Online SOC estimation of Li-FePO4 batteries through an observer of the system state with minimal nonspecificity

Abstract

An observer for nonlinear dynamical systems is presented. Both the uncertainty about the system state and the measurement noise are modelled by means of possibility distributions. The stability of the model is appraised through the nonspecificity of the posterior distribution of the state. The methodology is applied to build a fast estimator of the State of Charge of a LiFePO4 battery, and compared to stochastic alternatives as the Kalman filter on data obtained at the Battery Laboratory at Oviedo University. The new method improves linear filters in both speed and stability.