Li-Ion Battery Second-Life State of Health (SoH) Estimation with Polynomial Chaos

Traction batteries, such as those used in electric vehicles, demand precise energy and power specifications. Operating under extreme temperatures, experiencing a high number of partial cycles, and facing fluctuating discharge rates contribute to notable battery degradation in the initial years of use. Nevertheless, a battery that has reached the end-of-life status in the traction application can find a “second life” in less demanding scenarios, such as stationary applications in a smart grid or low-power electric vehicles like golf carts. Second-life applications typically lack access to extensive measurements and monitoring infrastructure, although some level of monitoring remains essential to prevent potentially catastrophic failures. This work presents an approach to monitor the performance of second-life batteries by estimating the State of Health using measurements obtained during the first life and solely considering terminal voltage during the second life. The approach employs a polynomial chaos series to learn the relationship between the extracted features and the battery's discharge capacity in its initial life. In the subsequent life phase, the trained model is applied to estimate the SoH.