Partial least squares model method for state of health prediction of lithium-ion batteries

The practical operation of battery packs in electric vehicles requires a continuous and accurate estimate of the state of health (SOH) either at cell or pack levels. On the other hand the stochastic operation in real applications of any ì battery pack accelerates and diversifies the aging process of each specific battery thus making hard a precise estimate of SOH and the prediction of the remaining useful life (RUL). In this study, an effective estimation method based on machine learning is proposed to achieve eliable SOH. Here, charge-discharge potential curves of Li-ion pouch cells from an on-line freely available dataset, were used as inputs for a linear regression model to predict SOH. Our experimental findings demonstrate that the suggested computational technique can accurately, steadily, and robustly estimate the battery SOH with an error that is smaller or comparable to other modelling approach based on multi-model-based algorithms.