This work presents a method to quantify and estimate the degradation level of lithium-ion battery cell cycle aged at different power levels. Experimental results previously reported by the authors are analysed further, now with artificial intelligence workflows to establish a method to estimate the cells degradation rate, in term of State of Health. The neural network estimation capability is dependent on the type of input signal used for training, and the relative proportion of training vs. testing data. The development of a feedforward neural network which elaborates the information of voltage and current differences during sudden power changes significantly increased the predictive capability of the method, reaching State-of-Health estimation best-case errors lower than 1 %, in line with more complex Artificial Intelligence approaches found in literature. In addition, the results obtained with the feedforward neural network are then compared with a regression learning – based estimation function, trained and tested over the same dataset. As last test, both these two methods, trained with dataset coming from cycle aging experimental tests, are used to estimate the State of Health of a lithium cell aged due to calendar phenomena only.

Neural network for the estimation of LFP battery SOH cycled at different power levels

Scarpelli C.;Huria T.;Lutzemberger G.;Ceraolo M.
2023-01-01

Abstract

This work presents a method to quantify and estimate the degradation level of lithium-ion battery cell cycle aged at different power levels. Experimental results previously reported by the authors are analysed further, now with artificial intelligence workflows to establish a method to estimate the cells degradation rate, in term of State of Health. The neural network estimation capability is dependent on the type of input signal used for training, and the relative proportion of training vs. testing data. The development of a feedforward neural network which elaborates the information of voltage and current differences during sudden power changes significantly increased the predictive capability of the method, reaching State-of-Health estimation best-case errors lower than 1 %, in line with more complex Artificial Intelligence approaches found in literature. In addition, the results obtained with the feedforward neural network are then compared with a regression learning – based estimation function, trained and tested over the same dataset. As last test, both these two methods, trained with dataset coming from cycle aging experimental tests, are used to estimate the State of Health of a lithium cell aged due to calendar phenomena only.
2023
Scarpelli, C.; Gazzarri, J.; Huria, T.; Lutzemberger, G.; Ceraolo, M.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11568/1174745
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 3
  • ???jsp.display-item.citation.isi??? ND
social impact