A Spatio-Temporal Modeling Approach for Battery Pack Capacity Prognostics

Battery packs are often used to provide energy at required level of current and voltage, in which cells are wired in series and parallel within a battery pack. The dependency among cells is critical to the useful life of the pack. Such dependency is attributed to many factors such as battery pack configurations and the mismatch of internal resistance. This paper proposes a general spatio-temporal prognostics method to model and quantify the dependency among cells through a class of selected spatial covariance structure. The covariance function in the proposed spatio-temporal model is estimated based on the cells capacity. Since cell-level capacity measurements in a battery pack may not be directly available, a capacity estimator based on sample entropy of current and voltage is proposed. The capacity degradation behavior of a battery pack with two parallel connected cells is investigated using the proposed method. In addition, the cycle life at pack level is assessed through simulation results from the estimated spatio-temporal model.