Parameter Estimation of Second-Life Lithium-Ion Batteries Through Subspace Identification Methods

The batteries of electric vehicles (EVs) are designed to operate until the moment that their capacity reaches 80% of the rated value defined by the manufacturer. After this point, according to standards, the battery should be removed from the EV. Nevertheless, these retired batteries still have a significant amount of capacity, and then, depending on their internal conditions, they can be used in applications with lower power requirements, this concept is called second-life. However, due to the high variability of the internal conditions of lithium-ion cells at second-life conditions, identifying their internal parameters is a challenge, and conventional approaches do not provide satisfactory results. Therefore, this work proposes a methodology to identify the dynamic model of lithium-ion cells in second-life conditions through subspace identification (SSI). Then, parameters related to the state of health (SoH) are derived from the identified models. The obtained results show that: (a) a good fit between the actual system and the identified model was achieved, where a root mean square error (RMSE) of 4.80 mV and a variance accounting for (VAF) of 99.1% were obtained; (b) it was possible to identify the system parameters with a relatively small error, and the estimated parameters can be used to predict the SoH of the given lithium-ion cell; and (c) due to the characteristics of the EVs drive cycles, the parameters of the lithium-ion cell can be identified during the operation of the EV, discarding the need for additional tests after the retirement of the LiB.