Insulation Fault Diagnosis of Battery Pack Based on Adaptive Filtering Algorithm

Insulation is the foundation for the safe operation of battery systems. However, the working condition of the battery system is complex, which challenges insulation fault detection. This article presents an online estimation algorithm of insulation resistance based on an adaptive filtering algorithm for a battery energy storage system (BESS). Specifically, the insulation detection model is developed based on the Thevenin model. Aiming at the problem of system noise, a joint estimation algorithm for battery parameters and voltage is proposed based on the recursive least square and unscented Kalman filter (RLS-UKF) algorithm. The full climate models of capacity and temperature are developed. Furthermore, an insulation resistance estimation algorithm is proposed based on the UKF algorithm. The proposed method is verified by different dynamic experiments. Experimental results show that the RLS-UKF algorithm has a better voltage filtering effect than the UKF algorithm with fixed model parameters. The proposed insulation resistance algorithm can accurately estimate the system’s insulation resistance under dynamic and static conditions.