关键词: 荷电状态, 粒子滤波, 分数阶建模, 多新息技术, 电池管理装备

Abstract: To address the limitations in lithium-ion battery State of Charge (SOC) estimation accuracy caused by particle degradation and time-varying model parameters in traditional Particle Filter (PF) algorithms under complex operating conditions,a joint estimation method based on the Fractional Order Multi-Innovation Adaptive Unscented Kalman Particle Filter-Extended Kalman Filter (FOMIAUKPF-EKF) algorithm was proposed. This method was based on a fractional-order second-order RC equivalent circuit model,in which Extended Kalman Filter (EKF) was employed for online parameter identification to compensate for time-varying effects. Multi-innovation theory and an adaptive noise adjustment mechanism were introduced to improve the Unscented Kalman Particle Filter (UKPF),effectively addressing particle impoverishment and enhancing nonlinear processing capability. Experiments conducted under Highway Fuel Economy Test (HWFET) and New European Driving Cycle (NEDC) conditions demonstrated that the FOMIAUKPF-EKF algorithm reduced modeling errors by 15 %~25 %,exhibited strong robustness under 20 % initial value deviation and 3 % noise disturbance,maintained the mean SOC estimation error within 1 %,and achieved significantly superior accuracy and convergence speed compared with benchmark algorithms such as PF and Fractional Order Unscented Kalman Particle Filter (FOUKPF).

Key words: State of Charge (SOC), Particle Filter (PF), fractional-order modeling, multi-innovation technique, battery management equipment