, i.e., physical models, empirical models, and semiempirical models [10], [13]. Mathieu Bressel et al predicted the degradation and RUL of a PEMFC based on the Extended Kalman Filter (EKF) and a new empirical model, and the uncertainty of the prediction was quantified [12], [14]. At the same time, a particle filter and a voltage degradation model were applied by Marine Jouin et al, and the degradation and RUL were predicted [15].…”