基于实车运行数据的锂离子电池健康状态估计

Accurately estimating the state of health of lithium-ion batteries is significant for the safety management of electric vehicles. Aiming at the problems of incomplete battery states, complex operating conditions, and poor data quality in real vehicle data, a joint SOH estimation method for extracting health factors in multiple operating conditions for real vehicle data is proposed. Firstly, the method of condition reconstruction of real vehicle operating data is proposed, which divided the real vehicle data into driving segments and charging segments to reduce the complexity of battery operating conditions. Then, the SOH evaluation models of driving conditions and charging conditions are constructed respectively for SOH estimation. For driving conditions, the internal resistance is selected as the SOH evaluation index, and SOH is estimated by the battery internal resistance modeling method based on Auto-LightGBM. For charging conditions, the capacity is selected as the SOH evaluation index and the battery capacity is calculated by extracting the constant-current charging segment. Then the influence characteristics of the capacity are extracted to establish the capacity model and estimate the battery SOH. The results show that the average absolute percentage errors of the modeling methods based on internal resistance and capacity are both less than 9%. Finally, a comprehensive evaluation model of SOH combining charging and discharging is established, and a joint estimation method of battery SOH combining charging and discharging segments is proposed. The SOH error based on real vehicle data is within 2%, and the reliability and adaptability of the proposed method are verified on laboratory data and multiple real vehicle data.