Logo
RECHERCHER

Online characteristics estimation of a fuel cell stack through covariance intersection data fusion

Téléchargements

Téléchargements par mois depuis la dernière année

Plus de statistiques...

Daeichian, A. et Ghaderi, R. et Kandidayeni, M. et Soleymani, M. et Trovão, J. P. et Boulon, L. (2021). Online characteristics estimation of a fuel cell stack through covariance intersection data fusion. Applied Energy, 292 . p. 116907. ISSN 0306-2619 DOI 10.1016/j.apenergy.2021.116907

[thumbnail of BOULON_L_213_POST.pdf] PDF
Sous embargo jusqu'au 16 juin 2023.

Télécharger (1MB)

Résumé

Employing semi-empirical models to estimate some characteristics of a fuel cell (FC) stack, such as power and polarization curves, is demanded for efficient design of a power allocation strategy in a FC hybrid electric vehicle. However, the multivariate nature of a FC system has made the design of an accurate model challenging. Since each semi-empirical model has its own pros and cons, this paper puts forward a data fusion approach for online characteristics estimation of a FC stack utilizing four well-known models, namely Mann, Squadrito, Amphlett, and Srinivasan. Despite the other similar techniques, the suggested one utilizes the strengths of each mentioned FC model while avoiding their drawbacks. Kalman filter is employed to identify the parameters of the models online to embrace the uncertainties caused by the alteration of operating conditions and degradation level. Considering the parameters, the output voltage given by each model as well as their covariance are computed. Then, a covariance intersection algorithm is proposed to fuse the estimated output voltages. The fusion of the models’ outputs leads to the estimation of fused characteristics curves. To underline the effectiveness of the proposed method, it is applied to four different experimental datasets extracted from three 500-W Horizon FCs. The obtained results demonstrate the superior performance of the suggested estimator in the sense of mean square error. On average, the mean square error of the data fusion method is 39.64% and 36.59% lower than other studied methods while estimating the polarization curve and power curve, respectively.

Type de document: Article
Mots-clés libres: data fusion energy management strategy modeling online identification proton exchange membrane fuel cell
Date de dépôt: 24 août 2021 18:23
Dernière modification: 24 août 2021 18:23
Version du document déposé: Post-print (version corrigée et acceptée)
URI: http://depot-e.uqtr.ca/id/eprint/9712

Actions (Identification requise)

Dernière vérification avant le dépôt Dernière vérification avant le dépôt