Battery State of Charge Estimation using An Adaptive Unscented kalman Filter for Photovoltaics Applications.

Abstract

Battery management system (BMS) is an electronic device responsible for all control and management operations of several battery parameters, especially SOH and SOC. The battery SOC acts like an indicator of the internal charge level of the battery, in order to avoid unpredicted system interruption and prevent the batteries from being over-charged or over-discharged. SOC estimation procedure is one of the most complex techniques caused by complex battery chemistry and its strong non linearity. In this paper we have chosen a Kalman filtering algorithm to estimate internal states of Lithium Ion battery and dynamically estimate the SOC by decreasing divergence due to parametric uncertainty of battery model, measurement and process noise by using an Unscented variant of this filter. To further enhance the UKF algorithm, an adaptive calculation of noise covariance is proposed to combine between a better convergence and robust results.  Experimental results indicate that the adaptive unscented Kalman filter based algorithm has better performance Battery State of Charge estimation. A comparison with other estimations techniques shows that the proposed SOC estimation method is the best choice in term of accuracy and robustness.