Model-Based State-of-Charge Estimation of 28 V LiFePO ₄ Aircraft Battery

Abstract

<div>This article introduces an advanced state-of-charge (SOC) estimation method customized for 28 V LiFePO₄ (LFP) helicopter batteries. The battery usage profile is characterized by four consecutive current pulses, each corresponding to distinct operational phases on the helicopter: instrument check, key-on, recharge, and emergency power output stages. To establish a precise battery model for LFP cells, the parameters of a second-order equivalent-circuit model are identified as a function of C-rate, SOC, and temperature. Furthermore, the observability of the battery model is assessed using extended Lie derivatives. The signal-to-noise ratio (SNR) of the open-circuit voltage (OCV)–SOC relation is analyzed and employed to evaluate the estimator’s resilience against OCV flatness. The extended Kalman filter (EKF) and the unscented Kalman filter (UKF) are utilized for SOC estimation. The results emphasize the significance of meticulously choosing process and sensor noise covariance matrices to achieve a resilient SOC estimator for LFP cells. Furthermore, the UKF demonstrates superior robustness against OCV–SOC relationships compared to the EKF. Lastly, the UKF is selected for testing across various aircraft usage scenarios at 10°C, 25°C, and 45°C. The resultant root mean square errors for SOC estimation at these different temperatures are consistently below 2%, thereby validating the effectiveness of the UKF SOC estimation approach.</div>