The lithium-ion battery degradation in electric vehicles is inevitable among its lifetime. Therefore, it is important to predict the remaining useful life (RUL) for battery management system. This paper proposed a RUL prediction method of lithium-ion battery based on dual filters with data-driven and model-based fusion algorithm. The key of the algorithm framework is that two fusion algorithm filters update the battery capacity degradation state and least squares support vector machines (LSSVM) model parameters with iteration online simultaneously. In fusion algorithm processing, the LSSVM model is used as the measurement equation of the unscented particle filtering (UPF) to provide the virtual measurement value of the future time for the UPF-LSSVM fusion algorithm. The uncertainty of the prognostic result is expressed by probability density function (PDF) in UPF processing. Finally, the simulation results show that the proposed dual filters fusion method is superior for lithium-ion battery RUL prediction.

电动汽车中的锂离子电池在其使用寿命期间的退化是不可避免的。因此，预测电池管理系统的剩余使用寿命 (RUL) 非常重要。本文提出了一种基于数据驱动和基于模型融合算法的双滤波器锂离子电池RUL预测方法。该算法框架的关键在于两个融合算法滤波器同时在线迭代更新电池容量退化状态和最小二乘支持向量机（LSSVM）模型参数。在融合算法处理中，采用LSSVM模型作为无味粒子滤波（UPF）的测量方程，为UPF-LSSVM融合算法提供未来时间的虚拟测量值。预后结果的不确定性在 UPF 处理中用概率密度函数 (PDF) 表示。最后，仿真结果表明，所提出的双滤波器融合方法对于锂离子电池 RUL 预测具有优越性。