Two Layer Markov Model for Prediction of Future Load and End of Discharge Time of Batteries

To predict the remaining discharge energy of a battery, it is significant to have an accurate prediction of its end of discharge time (EoDT). In recent studies, the EoDT is predicted by assuming that the battery load profile (current or power) is a priori known. However, in real-world applications future load on a battery is typically unknown with high dynamics and transients. Therefore, predicting battery EoDT in an online manner can be very challenging. The purpose of this paper is to derive a load prediction method for capturing historical charge/discharge behaviour of a battery to generate the most probable future usage of it, enabling an accurate EoDT prediction. This method is based on a two layer Markov model for the load extrapolation and iterative model-based estimation. To develop the proposed concept, lithium-ion batteries are selected and the numerical simulation results show an improvement in terms of the accuracy of the EoDT prediction compared to methods presented in the literature.