Model based condition monitoring in lithium-ion batteries

Abstract

In this paper, a model based condition monitoring technique is developed for lithium-ion battery condition monitoring. Here a number of lithium-ion batteries are cycled using two separate over discharge test regimes and the resulting shift in battery parameters is recorded. The battery models are constructed using the equivalent circuit methodology. The condition monitoring setup consists of a model bank representing the different degree of parameter shift due to overdischarge in the lithium ion battery. Extended Kalman filters (EKF) are used to maintain increased robustness of the condition monitoring setup while estimating the terminal voltage of the battery cell. The information carrying residuals are generated and evaluation process is carried out in real-time using multiple model adaptive estimation (MMAE) methodology. The condition evaluation function is used to generate probabilities that indicate the presence of a particular operational condition. Using the test data, it is shown that the performance shift in lithium ion batteries due to over discharge can be accurately detected.