Battery State of Charge Estimation in Automotive Applications using LPV Techniques

One of the most difficult problems in battery pack management aboard a P/H/EV is the estimation of the state of charge (SoC). Many proposed solutions to this problem have appeared in the literature; in particular, model-based extended Kalman filter approaches have shown great promise. However, the computational burden of implementing an extended Kalman filter is significant. Moreover, some parameters needed to make the extended Kalman filter function correctly are difficult to estimate from measured data. This paper proposes an SoC estimator design using linear parameter varying (LPV) system techniques that provides a low computational alternative to the extended Kalman filter. The stability of this estimator can be verified analytically. The performance of the estimator in terms of convergence and tracking is verified experimentally on an isothermal dataset taken from a lithium ion battery cell.