Multidimensional estimation of inhomogeneous lithium-ion cell aging via modal differential voltage analysis

Due to the increasing demand for lithium-ion batteries in automotive applications, understanding their use and degradation is becoming more and more important. This paper presents a new method to quantify cell degradation's multidimensional aging and inhomogeneity. The proposed algorithm is inspired by data-driven modal reduction. Offline findings from measured differential voltage curves and electrochemical simulations are encrypted in low-rank patterns, suitable for on-board application. A dataset of eighteen automotive pouch cells with graphite anode and NMC-622 cathode that were cycled for accelerated aging is used to generate orthogonal aging modes. The cycling is performed with two different load profiles, yielding homogeneous and inhomogeneous cell aging. The cell capacity is estimated for eight test cells with a polynomial function, making a Mean-Absolute-Error (MAE) of 0.3% over their entire lifetime. The loss of active lithium and loss of active material at the anode and the cathode are fitted in polynomials depending on references from physical- chemical simulations. Furthermore, the inhomogeneity of cell aging is estimated, recognizing a pattern of modal coordinate evolution that links the reference differential voltage curves to post-mortem findings from the literature.