随着锂离子电池在电动汽车中的广泛应用，准确估计其荷电状态（SOC）已成为保障电池管理系统（BMS）安全稳定运行并优化性能的关键因素。针对宽温度范围和全服役周期内动力电池内部状态难以精确估计的问题，提出了一种基于自适应最小二乘法（AFFRLS）和多新息无迹卡尔曼滤波（MIUKF）的锂离子电池SOC估算方法，实现了不同温度和老化状态下的精确估算。该方法采用二阶RC等效电路模型，通过自适应最小二乘法实现动力电池内部状态的实时监测和模型参数的在线辨识，并与传统最小二乘法（RLS）、卡尔曼滤波（EKF）和带遗忘因子的最小二乘法（FFRLS）进行对比分析。为提升估算精度，提出了多新息无迹卡尔曼滤波（MIUKF）算法，以解决传统无迹卡尔曼滤波（UKF）对历史数据利用率低的问题，并通过不同温度和老化状态下的SOC估算对比试验，验证了该方法的有效性。此外，在UDDS、LA92和HWFEF典型电池测试工况下进行了鲁棒性试验。试验结果表明，AFFRLS-MIUKF方法有效提升了历史数据利用率，能够在不同温度和老化条件下准确反映动力电池内部状态，SOC估计误差控制在2%范围内，显示出良好的鲁棒性。

With the widespread use of lithium-ion batteries in electric vehicles, accurately estimating the state of charge (SOC) has become a key factor in ensuring safe and stable operation of battery management system (BMS) and optimizing performance. To address the challenge of accurately estimating the internal state of power batteries under wide temperature ranges and throughout the entire service life, an SOC estimation method for lithium-ion batteries was proposed based on adaptive least squares (AFFRLS) and multi-innovation unscented kalman filter (MIUKF). The method can realize accurate estimation under different temperatures and aging states. By using a second-order RC equivalent circuit model, the method achieved real-time monitoring of the power battery＇s internal state and online identification of model parameters through AFFRLS, with comparisons made to traditional Least Squares (RLS), extended Kalman filter (EKF), and forgetting factor recursive least squares (FFRLS). To improve estimation accuracy, the multi-innovation unscented kalman filter (MIUKF) algorithm was introduced to address the low historical data utilization issue in traditional unscented Kalman filter (UKF). The effectiveness of this method was verified through SOC estimation experiments under various temperature and aging conditions. Additionally, robustness tests were conducted under typical battery testing conditions such as UDDS, LA92 and HWFEF. The experimental results show that AFFRLS-MIUKF method effectively enhances the utilization of historical data and accurately reflects the internal state of power batteries under different temperatures and aging conditions, with SOC estimation errors controlled within 2% range, demonstrating good robustness.